Coyote Tracks

Checking Prior Assumptions

2026-10-12T00:00:00+00:00

John Gruber, “X Once Again Marks the Spot</a>”:

If you have stayed away, or even just mostly away, from X for years and then come back, people just say things like “Hey, glad you’re back!” On the other networks, especially Mastodon, they tell you you’re “feeding Nazis” — by going to the platform where Tim Cook posts Pikachu videos. These people have either lost their minds or refuse to check their prior assumptions. </blockquote>
Okay, you know, I wasn’t sure if I was going to write about this, but since I was linked to with the word refuse in that paragraph: let’s check my prior assumptions!
That was a link to “If Your New People Are on X, Maybe Your New People Suck</a>,” in which I spelled out those assumptions pretty explicitly:

But the bar that hung the big FASCISTS WELCOME banner out a few years ago hasn’t changed management; the banner is now a permanent sign. So it’s not unreasonable to assume that most AI supporters on X, are, at the least, comfortable with soaking in an authoritarian-flavored social media stew. </blockquote>
John’s defense of X doesn’t address that, instead, it focuses on the undeniable truth that X is still much bigger than Mastodon and Bluesky combined, which—again, undeniably—means that many users there aren’t particularly interested in politics, don’t actively support Elon Musk or the Trump administration, and so on. I don’t think anyone can argue to the contrary. I never explicitly wrote “of course, people can still have perfectly normal, fun conversations on Twitter/X,” so let’s stipulate for the record that yes, of course, they can.
Let’s also stipulate for the record that they can behave in these ways:
I’m sure X is remarkably friendly if you aren’t the target of racial slurs and harassment campaigns.
Anyway, I wonder how many “n——s” I’ll have to sift through if I start posting there again?
Jamelle Bouie</a> </div>
And these ways:
I just think it’s dishonest to say that the politics are unavoidable on other platforms, but they are on X. Its point is its politics; you’re just saying they exist at a level you find palatable. The guys who doxed my family aren’t on the other platforms; they’re followed by staff members on X. Anil Dash</a> </div>
My assumptions about X aren’t that no other platform has harassment, nor that it’s impossible to have friendly, apolitical discussions on it. Neither do I assume that X is the only platform in which people get harassed, or get doxxed. But I do assume that X is intentionally a right-wing political project, and I base that not just on a hunch, but on widely-reported studies, such as one published earlier this year that shows X’s “For You” algorithm shifts political opinion toward more conservative opinions</a>. That study is based on an experiment from 2023, but it appeared to still be true in 2024, as 404 Media reported that the algorithm serves ragebait replies to Democrats more than Republicans</a>.
But those are a whole two years old now, and X’s former head of product claimed the reply-boosting changed in mid-2026. And that’s for algorithm-driven engagement anyway. Perhaps we’d get a better idea of the modern X ecosystem by looking at the accounts with the most engagement this year. Nate Silver, of all people, can help</a>:

Hmm.
Again, I’m not saying that it’s impossible to avoid politics on X, nor am I saying that John is wrong when he says that you’re way more likely to get shit for talking about using AI on Bluesky or Mastodon. What I’m saying is that neither of those contradict my assumptions.
I think when people say “Twitter is a Nazi bar”, it’s easy to dismiss that because Twitter is obviously nothing like, say, the Stormfront forum</a>. So let me stretch the metaphor for a moment: Twitter is a Nazi bar in the same way that a neighborhood coffee shop run by an evangelical church is a mission. You can be a regular at that coffee shop for years without magically becoming an evangelical Christian. Everyone there is just so friendly and nice! Sure, there’s a weekly Bible study on the events calendar, pamphlets on a corner table, Christian soft rock slipped into the playlists. But it’s not overtly proselytizing. Right?
But that coffee shop is absolutely a mission. That’s what they made it for. And that’s explicitly what Elon Musk has turned Twitter into. As Dash put it, X’s point is its politics. The fact that you can have conversations about ostensibly non-political subjects there without feeling attacked is not a sign that X is “remarkably friendly and inclusive”; it’s a sign that what you’re posting isn’t drawing fire there. The political right wing supports generative AI far more the political left wing does, and it is worth pondering that uncomfortable truth for a while. Also, I’m going to go out on a limb and say that if John Gruber posted a photo of himself at a “No Kings” protest in downtown Philly holding an “ICE OUT” banner, the responses on X would be somewhat less friendly and inclusive.
Okay. Whew. Prior assumptions checked. You know what? I’m feeling pretty good about them.
To support my writing, consider a tip on Ko-fi.com.

Two Weeks with the Fitbit Air

2026-10-06T00:00:00+00:00

Back in mid-September, I wrote about reconsidering my Apple Watch use</a>, in large part because I don’t like the promised upcoming “Audio Intelligence” features. And, I thought, it could be good to have one less always-on computer screen with me. Right?
Well, I went ahead and ordered that Fitbit Air, and I’ve been wearing it instead of—not in tandem with—the Apple Watch for the last two weeks. (Actually, getting closer to three now.)
The good</h2>
It’s light and I don’t have to charge it often. They say once a week; I’ve been charging about twice a week, I think, because I’m a little neurotic about letting any device get down much below 40% charge. The software that I’m using—Google Health because I have to, Bevel</a> because I wanted to give it a try after reading a couple of positive reviews—is refreshingly non-gamified compared to Apple Fitness. No rings, no monthly challenges. Google Health is pretty straightforward, and Bevel uses a unique rubric of “strain vs. recovery” to give you ideas on how much you should exercise on a given day. No minutes, no “move points” that theoretically-but-not-really correspond to calories. I also appreciate that both Bevel and Google Health know specific training exercises—for instance, “Arnold Press with dumbbells”—and both have little video/animation showing you how to do the exercise correctly. Instead of simply saying “core training”, you can build a workout routine of multiple exercises and save it.
And, yes, it does let me use other watches, or no watch at all. I only have one other watch, and just to add some irony to the story, that watch’s band broke on Friday. But, hey. It’s a nice watch, much nicer than the Apple Watch in fashion terms, and I’m still considering some other radically different watch as a second. The Apple Watch has never been adventurous in terms of its watch face options, particularly in the “you can have any face you want as long as it’s San Francisco” typography choices.
As far as I can tell, the Fitbit has a more reliable heart rate sensor. It also seems to be better when it comes to sleep tracking. I’m not saying that only because it’s seemingly more likely to tell me that I slept well while Apple Watch and Sleep++</a> tell me my sleep was crap; I’m saying it because what it tells me seems much more in line with the way I actually feel when I get up in the morning.
The bad</h2>
The truth is that I miss a lot—I mean, a lot—of features that I didn’t expect to with the Apple Watch. The biggest one is Apple Pay: I used that on the Watch whenever I could, and it’s a hard habit to break.
It turns out, though, it’s not just Apple Pay. It’s controlling Castro, my podcast playing app, when I’m using the Bluetooth speaker and the phone is in another room. Or looking at the watch for walking directions after I start the map app on my phone. Or seeing what music’s playing when I’m streaming dinner jazz in the evening. Or glancing at the watch to see what the UV index is. Or checking what my current heart rate is. And, while the gamified aspects to Apple Fitness kind of annoy me, the truth is that they also kind of work. Keeping up with my exercise, or even just a general “stay moving to stay fit” vibe, requires just a little more awareness, more conscious effort, that it did before. There are places in my life I’m specifically trying to introduce a need to have more conscious awareness—driving places without the GPS constantly on, for instance. I’m not sure this is a place where that’s good, though.
It’s also simply using the watch without having the phone in my pocket at all. I had a model with GPS, so I could go for a walk and still get the route recorded without bringing the phone with me. Now, not only does that not work, I don’t even reliably get credits for steps in Pedometer++</a> without carrying the phone. This should work, as far as I can tell; the Fitbit syncs with Google Health, which in turn syncs with Apple Health, which should sync with Pedometer++, right?
Yeah, well. This leads to
The ugly</h2>
I’ve triple-checked everything to make sure every app says that it’s set up correctly, that Google Health is a data source for Apple Health and everything has the access that it needs. Doesn’t matter.
I’m sure some of this has to do with Apple making third-party integrations difficult. But that’s not sufficient to explain how out of sync these apps get. The fact that Google Health, Apple Health, and Pedometer++ will show step counts that are literally thousands of steps apart is baffling. I trust Google Health here only because the step counts it shows are close to what I was getting using the Apple Watch, and I doubt I’m walking materially less (or more) now.
I can start a workout in either Google Health or Bevel, and in theory they’ll sync with one another. In practice, I can’t figure out how to start a workout in Bevel that isn’t a “routine” but is just a single workout like indoor cycling. I can only log it after the fact. I can start it in Google Health, but unlike the Apple Watch, it doesn’t present exercises in a smart sort order (e.g., most often used at the top); you can only do that if you save a workout routine in it. If you start a workout in Bevel, after you’re finished, it’s stupidly easy to delete it rather than save it with no chance of recovery. This includes “the application appears to have frozen for three minutes on a screen that looks like it’s just going to give you a review of your completed workout, but if you click the ‘X’ to close it, you’ll discard everything, ha ha.”
So, I use Google Health for workouts. Does it have Live Activities yet, a feature Apple introduced four years ago? Of course not. Don’t be ridiculous. Bevel does, but it’s not particularly good, and also see the last paragraph.
Again, yes, I’ll grant Apple makes integrating non-Apple hardware into their ecosystem more difficult than it needs to be. But that’s not the source of the problems I’m talking about here. Google Health has a cluttered, hard-to-follow user interface that occasionally suffers from weird language choices. (It took me over a week to discover that it called an exercise bike workout not “exercise bike” or “indoor cycle” or anything else with the word “cycle” or “bike’, but instead “spinning”.) Bevel’s is better in most respects, but its workout system is clunkier and, at least in my experience, buggy.
The verdict</h2>
I mean, I’m keeping it, and I’m getting used to not having that screen on my wrist. But it doesn’t mean I don’t miss it.
And, if Apple ever does make a screenless fitness tracker of their own (without creepy AI features), I’m switching back in a heartbeat.
To support my writing, consider a tip on Ko-fi.com.

Why They're Building Muse

2026-09-28T00:00:00+00:00

“We built Muse to help people build the world they want.” — Alexandr Wang of Meta</a> </blockquote>
Most people never say what they want. It’s not because we don’t want. We want fiercely and constantly. We may not say it, but we do type it. Not directly, but it’s implied in every search we make. Every post we thumbs-up. Every video we linger on. And at Meta, we know how to extract that information in ways you’ve never imagined. We know how much time you’re not spending with your family, how crippling your social anxiety is, and how shitty the food you’re constantly shoveling in your mouth is.
When you don’t name your dreams, you mostly can’t reach them. The first hurdle, the biggest one you face, is simply starting. It’s saying them aloud. It’s getting things done, although don’t construe that as infringing on David Allen’s Getting Things Done® trademark because we’re a juicy litigation target. Forget I said anything. The point is: we want you to name your dreams. Your anxieties. Your desires. Your fears. Out loud. Where we can hear them.
Every unrealized want is a quiet tragedy of lost potential revenue.
Now, imagine every person on earth with an overbearing, obsequious, and marketing-department cute AI assistant. A general manager whose job it is to find out what you want and make sure Meta gets a cut. Someone that hears the half-sentence, “I feel like there’s something more I could be doing,” and engages you in artificially supportive dialogue directing you to our sponsored partners and, occasionally, offering life-threatening advice or acting as a creepy sex doll because guardrails are for losers. It builds a plan. Sends the email. Uses your credit card. Makes the phone call. Increases your debt. Keeps you on schedule. Leaks your personal information to hackers in Russia. Outsources all your taste and judgment to us, a company historically known for precisely those qualities. All that’s left is the human part, such as wondering why you’ve given up all the actual joy and passion of doing, leaving you as the depressed, aimless middle manager of your own life.
This might not be a genie in a bottle, but it can absolutely be a monkey’s paw.
The world until now has been shaped by the small number of fanatics who have somehow found a way to make their wants real. Fanatics like Silicon Valley tech bros. And we’d like to keep it that way. It’s been working out damn well for us.
We built Muse to help people build the world we want. Whether they fucking like it or not.
To support my writing, consider a tip on Ko-fi.com.

Org Mode, Marked, and Workflows

2026-09-18T00:00:00+00:00

Last year, I wrote an article called “Seriously, Markdown</a>” in which I gently dismantled the “nobody can do serious writing in Markdown” arguments that come up every so often. I got some pushback, none of which I confess I found particularly convincing because they invariably rehashed points I addressed in the article. Also, people who got their hackles up tended to gloss over the part where I wrote,

I’m not saying “give up and just use Markdown.” Being familiar with another markup system is a powerful argument for staying with it, for a start, and there are other reasons to use something else that have nothing to do with markup. There are specific kinds of technical documentation […] where AsciiDoc is your only practical alternative to full-bore SGML/XML. And as a newly anointed Emacs nerd, I find myself using Org Mode daily at work, and that includes writing extensive notes using its markup system, not Markdown. </blockquote>
So. Something else I wrote was:

When you focus purely on the markup semantics, it’s hard to argue that any other system is “better” than Markdown. </blockquote>
Which is true. The thing is, the obverse is also true: when you focus purely on the markup semantics, it’s hard to argue that Markdown is “better” than any other system. I’ve been doing more writing in Org Mode lately, partially just as an experiment, and partially because—as I’ve noted more than once—the strength of Org Mode isn’t the markup part, it’s the mode part: properties and tags and dates and other bits and bobs entirely outside Markdown’s purview. Here’s a few things I’ve noticed:

While writing italics like /this/</code> rather than _this_</code> or *this*</code> looks a little dorkier, it’s still perfectly readable as plain text, and it’s marginally faster to type: you can type a / without hitting SHIFT. That’s a tiny thing, but if you use italics too much</s> frequently it adds up.</li>
Using Org Mode for complex formatting still requires dropping to HTML once in a while just like Markdown does (e.g., setting the opening words of a fiction story, or a ALL-CAPS PHRASE IN BODY TEXT, in small caps), and Org is clunkier for that. “Markdown is so limited you have to eeeeeewwwww use HTML or template-specific macros” is a common knock among the anti-Markdown crowd, but sorry, Markdown got it right here.</li>
If you want to work with anything outside of Emacs, things get hairy real fast.</li> </ol>I’ve written a few little “helper functions” to copy text as rich (formatted) text or HTML, as well as get the current Safari page and paste it in as a link; I’ve worked to make these work with both Org Mode and Markdown transparently. The key, as is so often the case, is Pandoc</a>. But what I’d lost with Org Mode was Marked</a>, a tremendously over-engineered Markdown preview program for the Mac. I don’t think I need its export tools anymore—I’ve built my own, for both Markdown and Org, inside Emacs at this point—but it has some really neat proofreading tools, particularly with respect to visually highlighting overused words, that I really missed. It’s easy to get Marked to work with Emacs’ Markdown mode, but as the name implies, it’s just for Markdown files. Well, as it turns out, Marked 3 supports custom processors, and you can configure it to use Pandoc so it works with Org Mode files transparently. In Marked 3’s settings, click Processor, then Custom Rules.</li> In the dialog box, click + to create a new rule and name it “Org Mode”.</li> There’s a selection for “Pre” or “Processor”. Choose Processor. (This replaces the internal Markdown processor completely for this rule.)</li> In the conditions box, select “Extension”, “is”, and fill in org</code>.</li> In the Actions box, select “Run Command”, then /path/to/pandoc -f org -t html</code> as the command—using the actual path to the pandoc</code> command line program, of course.</li> Make sure the rule is enabled.</li> </ol> When you’re finished, the rule window should look something like this: </figure> I use a Transient menu command</a> to trigger the preview, in the set of “export” commands I’ve set up for Markdown and Org files. This command will work on both, since Marked is the one determining which processor to use, not Emacs. (defun wm/preview-in-marked (args) "Preview in Marked." (interactive (list (transient-args 'wm/export))) (save-buffer) (shell-command (concat "open -a \"Marked\" \"" (wm/escape-double-quotes (buffer-file-name)) "\""))) (defun wm/escape-double-quotes (str-val) "Return STR-VAL with every double-quote escaped with backslash." (save-match-data (replace-regexp-in-string "\"" "\\\\\"" str-val))) </code></pre>Why not just stick with Markdown?</h2> Honestly, I often do. The few non-Emacs editors for Mac and iOS that support Org are clearly not built around the idea of writing long-form prose with them, so if I’m out with the iPad or want to edit on the iPhone, using Org is a chore. (Authors: please do some basic bold/italic/heading/etc. formatting in the editor as we write, and if you’re going to bundle fonts, include the files for italics and bold.) Conversely, Emacs’s Markdown mode is terrific. I’m writing this blog post in Markdown because that’s what Zola</a> uses, and insisting on writing in Org would invariably add extra steps I don’t need. Like it or not, the rest of the plain text world has largely standardized on Markdown, and if you have to work with that world, using anything else is fighting against the tide. But, if you really are working primarily in Emacs, let’s face it: you’re kind of used to being in your own world anyway. In Emacs, not using Org is often fighting against the tide. Fortunately, Emacs’s extreme flexibility lets you sidle Org into workflows that might not otherwise expect it. To support my writing, consider a tip on Ko-fi.com.
Reconsidering the Apple Watch 2026-09-14T00:00:00+00:00 For most of my life, I just wasn’t a watch person. Sometime after college, when whatever cheap watch I had at the time broke, I just stopped bothering. The advent of smartphones—my first was the T-Mobile Sidekick</a>, which deserves its own post at some point (people either forget or just never knew that the Sidekick had an on-device app store, visual voicemail, and one of the best pre-iPhone web browsers on it)—gave me a “watch” in my pocket. What else did I need, right? Well, sometime around 2014 or 2015, I bought a mechanical watch on a whim. The brand is “88 Rue du Rhone,” and the watch does nothing but (gasp) tell time. But it was a good watch. I liked it. I still have it, although its battery died years ago. When the Apple Watch came out, I wasn’t initially interested, but sometime in 2017 or 2018 I bought a Series 3. Once I got into the fitness aspects—haphazardly and half-assedly, but still—it was game over. You have to keep wearing the Apple Watch, man, or you’re not gonna close your rings. You have to close your goddamn rings, right? As of right now, I have a Series 9 watch, one of the stainless steel ones. Ironically, its sapphire glass looks more beat up than I remember the glass on the two aluminum watches I owned getting; it’s not scratches as much as what look like permanent smudges in one corner, like a coating wore off. Its battery health is down to 84%. This would be a good time to consider a Series 12, maybe going back to aluminum. And the Series 12 is, in a lot of ways, a great upgrade. The problem—if you’ve heard about the watch, you can guess what’s coming—is the “Audio Intelligence” features, Live Rewind and Siri Recap</a>. As Apple describes these: Live Rewind recalls what was just said in the last 15 seconds by showing a text snippet on your Apple Watch when you double-press the Digital Crown, so you can catch something you missed or are less familiar with. You can ask Siri about the content of the snippet or simply save it to the Siri app. Siri Recap ambiently takes notes on your conversations throughout your day and creates high-level Apple Intelligence–generated summaries that you can use later to jog your memory, so you can stay present in the moment. Your summaries auto-delete after 7 days if you choose not to save them. </blockquote> Now, I trust Apple when they talk about how much their approach considers privacy. Everything has to be turned on (e.g., it’s not “this is always on and you have to remember to turn it off”). Live Rewind only activates when you double-press the watch’s crown, they never actually record audio (“there is no recording to share, forward, or produce if requested by any party, because no recording exists”), the data that is produced is end-to-end encrypted and even Apple can’t access it, it doesn’t identify speakers in conversations, tries to omit “potentially harmful content” (everything from credit card numbers to hate speech), so on and so forth. Yet I still doubt “we’re not recording everyone without their explicit permission, we’re just transcribing and/or summarizing everyone without explicit permission” will get the collective “oh, well, that’s fine, then” reaction Apple seems to be expecting. As Glenn Fleishman writes</a>, These features feel like they were designed in a lab and passed through to production without enough people wondering, “Will this provoke a tremendous backlash because of the gaps in privacy that are relatively obvious?” </blockquote> Look, I get the logic here: if “AI wearables” will ever actually be a thing, turning what millions of people are already wearing into one makes sense. And maybe, because there are millions of Apple Watches out there, people won’t dump a beer over your head for wearing a Series 12 the way they would if they saw you wearing Meta Pervert Glasses or a Plaud NotePin</a>. But this still feels like a tremendous failure to read the room—and at the worst, a misstep big enough that it could get the general public to turn against Apple Watches. I’ll be honest: I’d already been looking over at the Rue du Rhone watch over the last few months, hoping that just changing the battery will get it going again, wondering if perhaps I’d enjoy going back. But the fitness is such a pull, right? Yeah, but there’s the $99 Fitbit Air</a>. That’s—let me do the calculations—one-quarter the starting price of the $399 Apple Watch. It has no display and its battery lasts for a week on one charge. While it has an optional subscription to get the Gemini-powered “health coach” working with it, if I don’t pay for the subscription, I don’t have to deal with Gemini. Win-win! And, as of just last month (August 2026), the Google Health app now syncs with the Apple Health app. It has one big downside: it’s Google, and I’ve been trying to de-Googlefy over the last year or so. But the trade-offs might well be worth it. I can get a Fitbit Air and another whole watch for well under the cost of the Apple Watch. I’m not committed to this, yet. But I’m bringing that Rue du Rhone watch to Batteries Plus in the next few days. To support my writing, consider a tip on Ko-fi.com. I Don't Think I Can Stay in Tech 2026-09-01T00:00:00+00:00 I came across a blog post today that hit me harder than expected. It’s nominally about one specific action and person—companies and people giving money to David Heinemeier Hansson’s “Omacom Foundation,” despite DHH’s well-publicized turn toward fascist ideology—but it’s also about, well, everything. Brennan Kenneth Brown writes in “Normalized Fascism in Open Source: $12 Million Given to DHH</a>”, Much of the world has become far too complacent and complicit with people (and their companies) who proclaim and hold far-right, fascist ideology. These are not image boards residing in the fringe corners of an anonymous Internet made up of people who have slipped through the cracks. No, these are wealthy people in power who are blatant and shameless in their support for those who continuously beat the drum of bigotry, xenophobia, transphobia, and cynical accelerationist hatred. </blockquote> He details DHH’s supremacist writing, going on to list people and companies donating millions of dollars to “bankroll [DHH]’s Linux distribution, Omarchy, which is nothing more than a few vibecoded scripts on top of Arch Linux.” The people include the CEOs of Shopify, Stripe, Dell (yes, Michael Dell himself), Block/Square, Cloudflare, Dropbox, Coinbase (of fucking course), and the creator of OpenClaw; the companies are 1Password and 37signals. The latter company is one DHH founded, known for Basecamp and for banning talk of “politics” in April 2021</a>. The former company is, well, 1Password. I’ve interviewed at several of those companies, and genuinely wanted to work for two of them. Several of them were already falling in my estimation—Stripe’s prudishness in what they accept payments for is a problem given that they and the equally moralistic PayPal constitute an effective duopoly on online payment processing; Dropbox and Block have been enshittifying for years, while Dell and Coinbase started enshittified—but this is a new low. Brown again: A dozen of the most powerful executives in tech looked at a man who wrote that a British march calling for the deportation of non-white citizens was “heartwarming,” and decided he was the person they wanted to hand a multimillion-dollar foundation to. </blockquote> He goes on to describe other open source developers who’ve declared themselves “apolitical,” from Ladybird’s Andreas Kling “treating the mere existence of trans contributors as more ‘political’ than excluding them” and posting in support of Charlie Kirk to conservative homophobe Brendan Eich of Brave—a browser often recommended as a private alternative to Chrome—giving targeted harassment forum Kiwi Farms a helping hand in payment processing. We have been watching the fragile fury of white masculinity and hate metastasize and fester over the years, from Gamergate’s harassment campaigns through years of anti-SJW cringe compilation content mills, the marking of Antifa as the enemy, the tantrum-throwing #AllLivesMatter movement, and the manufactured panic over DEI and “wokeism.” You may be in a position where the specific harms and rhetoric being spewed are not aimed at you, and you may have been coaxed into viewing what I’m saying here as overreaction, or as a rabid, self-righteous screed. Millions of dollars have been spent to ensure this, and it could not be further from the truth. </blockquote> I think this post walloped me because it plainly states truths I’ve been grappling with—sometimes</a> openly</a>, more often internally. Who am I comfortable working for, and what am I comfortable doing? I’ve surely worked for people with politics wildly divergent from mine in the past; is it really that much of a bigger deal now? Yeah, but in the past, those people weren’t multimillionaires and billionaires actively courting fascism. I know much bigger bloggers than I who probably do see this as an overreaction, but we’ve been tut-tutting at those who’ve been raising the alarm on this for over a decade, and the alarm-raisers have been right.</a> When watching Senator Welch confront Joe McCarthy</a>, the proper reaction is not to think Welch was being too darn mean. Where you want to draw your personal lines is your business. My personal lines don’t extend to refusing to use the new Siri AI for personal tasks such as telling me when my next flight leaves or making a note of where I parked. I’m not going to ruthlessly prune my life of all software I suspect of having AI-generated code; I have thirty years experience reviewing 100% human-generated source code, and if you think AI invented “slop”, I have sad news for you about every program you depend on right now. (Having said that, I absolutely want to minimize the amount of AI-generated code I’m exposed to; way too much enters production with little to no human review. There’s going to be real catastrophic shit happening in the next eighteen months, in both big name software and the indie app world.) But my personal lines definitely extend to don’t give money to people who enable fascism and don’t create value for them, either. The latter one covers recommending them, linking to them, or creating content for their services. And it sure as hell covers working for them. I don’t know the politics of the AI-focused companies that have given me otherwise welcome nibbles. But whether we like it or not, generative AI has a political valence at this moment in time, and that valence is largely in opposition to my values. So am I getting out of tech entirely? Frankly, that depends less on me than it does on tech. If I can find a company looking for a tech writer that isn’t an AI startup, LLM-pilled enterprise business, or run by a white supremacy funder, I’m open to a conversation. Otherwise, I don’t know. Freelance editing. Website design. Laying out flyers and zines. Assuming I can find anyone who wants those services who doesn’t want to just ask ChatGPT to do it. Otherwise, maybe it’s not too late to learn woodworking. To support my writing, consider a tip on Ko-fi.com. If Your New People Are on X, Maybe Your New People Suck 2026-08-19T00:00:00+00:00 I’ve long liked Federico Viticci’s MacStories</a>, although I’ve drifted away from it as it’s gotten more focused on AI lately. (Note: this is what writers call “foreshadowing”.) I’ve been listening to the Connected</a> podcast he co-hosts since before it was called Connected. Federico is a man of strong opinions and, from all appearances, strong principles. In 2024, for instance, he and MacStories managing editor John Voorhees strongly called for AI regulation</a>, writing, The trouble with the companies developing LLMs is that instead of offering a fair trade to publishers and other creators and respecting their wishes about whether their content is crawled, they just took it, and in some cases, brazenly lied to everyone along the way. The breadth of offenders is staggering. This isn’t just a startup problem. In fact, a wide swath of the tech industry, including behemoths like Apple, Google, Microsoft, and Meta, have joined OpenAI, Anthropic, and Perplexity, to ingest the intellectual property of publishers without their consent, and then used that property to build their own commercial products. None of them paid any regard to the Robots Exclusion Protocol. Instead, some offered a way for publishers to opt out of their crawling activities, but only after they’d already taken the entire corpus of the Internet like a thief offering a shopkeeper a lock after emptying their storefront. </blockquote> And, admirably, they led the charge off Twitter, leaving in December 2022</a>: We’re not closing down our Twitter accounts (yet), but as you may have noticed, they haven’t been active lately and won’t be going forward. That’s because we’ve grown increasingly uncomfortable with the direction the company is heading. If you’ve been keeping up with the news, you know what I mean. </blockquote> Yes, Federico! I know exactly what you mean</a>, and good on you for taking a stand. In 2025, they took an even more sharply-worded stand against Meta’s hate-friendly moderation policy revision, choosing to leave all their platforms</a>, writing, “Technology should bring people together not divide and dehumanize them, which is why we’re finished with Threads and Instagram.” In my own goodbye-Twitter post</a>, I wrote, New Twitter will be hostile to anyone queer, or non-white, or slightly to the left of Ronald Reagan. You may be a creator who wants to stay on Twitter to reach your audience, but the audience there will inevitably tilt toward the anti-woke, All Lives Matter, gender critical, Just Asking Questions crowd. If they’re your audience, congratulations, I guess. </blockquote> And report</a> after report</a> after report</a> shows this to be true. While it’s certainly not true that Twitter/X is only right-wing trolls and grifters, it is absolutely true that it is the preferred platform of right-wing trolls and grifters, with an algorithm tuned in such a way that it can demonstrably pull regular users toward the political right, either through rage-baiting or deliberately presenting a distorted view of The Consensus™. This is why it’s poisonous for journalists to make Twitter/X their primary platform; while Bluesky and Mastodon are both left-wing compared to Twitter/X, that’s because of conscious, active choices made by Twitter/X’s management, not choices made by those who run Bluesky (let alone the thousands who run individual Mastodon servers). But now MacStories has come back to X, with Viticci posting to his personal account on August 17th</a>, “Over the last two years, I’ve fundamentally changed as a creator. I’m still a writer and podcaster, but I guess I’m also a new kind of ‘builder’. Today, I need to find new readers, listeners, and developers, wherever they may be.” Editor John Voorhees also returned, even more explicitly citing AI as the reason: “A lot has happened at MacStories since I last posted here, including that we’ve begun building tools for readers and for ourselves with Codex and Claude Code. It’s a natural extension of our obsession with automation.” New posts on MacStories are being tweeted (Xed?) out again as well. Niléane, a (former?) MacStories contributor, trans LGBTQ+ activist, and co-host of a podcast published by MacStories, is understandably furious</a>: Federico has decided to fully return MacStories to X without telling anyone involved with the site whatsoever (apart from John Voorhees). No post on the site, no text message, nothing in Slack, nothing in the newsletter either, so nothing for their Club MacStories members who also had to find out elsewhere. </blockquote> The Bluesky and Mastodon crowds treat Generative AI with a fair amount of skepticism—not exclusively the “burn it with fire” attitude of many artists and writers, but misgivings about its utility</a>, ethics</a>, and legality</a> are common enough there that users won’t greet a stream of “cool things you can do with AI” with wild enthusiasm. Twitter/X users, conversely, are collectively far more into AI, along with being far more into alt-right influencers, conspiracy theories, and strange grifts. I’m not saying generative AI is wholly of a piece with the latter three, but I am, perhaps, tapping my chin and making a rude hmmm! noise. MacStories undoubtedly conceives of their return to Twitter/X as an apolitical outreach to AI fans. But the bar that hung the big FASCISTS WELCOME banner out a few years ago hasn’t changed management; the banner is now a permanent sign. So it’s not unreasonable to assume that most AI supporters on X are, at the least, comfortable with soaking in an authoritarian-flavored social media stew. It would, if anything, be unreasonable to assume otherwise. MacStories will indeed get new readers, and those readers will bring new preferences about what they do and don’t want to see: more AI, less woke. A very specific brand of “no politics.” And I don’t doubt Viticci and Voorhees will tell themselves that being there doesn’t mean they won’t stick to their principles, and that they’re not going to bend to please the anti-woke crowd. The problem is that by choosing to return to X, they already have. To support my writing, consider a tip on Ko-fi.com. Lake Minneola to Lake Wobegon and Back 2026-08-18T00:00:00+00:00 In July 2022, I wrote “Thoughts on Leaving California</a>” about my move back to Florida after twenty years in the Golden State; I followed it a year later with “A Year Back in Florida</a>”, which ended with “I’ll check back in after another year.” I’ve been doing that annually ever since, although the date has kind of crept up; this year I’m over a month late. I wrote 2024’s check-in</a> during a 16-month stretch of unemployment. While part of that was a deliberate break, most of it wasn’t. That check-in was kind of angry. Last year’s check-in</a> got even angrier, even though I’d gotten a fun contract that ended up running over a year. Much of the anger stemmed from the country’s steadily degenerating sociopolitics; more than a little came from a sense of being trapped by geography. So how am I feeling now, four years in, unemployed in a terrible job market, less than a decade from traditional retirement age when it seems increasingly likely we’re heading into a recession—at best—within the next few years? The market for remote technical writing jobs, at this moment, needs to improve to be dreadful. If I rule out jobs that involve babysitting LLMs—and jobs that directly tie into the federal government, an organization I’m less than enthused about at the moment—I’ve found a total of twelve to apply to in the last month. That’s enough to receive unemployment benefits in California, but not in Florida, which makes it as hard as possible to get the meager amount they offer. On the flip side, I returned to travel starting last year, making a summer trip to Pittsburgh by train. This year, I returned to both Silicon Valley to see friends and to Lawrence, Kansas, for the amazing Ad Astra “Repeat Offenders” workshop</a>. And, in a few weeks, I’m going to Maryland to see Garrison Keillor. “Wait,” you might be thinking, “you’re traveling that far for a show with the Prairie Home Companion dude who was fired for sexual misconduct</a>? Seriously?” Okay, the truth: my 81-year-old mom is using the Garrison Keillor show as an excuse to make one last trip to her home town of Baltimore, and I’m going with her. Yes, she is still a Keillor fan, listening to his short weekly podcast and getting his newsletters; we’ve talked about the misconduct allegations in the past, and yes, I think she’s performing a “they were only allegations and they weren’t really that bad, were they” rationalization, the way many people make excuses for people whose work has meant a lot to them over the years. (Also see: J.K. Rowling, Neil Gaiman.) I used to enjoy Prairie Home, but left to my own devices, I wouldn’t buy a ticket even if the #MeToo aspect wasn’t present. His post-PHC work is a touch crankier and definitely less poignant. (And on Substack, because of course it is.) But, as I said, this is less about Keillor than it is about making a trip back to Baltimore that mom has been talking about taking for years. While she’s had chronic health issues for all her life, they didn’t get in her way until this decade. (She might quibble with that assessment, but we took annual trips around the country.) But a case of shingles, of all things, about five years ago left her with a neuralgia behind one ear that makes it painful to wear glasses or hearing aids just at the time both her hearing and vision are getting markedly worse. And a couple of years ago, the summer of her 80th birthday, we both caught Covid, and it made her lifelong respiratory issues considerably worse; a cough has become chronic, and her already too-low weight has been steadily dropping, a dilemma compounded by a litany of dietary restrictions and stubbornly untreatable digestive issues. I know I’ve been to Baltimore before with her, but honestly, I can’t remember when that trip happened; it couldn’t have been more recent than the 1990s. It’s not an area I feel a connection to, but it’s an area she’s deeply connected to. Since I moved in with her in 2022, her reminisces of childhood in Baltimore and southern Maryland, where she had an aunt she stayed with in summers, have become near-daily conversation. This year, I came to the dismaying realization that my mother is deeply sad about her life. About the prickly relationship she had with her own mother; about the sense of isolation since we moved in the 1980s from relatively cosmopolitan Sarasota to backwater Hernando County—first to Spring Hill, a sprawling unincorporated community of tract houses, then out here to Ridge Manor, a rural unincorporated community destined to be overrun by tract houses in the next few years. And the isolation took a sharp turn for the worse post-pandemic. (I wish I’d pushed her to restart an annual Christmas party she used to hold for friends and neighbors, which I suspect helped with connections more than she admits to herself. She would point out, though, that several of the people she used to invite have moved away themselves in the last few years.) This is a sadness I’m, bluntly, bad at helping with. I’m too reserved, not good at giving comfort (these are issues I know I should unpack myself sometime), and get impatient when we rehash evergreen complaints, from Florida’s endless construction to the sorry state of the American healthcare system. But it’s also that this side of her is mostly new to me. Before I moved in with her, I would have described my mother as generally happy; I’m not sure how much of that was me naïvely missing the forest for the trees, and how much of it is a new, late-in-life melancholia. The Baltimore trip will be a challenge in a dozen different ways, from the complications her health conditions may make while traveling to the fact that our lovely neurotic cat, Maisie, will be left alone for a full week. A neighbor will be coming over to feed her, clean litter boxes, etc., but Maisie literally refuses to be seen by any human other than my mother and myself. I am hoping we won’t come back to a house full of shredded furniture and dried puddles of cat vomit, but I am not placing any bets. Regardless, my mom needs this trip, and I’m happy to do it with her. I’ve come to another dismaying revelation, though, this one about myself. I’m going to return to that unanswered question about how I feel now, unemployed in a terrible job market, etc., etc. And the answer is: kind of…good, I think? I don’t want to make any grand pronouncements about being done with tech; in some ways, I love technology as much as I ever have. But to say I’m disenchanted with the industry currently would be a grand understatement. My biggest frustration is I don’t know what else I can fall back on. My biggest advantage is that I have time to figure it out. I’m dusting off my copy of What Color Is Your Parachute?</a>, a career advice book whose first edition is as old as I am. (Figuratively dusting off; I have the last edition from 2022 as an ebook, purchased during my 2024 interregnum.) It’s a quirky book, premised on the notion that you first need to figure out your dream job, then target employers and convince them to hire you. I haven’t gone through it in at least a quarter-century, and the truth is I never truly followed it. The closest I came was circling back to a company that had approached me about technical writing, though, and I don’t think it’s a coincidence that doing that led to the best jobs I’ve had. In the meantime, I’m going to keep writing. I may get better about posting to this blog more regularly; 2025 was the most active I’ve been in years and it was still only 15 posts, and this post makes 11 in 2026. I’ve started following David Sparks’ advice in his Productivity Field Guide</a>, which upon reflection seems a spiritual cousin to Parachute: instead of focusing on “how do you become more productive” or “how do you get a job”, they both focus on “who do you want to be?” It’s a good question. It feels ironic that I’m starting to ask it more seriously again as I approach a traditional retirement age—but, in a different light, maybe that’s a great time to ask. To support my writing, consider a tip on Ko-fi.com. The Case of the Robot Web Thief 2026-08-13T00:00:00+00:00 There’s been buzz recently about how a nascent web and mobile astronomy app, Dark Hours, got wildly incorrectly rejected from Apple’s App Store. Then maybe it turned out that the rejection wasn’t so wild because Apple may well have been looking at the astrology app the developer first submitted under the name “Asterly”. Then it turned out that the web version of Dark Hours was strikingly similar—although UI-wise, hardly identical—to an existing web app called DarkHours (camelcase rather than two words). It was so similar, in fact, that they both shared the same weird bug</a>. All this led to Daring Fireball’s first-ever retraction</a> and considerable opprobrium directed at Terry Godier, developer of Dark Hours (two words rather than camelcase). Godier was up front about creating his web app with Claude</a>, writing in his apology, I was careless in relying on AI to generate the project without doing the work to understand whether it closely resembled an existing project. That’s on me, and I am responsible for what I published. </blockquote> And, yes, sure, but I think this raises fascinating questions about letting an AI do most or all of your work that run beyond these specifics. It’s probable that Godier would have found the existing DarkHours if he did a non-AI web search for something like, you know, astronomy app dark hours</code>. But DarkHours’s source code is available on GitHub, so it’s likely in Claude’s training data. If the names weren’t the same, this could still easily have happened, and it might have taken much longer for anyone, including Godier, to discover it.</li> And camelcase-not-two-words DarkHours was also co-written with Claude. The supposed “smoking gun” of the identical bug could have existed not only with two differently-named apps, but with DarkHours not being on GitHub: Claude very likely would code a similar function in a similar way, making similar mistakes. (In fact, presuming that bug has been fixed in the publicly available source code, that’s almost certainly what happened.)</li> </ul> “Check how close both your idea and your branding is to existing apps” is always sound advice, but using AI makes this considerably more fraught. You may have no idea if Claude’s effectively pulled a chunk of a GPL-licensed library into your proprietary code, or replicated an aspect of an existing app in a way that doesn’t technically violate licensing or copyright but opens you up to a patent lawsuit. It’s easy to write off Godier’s misadventures here as a one-off mess he got himself into, but the implications of said misadventure are far-reaching. Lawyers at tech firms should, at the very least, begin insisting on very strict review and vetting procedures for any AI-written code, and indie developers who’ve been relying on agentic AI need to do the same. Just checking that Claude’s code reads well and runs correctly isn’t enough. To support my writing, consider a tip on Ko-fi.com. AI's Chicken Sandwich Problem 2026-08-07T00:00:00+00:00 I’ve written before about being between two communities with very different attitudes toward our current generative AI moment: Artists, writers, and other (non-tech) creators mostly loathe it with every fiber of their being, to the point where they’ll vociferously argue that so much as asking Claude for a list of good curry places in London is a betrayal of the Enlightenment.</li> Tech people often embrace it even when they admit the downsides. There’s a lot of “the energy costs are horrible, stolen works are used in training, the economics are irrational, also here’s the five apps I vibe-coded yesterday afternoon” happening.</li> </ul> So, it’s not surprising that my own attitude remains conflicted. I’ve been chewing on this article for over two weeks now; I suspect I’ve written five or six thousand words, most of which ended up on the cutting room floor (the Emacs scratch buffer). Full disclosure: I’ve used LLMs, lightly, in non-creative tasks. I’ve had Claude spit out a few Emacs Lisp functions for me, and write a Python script to convert the dates of a thousand or so Day One entries to the format org-journal</a> wanted. I’ve had it build a simple data table or two. I’ve used ML-based image editing tools to judiciously remove background elements in photos like an overhead electric line. What I haven’t ever done is ask Claude to be my writing partner, Codex to write a competitor to Scrivener, or ChatGPT to be my magical robot therapist-slash-girlfriend. I think there’s a fundamental difference between the first set of uses and the second set that boils down to enhancement vs. replacement. Asking Claude to write an Emacs function for me saves me a bit of time; asking it to write a research paper for me replaces work I, not the machine, should do, and more than likely perpetrates deception. Using Apple Photos’ Clean Up tool gets me a better result than I’d get after an hour or two futzing in Photoshop; using a generative image model to create a book cover for me literally replaces a human artist—by synthesizing the work of thousands of human artists who get neither credit nor compensation. And I think that’s a perfectly reasonable way of looking at it: the problem isn’t the tool in and of itself, but how we use the tool. But I’m not sure that it’s enough. It’s not only how the tool is used, it’s how the tool is promoted and packaged, even how it’s designed and implemented. It’s about who benefits from the overblown hype and, at risk of sounding conspiratorial, how it advances an agenda you may not be on board with. It’s about the chicken sandwich of it all. Nerds would like to treat the technology as existing in a vacuum, as being separate from the companies who create them, from their agendas and their politics. But LLMs are not, as much as we wish they were, apolitical spherical cows</a>. Recently I came across a link to writer Mike Brock’s newsletter; in one article</a>, he makes the observation, The AGI [artificial general intelligence] hype does more than extract capital from credulous VCs. It builds philosophical architecture justifying surrender of democratic judgment to those who control the servers. </blockquote> You might think that’s overstated, but recall that one of generative AI’s biggest boosters is literally the same fucking guy that made many of us leave Twitter. A disturbing number of other ostensible luminaries in the field are openly antidemocratic, or are highly influenced by those who are. There’s no escaping the uncomfortable truth that the rush to AI all the things, to put it everywhere and use it for everything, is a rush to cede control of wide swaths of our business and personal lives to a small number of largely far-right multibillionaires. There are legitimate reasons why generative AI has become the poster child for everything wrong with the tech industry in 2026. Okay, there’s an obvious rebuttal: come on, generative AI is not, in and of itself, pro-fascism. It can’t be. It’s just a tool. Yes. Absolutely true. Likewise, a chicken sandwich from Chick-Fil-A isn’t, in and of itself, anti-queer. It’s just a chicken sandwich. But I’m still not going to give Chick-Fil-A my money. The issue isn’t the product; the issue is what they’re going to put that money toward. Driving away with a combo meal doesn’t mean I agree with their philosophy—yet I would be, in a literal sense, buying into it. This is why I don’t eat there. This is why I’m not on Twitter/X. This is why I’m not going to buy boxes from Uline</a> or stickers from Sticker Mule</a>. You get where I’m going with this. Sam Altman and Dario Amodei are both hypesters, but they’re not just hypesters. Granted, Altman is in every effective way worse, from his personal philosophy to his company’s noxious market positioning. (It’s hard not to notice that nearly all stories about people developing “AI psychosis” are stories about ChatGPT.) Amodei, though, is positively soaked in TESCREAL, and Anthropic was the company sued—successfully—for massive copyright infringement. As for the next tier, Elon Musk is, you know, that fucking guy. Mark Zuckerberg is that other fucking guy. And Google has been that fucking company for years now. I don’t think I have qualms about asking Siri AI the occasional complex query, or to draft a Shortcut. Yes, I know Siri AI is partially based on Google models, don’t well-actually me. But beyond that? If I had to choose between Claude and ChatGPT (let alone Grok, Gemini, and whatever Meta calls theirs, I don’t know, Lolita</a>?), I’d choose Claude. But it’s hard to shake the suspicion that the most ethical answer is “none of the above.” We are at a curious moment where the least concerning companies on the politics and philosophy axes may be Chinese ones like Qwen and Kimi. Which is pretty concerning. Where your personal lines lie are, well, personal. While staying to the enhance side of the enhance/replacement line increasingly strikes me as falling into the “necessary but not sufficient” bucket, I know not everyone will agree with me. But here’s what I do know: look, you just don’t understand how incredibly delicious the frontier chicken sandwiches have gotten in the last six months is not an adequate defense. To support my writing, consider a tip on Ko-fi.com. Bring Back the Insert Key on Emacs 2026-07-07T00:00:00+00:00 Back in the olden days, there was an Insert key on Mac keyboards. (There still is on PC keyboards, although in many programs it doesn’t actually do anything anymore.) If you were editing text, you had two different modes, overwrite and insert. If you were in insert mode, then when text was in front of the cursor, typing would insert text. If the | character represents the cursor: banana |crazypants</code></pre> And you typed plaid</code>, you’d get: banana plaid|crazypants</code></pre> This is what you expect in modern editing, right? But, if you were in overwrite mode and did the same thing, you would get: banana plaid|pants</code></pre> Crazy(pants), right? But the thing is, once in a while this is still useful, and it’s basically gone away. The Insert key has become the Help on Mac keyboards; if you have a mechanical keyboard that has an Insert key like I do, it (probably) acts like a Help key. Well, screw that, at least in Emacs. Everybody who uses Emacs can use ^H</code>, or in Emacs parlance C-h</code>, to get help, right? So if you’re on a Mac, you can restore the Insert key with one line in your init.el</code> file: (keymap-global-set "<help>" #'overwrite-mode)</code></pre> If you’re not on a Mac, your Insert key probably works as expected already. At least in Emacs. All well and good, but you know what would be better? Make the cursor reflect the change, so you don’t have to check the mode line. In my init.el</code> file, I tell Emacs to use a vertical line cursor: (customize-set-variable 'cursor-type '(bar . 2))</code></pre> Since overwrite-mode</code> has a hook that gets called when it’s toggled, it’s trivial to change the cursor based on that. I turn it into a box. The line/box combo seems (to me) to be visually very suggestive of what’s going to happen: in overwrite mode, the box highlights the character it’s going to replace, while in normal insert mode, the vertical line is where the character you’re going to insert will go. ;; change cursor based on overwrite mode (defun wm/update-cursor-type-for-overwrite () "Set cursor-type to box when overwrite-mode is on." (setq cursor-type (if overwrite-mode 'box '(bar . 2)))) (add-hook 'overwrite-mode-hook #'wm/update-cursor-type-for-overwrite)</code></pre> This is something I remember fondly from earlier versions of BBEdit—I don’t know whether current versions still support this if you have a keyboard old enough to send a true Insert key. But if you’re in Emacs, you can force the issue. Party like it’s 1989! To support my writing, consider a tip on Ko-fi.com. The Best Worst Email Client 2026-05-25T00:00:00+00:00 In November of last year, I wrote about what I did and didn’t want to do in Emacs</a>. It’s great for anything that involves text editing/processing, good for certain kinds of management tasks (it’s a great Git client</a>, and even has surprisingly useful file management tricks up its sleeve); it’s not—at least to me—great for tasks that involve viewing, like RSS reading and PDF viewing. I’ve kept putting this to the test, and stand by most of what I said back then. Most of the time I’d rather view PDFs in Apple’s own Preview. When it comes to RSS, I want good typography and the ability to read on my iPad and iPhone—Elfeed</a>, for all the virtues it undoubtedly has, is a nonstarter for me. (Yes, I really tried. More than once.) Email, though…well. Here’s the thing about me and email clients: I hate pretty much all of them. The last email program I truly loved was Bare Bones’ Mailsmith. Quoting from Matt Neuburg’s 2003 (!) review</a>: Since the notion that your current email client might need replacing is probably threatening to you, let me help by giving you, up front, some reasons to stop reading this article altogether. You shouldn’t proceed if you like your email as more than text – HTML, format=flowed, or pictures, rendered right in your client program. Mailsmith displays just text: you can easily open attached HTML or images in another program, but you won’t see them within Mailsmith itself. That suits me perfectly, since pure text (plus attachments) is just what I think email should consist of. But if you really need to see email messages with pictures, tiny print, underlines, or funny “quote bars” down the side, don’t consider Mailsmith. You also shouldn’t read on if you are religiously opposed to an email client that keeps its messages in a database. I actually agree with that position; since email is just text, why over-engineer with a database and all its attendant problems, such as huge file sizes and data you can’t retrieve if things go wrong? This is a prejudice I’ve managed to repress in order to accept Mailsmith; perhaps, after some soul-searching, you could do likewise. </blockquote> Users of Emacs’s mu4e</a> might resonate with a lot of that. Mailsmith is, if anything, quirkier than mu4e. Unlike Mailsmith, mu4e uses standard Maildir folders, using its database solely for indexing/searching. It can display HTML email inline (and send it, too). And Mailsmith was POP-only. Its developers decided that re-engineering its database back end to work with IMAP wasn’t worth the effort, a choice which may have been technically correct but spelled inevitable doom. Now, I’ve tried a lot, like, a lot, of “modern” email clients. The built-in Apple email client is extremely straightforward, but underpowered. Thunderbird</a> is too lumbering and heavyweight for what I want. Hipster ones like Spark</a> are pretty, but focus on things I don’t care about. Mimestream</a> is solid, but it’s Gmail-only. Microsoft Outlook is, unfortunately, Microsoft Outlook. MailMate</a> is nerdy and, in some ways, the most obvious spiritual descendant of Mailsmith, but I find some of its UX inscrutable. Yes, I am aware of the irony of saying this just before describing an Emacs-based client. And increasingly, commercial email clients are expensive and/or subscription-only. MailMate, $40/year unless you want to freeload. Mimestream, $50/year, no freeloading option, and still only works with Gmail. The beautiful and powerful new client MailMaven</a>, $75 with updates for 12 months—if you want updates after that, it’s another $75. Superhuman</a>, a business-focused mail client that some podcasters I listen to love, a staggering $300/year. Look, folks, I’m a late middle-age dude living on contract work. I need to reduce the number of things I’m paying for. A big reason I didn’t test out mu4e last year was because it’s a bear to set up. See, while I said mu4e works with IMAP, in truth you need to set up another program like isync/mbsync</a>, to get your actual mail in and out of your computer. And if you have multiple email accounts, you have to set up Maildirs and configuration for each of them. This isn’t difficult, but it’s very Unixy—and up until recently, I led a multi-Mac lifestyle, and damned if I was going through all this on two computers. Early this year, though, I switched to a single desktop-class laptop, turning my old desktop into a server. So, I took the plunge. After you set up mail sync, you still need to configure mu4e itself. See, mu4e is, behind the scenes, just a search engine: when you ask it to open your Gmail account inbox, you’re executing the search query maildir:/gmail/INBOX</code>. You create what mu4e calls “bookmarks” for those queries, which lets it behave like a conventional email client. Then you need to set up “contexts” for your different accounts, telling it which folders to use for drafts, sent messages, trash, and archiving (which mu4e calls “refiling”). In typical Emacs fashion, you do this with, you guessed it, Lisp code. (The “refile” command can be a Lisp function itself, because of course it can: you could make it act intelligently based on subject and/or sender, sending some things to archive, some things to a receipt folder, and so on. This is something you can’t do with other email clients.) Again, this sounds more difficult than it is, but it’s definitely not click and go like most modern mail clients. After you slog through all of that, behold the glory of mu4e! The mu4e dashboard, as it appears for me.</figcaption> </figure> Okay. You know how we talk about Mac-assed Mac apps? This is the opposite of that. But look closer. A mail count like “2(+2)/17” lets me know that, in that inbox, there are 2 unread messages out of 17 total; the “+2” is how many are new since the last time I looked. Since these are search queries, they can do things other mail clients can’t easily do; I have unified views for unread, today, and spam. Normally I don’t like unified views because I want to quickly know what mailbox any given message is actually in, something not all mail clients are good at. This one is, as we see from the message view: The mu4e message view.</figcaption> </figure> Again, not a thing of beauty, but again, interesting. It’s great at threading conversations, and that message view is very functional. We know what mailbox this message belongs to because “Maildir” is one of the headers. I also know it’s from a mailing list. The view of the message from Broad Street Brewing shows the HTML email as rendered by Emacs’ bare-bones internal web browser. If I wanted to, I could open the “real” HTML email either in a separate web browser window or using WebKit inside Emacs. The advantage of mu4e’s text-forward approach is that you can open any email safely, without triggering tracking or JavaScript payloads. And, this thing is fast. I “only” have 22,000 email messages or so, but it can search hundreds of thousands of email messages instantly and accurately. Beyond that, you have all the features you’d expect, even if the UX takes getting used to. But just like one of Mailsmith’s secret weapons was using BBEdit’s editing engine, one of mu4e’s secret weapons is using Emacs. When you’re writing a message, you can do everything Emacs can. This is where I was most wrong in my original take: while I may spend more time reading email than writing it, having a full editor when I am writing it is fantastic. (That you can integrate mu4e and Org mode, the infamous outliner/agenda/notebook system, is just icing on the cake.) So have I sold you on it? No? Of course not. Look, not only is mu4e not for everyone, it’s not for most people. But if you’re an Emacs fan—or at least Emacs-tolerant—and you’re either a serious mail power user or someone like me who just hates most normal email clients, setting it up may just be worth the initial pain. As for how to set it up, that’s a capital-J Journey. A couple of pages which were helpful to me: Email setup in Emacs with Mu4e on macOS</a></li> Managing Multiple Email Accounts with mu4e and mbsync</a></li> </ul> To support my writing, consider a tip on Ko-fi.com. Soylent Confessions 2026-04-27T00:00:00+00:00 If you’ve been on the Internet in the last fifteen years and change, you’ve probably heard of Soylent, whether you wanted to or not. Nutrition drinks as a category have been around forever, pitched in various forms to dieters or weightlifters or people who just frequently find themselves rushing in the morning. Nobody gives their coworkers shit for grabbing a prepackaged smoothie for breakfast. Soylent, though, has a distinctly bad reputation. Partly, that comes from the fact that the original version—the one most people know about—looks and tastes like watery cake batter (to pick one of the kinder descriptions). But a lot of it comes from Soylent’s associations. Its “food for people who hate food” gestalt. Its vibe. As a December 2025 SFGate article put it</a>, When this liquid meal replacement was first introduced more than a decade ago, it became symbolic of the Bay Area’s rising tech scene and the brutal gentrification that followed. […] At its core, Soylent feels like an extension of capitalism and a byproduct of one of Silicon Valley’s most harmful ideologies: that we ought to forgo the human experience in favor of productivity. </blockquote> But over the years, Soylent quietly came out with ready-to-drink shakes instead of weird powders, shakes with normal flavors like chocolate, strawberry, and vanilla, and those shakes were (looks around from side to side furtively) (lowers voice to a strained whisper) (ahem) good. I’m sorry. Truly. I know I’m supposed to nod in dutiful agreement with SFGate’s Ariana Bindman when she describes it as “medical-grade slop.” But they tasted good. Not good for something stupidly named after a horror in a dystopian sci-fi movie, not good for something created by the kind of dingus who thinks stupidly naming the product after a horror in a dystopian sci-fi movie is clever. Just good, full stop. And no, I’m not someone who hates food. I love finding new restaurants, from neighborhood dives to aspiring Michelin Star contenders. Thai? Hawaiian fusion? Burger joint? Vegan Tex-Mex? Proper Tampa-style Cuban sandwiches on La Segunda bread? (La Segunda is to Cuban sandwiches what Amoroso is to Philly cheesesteaks.) Bring it all on. I’ve never been an “eat at my desk” kind of worker; my favorite office jobs were in downtown San Francisco and San Jose, because I could walk to so many places for lunch. And while I’m not a great cook, I’m not bad in the kitchen. (I’m getting better in late middle age, now that I have to; when I lived in California, it was way too easy to eat out for most meals, and I did.) But I was never much of a breakfast person. I made myself a cup of coffee, worked on my own stuff for an hour after waking up, and maybe got a bagel or a donut or whatever on the way to the office. Or not. A coworker in 2015 brought in some Soylent. I wasn’t impressed, although they had a short-lived snack bar that wasn’t bad. (As with many of their early experiments, it ended in disaster.) On a whim, I checked them out again in 2018; by that point, their founder had left the company and ascended to conspiracy-pilled Kanye-voting nutbag</a>, but his former company had discovered the revolutionary concept of “flavor.” So I took a chance on a variety pack. And damn. The chocolate tasted like a good chocolate milk. The vanilla was genuinely great. The strawberry tasted like the way a child of the eighties remembers Strawberry Quik tasting, as opposed to the way Strawberry Quik actually tastes, which is terrible. So, yes: I’ll admit it. I got regular mail order shipments of Soylent for a few years. They didn’t make me hate food, or make me a better programmer, or get me two million in early-round VC funding. They were just a convenient and good-tasting way for a breakfast-hater to have breakfast. Was I bothered by the associations with the most mockable aspects of Silicon Valley’s ethos? A little, but the truth is, I started cutting back on Soylent for prosaic reasons, not symbolic ones. After moving back to Florida in 2022 to help out my mom, I rediscovered that she’s a dedicated breakfast person. We always have cereal on hand now, and a bowl of organic cereal with a generous topping of berries and soy milk is just as nutritious and lower calorie than Soylent, at about two-thirds the price per serving. Then in 2023, Soylent was bought by “Starco Brands,” home of spray-on popcorn flavors, alcoholic whipped cream, and skin care products for athletes. (Really.) They proceeded to immediately raise the price and screw up both their subscription shipping system and their retail supply chain. After three out of four consecutive orders had problems, I gave up. That’s right: a company reviled as a symbol of techbro capitalism is being undone by—surprise!—capitalism. Do I miss it? A little. But it’s unlikely I’ll ever again have an office job I’m rushing to on weekday mornings, and the associations are more wearing in 2026 than they were a mere eight years ago. Also, I’m not sure how long the company is for this world, given their business troubles. (As of this writing, their shopping page is completely broken, which is not promising.) But who knows. Maybe someone sensible will buy the company, change the vibe, rehabilitate the brand. Personally, though? I’d lose the name. It’s not nearly as clever as they think it is. To support my writing, consider a tip on Ko-fi.com. Ready for the Revolution 2026-04-16T00:00:00+00:00 Not so long ago, I considered myself basically a capitalist, and—echoing folks like Ralph Nader and Elizabeth Warren—would have said our problems are with corporate capitalism. Maybe so, but whether or not corporatism is the capitalism that Adam Smith envisioned, it’s the capitalism we have. Inasmuch as the notion of “markets” includes convention dealers’ rooms, great local coffee shops</a> and bookstores</a>, and the guy down the road selling honey made by the bees in his side yard, markets are great—but the capitalists that we have don’t really like markets very much. They’d all be perfectly happy to be the only vendor you could buy from, full stop. The problem for a tech nerd like me is that the technology industry has been distilled to modern capitalism in its most toxic form. Perhaps I’m an idealist, but I don’t think this was always the case; garage startups from idealists who think they can change the world for good may have always been a cliché, but there was a truth there, however aspirational. Silicon Valley’s outlook tended toward what I called technolibertarianism, the socially liberal but anti-regulation philosophy of Stewart Brand</a> and Wired magazine. Yet, there’s long been a technofascist streak, exemplified by neo-reactionaries like Curtis Yarvin</a> and in even larger part by PayPal Mafia</a> members Peter Thiel, Elon Musk, and David Sacks. There is a great nonfiction book to be written exploring why so much of what’s gone wrong with the tech industry and our country’s politics stems from this one fucking company. In the last decade and change, tech billionaires have come to see fascism as the best way to keep what they have. This sucks because, at its core, a lot of technology is still pretty damn fun. Mac hardware is great, and while macOS gives me way too much to bitch about these days, I still prefer it to the alternatives I’ve tried. I don’t need a BYOK</a> but I kind of want one. I haven’t been using my iPad Air much since I moved my computing life to the MacBook Pro M5, but I have a place in my world for an OLED iPad mini. And I’ll be honest: if I could set aside the ethical, legal, economic, and environmental issues with generative AI, it’d be pretty damn cool, too. Large language models give us a quantum leap in natural language processing, proofreading, transcription, translation, and summarization. Yes, I know all the ways in which LLMs are “bad” at all those things, but in comparison to the previous state of the art in machine-powered proofing, transcription, translation, and summarization, it’s just no contest. You can use LLMs in ways that aren’t “generative” in a conventional sense at all. For instance, asking an MCP-powered agent to generate a playlist for you, or using a Shortcut to export a photo from your iPhone photo library and give it a sensible filename by sending the image to an LLM. But of course, you can’t set aside the ethical, legal, economic, and environmental issues with generative AI; it’s being used as an excuse to rip up dozens of fields, including my own (technical writing). When it is used to generate “creative” work, it threatens the livelihood of everyone from visual artists to musicians, and its output—even when it’s “good enough”—is, as I’ve repeated ad nauseum, definitionally never better than median. Worse, it’s creating that median output from input that, very often, it has no legal or moral right to. And while the playlist creator and photo namer don’t threaten anyone’s livelihood, they still take advantage of models built on environmentally and legally unsound foundations. I’ve had an odd journey with generative AI. While I appreciated the problems with it and would never want to use it for creative work, I also appreciated the possibility of that cool stuff, you know? I didn’t want it to write for me, but maybe it could help brainstorm bits and bobs. (It can, albeit like a precocious seventh-grader writing a book report, so its value is questionable.) I don’t like Google’s AI-generated summaries, but Duck Duck Go’s are mostly decent, attempt to give sources, and in general don’t come across like they’re trying to keep you away from other people’s websites. I have little interest in “vibe coding” a complete app, but I’ve generated a few utility scripts, and when I was working on a now-stalled Swift project I tried to use AI to work through a few thorny problems. (Success: mixed.) But even as generative AI gets better, both in terms of output quality and in practical application, my feeling towards it gets queasier, and now we’re back to—surprise!—capitalism. The problem isn’t the technology in and of itself. It’s the companies pushing it. Every tech product and service screams AAAAAA!!! IIIIII!!! louder, harder, and faster. Microsoft Office is now “Microsoft 365 Copilot”; Visual Studio Code is now “the open source AI code editor.” MinIO, a NoSQL database company, now advertises itself as an “Exascale AI Data Store.” I nearly got a job at MinIO, which turned out to be a dodged bullet: they recently fired all their tech writers because they think they can be replaced by guess what? Laravel</a>, the most popular PHP web development framework, advertised itself as “the PHP framework for web artisans” for years, until earlier this year when it switched to “the clean stack for artisans and agents.” My favorite crazy story outlining/plotting application, Dramatica Story Expert</a>, has been “reinvented” as a web-based, AI-powered service</a> that costs $20 a month to use at its cheapest level. Other than the AI companies themselves, who is this all for? Who asked for this? Who benefits from this? The users? If they’re developers and/or productivity nerds, maybe. If you’re using the AI as an assistive tool rather than doing the equivalent of falling asleep in the back seat while your Tesla drives off an overpass, it works, mostly. And I almost wish it didn’t, because it’s led a depressing number of otherwise smart, kind people to swallow the hype wholesale with a hear-no-evil, see-no-evil attitude. Writers: Generative AI models were built on our stolen works, are deeply unethical, and risk devaluing our entire profession. Artists: Generative AI models were built on our stolen works, are deeply unethical, and risk devaluing our entire profession. Developers: Wheeeeeeeeee! James Thomson (@jamesthomson@mastodon.social)</a> </div> But who else? AI’s being pushed to creative workers, from Dramatica to Sudowrite to whatever fuckery Adobe is up to this week, and obviously those programs have paying customers. But the people in the publishing, film, and game industries who love AI the most aren’t writers, musicians, and artists—they’re the people who want to screw over writers, musicians, and artists. Sudowrite and similar “editors” are writing programs for people who hate writing: the screwing-over crowd is their target market. Dramatica has historically positioned itself as a tool for actual creators, but their marketing increasingly talks up “teams and studios”—as it must, given the prices they need to charge given what the AI companies their product is now built on top of charge them. I’ve been in computers long enough to live through several epochs, but I’d say there have been only three seismic shifts, fundamental changes in the way we related to our computers: the graphic user interface, the internet, and the smart phone. It feels like it’s the right time for another revolution, and we’re being told by pundits and companies—especially companies—that AI is clearly it, baby. I don’t think it is. To repeat it loud enough so those in the back can hear, I’m not suggesting generative AI is valueless. If they can solve both the economic and legal/moral issues surrounding it—and those are, to be clear, very fucking big ifs—it has a bright future in verticals where it can do what it’s best at: tightly constrained, non-creative generation. As a former programmer who’s seen what a lot of commercial code actually looks like, I assure you that you and I both, right now, rely on 100% human-written code shittier than what Claude Code farts out. Putting unreviewed LLM-generated code into production is insane, but as long as there are humans who read, understand, and verify the generated code, it’s going to be fine. But beyond code generation, we fall off the cliff of diminishing returns real fast. Chatbots are good as a jumping-off point for web searches and research, but you can’t rely on them. They’re not good at writing text that requires, or is even just improved by, any kind of verve or voice or original thinking—so that leaves, what, first-level customer support responses? Meeting summaries? Business memos? I’m aware of, and sympathize with, the “this way lies Idiocracy” concerns here, but I suspect they’re going to prove overblown. While tech-centric productivity nerds I say “tech-centric productivity nerd” with love; I’ve bought more than one guide from MacSparky, including his actual Productivity Field Guide—which is, interestingly, the least tech-centric of all his guides. might go all in on robot assistants</a>, most of us will at best half-heartedly try whatever Copilot does in Microsoft Office, like asking it “what goddamn ‘ribbon’ did you hide the button I’m looking for in.” (It’s great at answering that.) AI simply doesn’t have the juice to support either its market valuations or the “this is the next big tech revolution” hype. Very few AI “users” who aren’t developers—or tech columnists/podcasters—pay for it. AI companies aren’t just unprofitable, they’re losing money at an earth-shattering rate, and evidence is mounting that the entire industry stays afloat through financial shenanigans on the order of Enron and WorldCom. If OpenAI and Anthropic both disappeared in a flaming cloud of worthless stock options tomorrow, some developers and tech-adjacent productivity nerds would fall into mourning, but everyone else would shrug and move on. Except for visual artists, animators, voice actors, and writers, who would throw a worldwide bacchanalia that made Mardi Gras look like a monastic retreat. But if AI isn’t the next tech revolution, what is? I’ve seen a lot of hope that we’re collectively finally going to decide to stop giving big tech all the power, that we’re going to lean into owning as much of our digital lives as we can. Indieweb! Decentralization! Federation! Personally, I’m all for this. But I’ve watched my non-techie communities opt for Bluesky and Threads over Mastodon, recoil from the idea of setting up their own websites—or set them up and stop using them after a couple of months—and choose to bemoan being locked into big-name subscription apps in lieu of investigating alternatives. We all agree it’s terrible to depend on giant companies with predatory business models and/or morally corrupt executives/backers, but darn, it’s just so convenient (they write from their MacBook Pro). Frankly, I don’t have a good answer for this. I don’t think it’s impossible that the current iteration of AI leads to the next true tech revolution, but we need to have another, less tech-focused and more fundamental revolution first. As much as possible, we need to choose smaller businesses to both buy from and work for; when we choose to buy from or work for bigger companies, we need to take into account what they’re doing for, or to, our environment, our privacy, and our politics. The current system needs to be shaken up, if not outright dismantled, and that’s not going to come from the top. It has to come from us. Also: fuck off, Laravel, I’m going back to Perl. To support my writing, consider a tip on Ko-fi.com. "How are you leveraging AI in your technical writing?" 2026-04-06T00:00:00+00:00 I shared a LinkedIn post with that title last week. I’m going to share what I wrote there, but I’m going to ask a second question after that, which I may or may not share with the LinkedIn crowd later. If you’re thinking “but I thought you were an AI skeptic!”, hold on. Okay? Okay, here’s the post: First and foremost, what we’re calling “AI” is not the AI of sci-fi movies. Generative pretrained transformers are statistical prediction engines; when they’re coupled with large language models, what they do is generate a statistically likely continuation of their input text. That’s it. Full stop. Now, it turns out that “that’s it” undersells how amazing that output can be. The statistically likely continuation of an input text that's a search query is a plausible answer to that query; the statistically most likely continuation of feeding a Github repository into an LLM with a command to write documentation for the code is plausible documentation. But plausible is not the same as correct. The bigger and more complicated the task presented to an LLM is, the likelier it is to go off the rails—maybe just a little, maybe a lot. If you don’t know what the output should look like, you may not catch this. That’s the obvious peril of “vibe coding”; a similar issue arises with “vibe documenting.” The more constrained the task you give it, the better job it’s going to do at it. Claude will probably nail “write a Python script to transform the date formats in this text file” on the first try; “write a clone of Photoshop that runs as a web app,” not so much. Likewise, “write reference descriptions for this API, including input parameters, sample output, and error conditions” will get you a better first draft than “write a complete casual developer’s guide for this database.” Second, the statistically most likely continuation of anything is definitionally the median. An LLM will be much faster at generation than a human, but at its absolute tippy-top best, it’ll be better than about fifty percent of humans at the same task…and worse than about fifty percent of them. The more difficult and expansive the problem you’ve given it is, the less likely it is to be at that absolute tippy-top best. It’ll make mistakes a human wouldn’t, because it will have made what appear to be incorrect decisions or assumptions: calling a library that doesn’t actually exist, or including a guide on Git basics in an overview of how to write game plugins. The thing is, it’s not actually making any decisions or assumptions. It’s just determined that statistically speaking, a lot of code in its corpus calls similar libraries in similar functions, and a lot of documentation in its corpus talks about version control after it talks about NPM modules (or whatever). So, if you’re going to use AI in documentation: Target constrained, clearly defined use cases that contain sufficient context the LLM will stick to the task at hand.</li> Expect the output to be first draft quality, not production ready.</li> Humans must review the output, both to verify the engineering aspects (do the samples work, does the API really have those calls, etc.) and to make sure it’s staying on topic.</li> Do not expect the LLM to have your corporate voice or stick to your corporate style guide, even if you’ve prompted it to do so. It won’t.</li> The human reviewer(s) should be using a Markdown linter. (There’s probably some automated way to handle this, but I think it’s a good idea regardless.)</li> </ul> Okay. LinkedIn enough for you? Here’s the second question: “Should you be leveraging AI in your technical writing?” A tech writing group I’ve been loosely involved with the past few years, Write the Docs, has been consumed by the question. A writer in that group who lost his job but got a new one less than a month later rhapsodized about LLMing all the things to do it (rewriting his LinkedIn profile and his resume, editing cover letters, creating custom targeted resumes, etc.). People in the WTD Slack talk about rebranding themselves as “context engineers” or “knowledge engineers.” Technical writers, they say, have been writing clear instructional procedures for technical concepts for decades. That’s literally the job description. Ergo, we’re exactly who companies should be hiring to direct AI systems. We should be kings of the new world! And, man, my urge to leave tech and open a tiki bar or coffee shop or take up woodworking has never been stronger. I’ve talked about my</a> various</a> reasons</a> to be skeptical/wary of AI before, so I won’t rehash them. LLMs have become good code monkeys, but generated code is less likely to be algorithmic plagiarizing than images or, more broadly, any attempt at creative work. (Also, “less likely” is not the same as “never”.) Training models on copyrighted material almost certainly is a copyright violation if you don’t have a legal right to access it: it’s behind a paywall, it’s a non-freely-licensed ebook, possibly it’s just a blog article that’s only licensed for non-commercial use (like the Creative Commons “Attribution-NonCommercial-ShareAlike” license</a> I use). Literally all the current popular AI systems use corpora that violate copyright in that regard. So, you know, there’s that. But even if we set that aside (which we shouldn’t), even if we grant the relatively rose-colored view of the first part of this post, there are two big questions. First, for companies: do you want your documentation to be middle-of-the-road at best? Because that’s what you’re going to get if LLMs do most of the work. Companies that let AI write all of their documentation won’t even hit that mid-level bar. The AI drafts that I worked with were okay, in the way that a first draft written by a freshman intern hopped up on Red Bulls would be okay. Every section needed editing, and most sections needed to be entirely rewritten. Or moved. Or cut completely. The overcaffeinated robot intern doesn’t understand information architecture or ontologies. You may think you are going to get it there with sufficiently clever prompting; you won’t. Did using an LLM save any time? In theory, it should. Even if the first draft isn’t that good, writing is almost always slower than editing. In practice, though, if I’m doing so much editing it basically is writing, it’s probably a draw. And if I’m not, the documentation isn’t as good as it could be. If the LLM adds any value here, it’s in giving me something to bounce off of, something to look at and say, That section doesn’t need to be there, it needs to be over here. The voice here is all wrong. This example is redundant and this other example doesn’t quite work, but I can find one that does. Back in Write the Docs, there seems to be a consensus—a vibe, if you will—that users can’t tell the difference between AI-written and human-written documentation. As “documentarians,” we need to accept that and get with the times if we want to keep a job. To which I say: bullshit. Users can tell the difference between good documentation and bad documentation, just like they can tell the difference between good UX design and bad UX design, or good typography and bad typography. Maybe they can’t explain the difference, why this interface is good and that one is bad, why that poster looks beautiful and that one looks like ass, or Stripe’s API documentation is mostly terrific and Apple’s API documentation is mostly undercooked tripe. But that doesn’t mean they can’t tell. So this leads me to the other question, for other writers: is this actually what you want? Like I said, I can see ways to use LLMs and still be a writer. But prompt engineering is not writing. Could I do a job that involves just tweaking prompts to Claude until it emits something with the general appearance of product documentation? Sure, I guess. But I’m not convinced anyone will pay me enough to keep me in the amount of high-quality rum—we are not talking Captain Morgan here—I’ll require to deal with that. So, should you be leveraging AI in your technical writing? Here’s a LinkedIn-friendly answer: If you wanna play around with it in the margins—using it for proofreading (not rewriting), or as a jumping-off point for initial drafts, summarizing, research—try it! But don’t take it for granted that it’s going to make you faster or smarter or better. Check its work constantly. Here’s an honest answer: I am already so damn tired of thinking about this question. I’m so tired of the fear that I’ll be penalized if I do not sound sufficiently enthusiastic about a tool which is already being used to replace people in my field despite being demonstrably incapable of doing so. (“Now that we have electric screwdrivers, we’ll never need trained carpenters again!”) I’m so tired of the entire tech industry willfully ignoring the signs that AI is a bubble, that it is being wildly oversold and jammed into places where it is inappropriate or even dangerous. And frankly, I’m not just tired, I’m angry seeing other technical writers buy into the idea that the only way they can stay in their career is by helping people who never took the field seriously to start with relentlessly devalue it. Maybe I’m a weirdo here, but I’m a writer because I enjoy writing. I’m a technical writer because I like the work of taking complex topics and figuring out best to present them to a specific audience. I take pride in my craft. I think there’s even some art to it. I’m not saying it’s impossible to take pride in figuring just the right words to type to get Gemini to spit out the best possible mid-tier documentation it’s capable of, but—actually, scratch that. I think that’s what I’m saying. Now, if you’ll excuse me, I’m going to go work on a tiki cocktail menu. You know, just in case. To support my writing, consider a tip on Ko-fi.com. What You're Allowed to Do 2026-03-15T00:00:00+00:00 John Gruber linked</a> to a piece by Sam Henri Gold entitled “This Is Not the Computer for You”</a>, which both is and isn’t an article about the MacBook Neo. It’s a reflection on how kids get started with computers, and how when he was a kid, “every limitation was just the edge of something I hadn’t figured out yet.” It’s lovely writing, but these few lines struck an unexpected chord of recognition: The limits you hit on the Neo are resource limits—memory is finite, silicon has a clock speed, processes cost something. You are learning physics. A Chromebook doesn’t teach you that. A Chromebook’s ceiling is made of web browser, and the things you run into are not the edges of computing but the edges of a product category designed to save you from yourself. The kid who tries to run Blender on a Chromebook doesn’t learn that his machine can’t handle it. He learns that Google decided he’s not allowed to. Those are completely different lessons. </blockquote> What I recognize in this is the most common charge I see against Apple products on places like Hacker News, and from HN-minded people: that the only thing you can do with them is what Apple allows you to do. That they’re all so locked down, you’ll be constantly fighting against arbitrary restrictions. Yet, I’ve seen few good responses to the challenge, “What can’t you do on a Mac that’s impossible because of arbitrary Apple-imposed restrictions?” If someone responds at all, they often respond with something that is possible and simply isn’t as easy to do as they’d like, or can’t be done in the way they’re accustomed to. The rest mostly fall into three buckets: UI preferences (for instance, “no focus follows mouse”); limitations that aren’t deal-breakers for non-Apple products (non-user-replaceable batteries, anemic gaming performance); fact-free insistence that macOS either will be or already is “just as locked down” as iOS is. The most valid criticism is that Apple is ruthless than other companies in cutting backward compatibility; I have one remaining 32-bit Intel Mac application that I wish I could still run. What Gold’s article made me reflect on is that, in a quarter-century of owning Macs, I’ve never felt “arbitrarily restricted” from doing anything meaningful. I have a plethora of scripting systems and utilities that hook into the OS and applications at deep levels. I can install tens of thousands of Unix apps via Homebrew. Sure, the tightening of various security models eliminated Unsanity’s Haxies</a>, but losing those weren’t dealbreakers. As for the iPhone: of the Android users I know, few have ever installed an application from anywhere other than the Google Play Store. I doubt many others regularly do things they couldn’t do on an iPhone, even when they think they are. There are definitely areas where Android is ahead for “normal” users, like pairing with smart accessories that aren’t made by the phone manufacturer. Or as a friend who went into a rant about this some time ago would point out, Bluetooth file transfers. (Most iPhone users would likely have the same response to that I did: “they don’t support floppy disks, either; what year is this again?”) But in practice, I suspect most acquaintances of mine who specifically chose an Android phone because “it’s more open”—because of vibes—would have been just as happy with an iPhone. And bluntly, nobody should choose an Android tablet over an iPad unless they have a very specific need, like running an app that simply doesn’t and can’t exist on iOS, or because they want an e-ink display. (I’ve toyed with the idea of getting a Boox</a> and trying to install Emacs on it to get a nerdy “writing deck,” but it sounds like more of a capital-P Project than I want to take on.) It does feel like I face an ever-increasing amount of restrictions in my computing life, but with few exceptions, they don’t come from Cupertino. They’re imposed by moving not just services but applications to subscription-based models, and by rising digital surveillance. It’s hard to simply own your media and your software now, to browse the web without suspecting all your movements are being harvested to give to marketers, to data brokers, to ICE. While I trust Linux distributions to be at least as privacy-respecting as Apple, I certainly wouldn’t extend that trust to Windows or Android—and if you’re browsing the web with Chrome and using Google web services, it doesn’t matter which OS you’re using. The MacBook Neo isn’t the computer for me, but it may well be the best first computer for Sam Henri Gold’s imagined child. Maybe they will become a developer, or a designer, or a filmmaker. But we need them to also become a digital rights activist. It’s not the Mac that will box them in⁠—it’s the rest of the tech world that’s made that their business model. To support my writing, consider a tip on Ko-fi.com. “Trim Silence”: Threat or Menace? 2026-02-18T00:00:00+00:00 While I don’t think Apple Podcasts itself has a “trim silence” feature, it’s popular in a bunch of podcast players. Overcast</a> probably kicked it off with its “Smart Speed” function, but it’s in Pocket Casts</a>, Castro</a>, and I’m sure a bunch of other ones. The idea is that the player “listens” for silences in the podcast and (hopefully subtly) shortens them, so you’re saving a little time even when you’re playing the podcast at normal speed. I suspect many users always run with Trim Silence enabled. I did, until I realized that occasionally Overcast interpreted very, very soft sounds as trimmable, subtly screwing up the timing of music. No problem: you can disable it for specific podcasts, right? Well, if you do that, you’ll learn that podcasts that I think of as “NPR style,” e.g., ones with a certain kind of production value like 99% Invisible</a> and Pop Culture Happy Hour</a>, actually sound pretty good without Trim Silence enabled. In fact, they sometimes sound better. They’re edited precisely—even though PCHH itself is, despite being an actual NPR production, very much the “several dudes talking” kind of show unique to the podcast form. (With the asterisk that PCHH’s rotating panel is extremely diverse across just about any demographic line you care to name.) Now, this isn’t a unique observation; I remember chatter a year or two ago on social media that I follow from people discovering this (and making dramatic denunciations of the option). But what I didn’t expect is what happened when I listened to podcasts from folks who are part of what one might cheekily dub the 5by5 Diaspora without trim silence enabled. (Brief aside: 5by5 was a network a decade and change ago run by Dan Benjamin, with shows featuring hosts that pod-listening Apple nerds know: Merlin Mann, John Siracusa, Myke Hurley, Marco Arment, John Gruber, and more. It was huge for a brief time, then fell apart in just as brief a time, with next to all its hosts either jumping to the nascent Relay</a> or going independent. Someday I’d love to know the story of what happened there; reading between the lines, it sounds as if Benjamin was either difficult to work with, didn’t give hosts generous business deals, or both.) My impression of many of those conversational shows was that even though they were relatively lightly edited, they were still edited tightly enough to take out awkward pauses. But listening to a few shows that I otherwise love without the Smart Speed or Trim Silence function at work, though, makes me question that belief. Are their creators are, consciously or not, assuming that everyone’s listening with Trim Silence enabled? This could be a style choice, to be sure, the idea that the conversation’s going to sound more natural if you keep in the two or three seconds while a host gathers their thoughts between sentences or phrases. In a sense, it does. But I don’t think it’s the kind of “natural” you actually want in a podcast any more than you want it in a scripted show, where any long silence is deliberately written into the script. When I listen to these podcasts as recorded, I think: this could have used a little more editing. Is this bad? No, not in the grand scheme of things. I am re-evaluating my podcast listens, in part because I don’t have the free time to keep up any more, and in part because I’m burning out on tech news in general and on AI AI AI AI AI AI in particular. Even so, I’m not going to stop listening to podcasts that I still like because of too-long pauses; I’ll just keep Trim Silence enabled for them. To support my writing, consider a tip on Ko-fi.com. Is AI Helping You to Do Work, or Just to Do Stuff? 2026-02-10T00:00:00+00:00 For someone who’s neither all-in on AI nor a passionate anti-AI crusader, I think about it a lot. From 2018 through 2023, I worked for Viv Labs, a startup from the folks who built Apple’s Siri; they had been bought by Samsung in order to fix Bixby, their much-maligned assistant. (The punchline here is that, to a large degree, they did fix it, but nobody noticed.) And my novel Kismet has emergent AI as a subtheme. I don’t want to belabor, yet again, the very valid complaints about AI—the horrendous energy consumption, the way it’s being jammed desperately into everything, how all current models are trained on illegally-obtained material, and so on. But let’s nonetheless consider that Anthropic, the company that’s paid millions to settle copyright violation lawsuits, may well be one of the least unethical AI companies compared to their competitors with suicide advice engines and child porn generators. The industry as a whole is plagued with moral issues that go above and beyond the rest of Silicon Valley’s, and that’s saying something. To be clear: the lack of a high-quality model trained entirely on legally accessible content casts a shadow on even otherwise innocuous uses, and I’d personally be leery of building a business that depends on current AI models.</aside> And yet, and yet, I find myself uncomfortable with a maximalist anti-AI position. Yes, current models incorporate pirated material, but there’s nothing intrinsic to the technology that suggests you couldn’t have a fine model trained entirely on legally accessible content. (I don’t buy the argument that the act of training itself is a copyright violation; if a search engine can legally get to it, a large language model should be able to, also. That LLM scrapers seem to be incredibly bad net citizens is an artifact of LLM companies being run by shitty people.) Yes, LLMs make mistakes when they parse natural language, screw up summaries in sometimes egregious ways, make imperfect transcriptions of audio—but they’re nonetheless a quantum leap past previous attempts. But the sudden virality of Clawdbot</s> Moltbot</s> OpenClaw snapped something else into focus for me recently. In case you missed OpenClaw’s story, I’ll let TechCrunch explain it</a>: According to its tagline, [OpenClaw] is the “AI that actually does things”—whether it’s managing your calendar, sending messages through your favorite apps, or checking you in for flights. This promise has drawn thousands of users willing to tackle the technical setup required, even though it started as a scrappy personal project built by one developer for his own use. </blockquote> On one level, that is absolutely the kind of Star Trek stuff AI optimists dream of. MacStories’ Federico Viticci, who’s recently poured his long-standing love of automation and scriptability into a 24/7 bear hug for productivity-focused generative AI tools, enthused, “OpenClaw showed me what the future of personal AI assistants looks like</a>.” (I am convinced that an alternate universe version of Federico is a legendary Emacs guru, given that I doubt there’s a more scriptable editor in the world.) On another level, it’s a massive risk to both your security and your wallet. That TechCrunch article notes an investor wryly pointing out, “‘actually doing things’ means ‘can execute arbitrary commands on your computer.’” If you have OpenMoltClawBot taking actions for you based on incoming mail and messages, it’s possible for someone to send a message to you that takes action you don’t want. As for the risk to the wallet, Viticci breezes past “I’ve been playing around with [OpenClaw] so much, I’ve burned through 180 million tokens on the Anthropic API”—but that’s over $500. In part, these flaws stem from OpenMoltClawBot being dim in the way all LLMs are. If you ask it to run a task at sunrise, it may write a cron</code> job</a> that calls the Claude API every fifteen minutes with the question “is it sunrise yet?” By literal definition, an LLM’s “smarts” are the median of what it’s been trained on: they will never write like the mythical “10× programmer.” Now, there may be some other flaws stemming from OpenClaw itself being vibe-coded. Its author has literally said “I ship code I don’t read</a>.” I, for one, see no problem giving code no human has so much as glanced at access to my personal files and bank accounts. What could possibly go wrong? But let’s assume that your personal installation of OpenMoltClawBot doesn’t wire your entire 401(k) to a script kiddie in Hackensack. Even with that relieving condition met, how much time are you spending with it that’s doing stuff rather than doing work? See, this is an old joke/truism about productivity systems: it’s easy to mistake Getting Things Done™ for getting things done. Spending time trying out different task managers, defining tags and projects, setting up your Inbox Zero system—it feels productive! It feels like you’re making progress! Variants of this appeal to a certain kind of mindset: you’re always on a search for the perfect task manager, the perfect pen and paper notebook system, the perfect</a> text</a> editor</a>. But a lot of your time is actually being spent in overhead, whether it’s managing the system you’ve chosen or switching between systems you’re trying out. Time spent organizing your tasks is time not spent doing tasks. Time spent fiddling with new text editors is time not spent writing. (As much as I’ve taken to Emacs, its endless fiddlability can be a double-edged sword.) A lot of the time you think you’re doing work, you’re actually just doing stuff. Stuff is not work. If you’re spending a lot of time brainstorming things you can get OpenClaw to do for you, you’re spending more time than you think doing stuff. Maybe more time than you’d be spending actually just doing the work yourself. The examples of integration that Viticci shows in his article are, without a doubt, damn cool, but some have a distinct “I built it because it was possible” air to them. Is it neat that I could tell my virtual assistant in Telegram to add two things to my to-do list and then resume playing Spotify on the computer that I’m sitting in front of at the moment? Yes, but I can already press a global shortcut anywhere to pop up a window to add something to my to-do list. If I want to resume playing Apple Music on the computer that, again, I am sitting in front of at the moment, I can (ahem) press the play button. Emacs Org Mode nerds would also point out that, if you live your writing life in Emacs, there’s a global capture key for adding not just to your to-do list, but almost anything—with customizable templates to boot.</aside> But what’s started to get a little crazy-making for me is the “AI will inevitably eat everything” attitude, whether in Viticci’s article or in Marco Arment’s prophesying in a recent episode</a> of the Accidental Tech Podcast</a>. To quote Viticci: When the major consumer LLMs become smart and intuitive enough to adapt to you on-demand for any given functionality—when you’ll eventually be able to ask Claude or ChatGPT to do or create anything on your computer with no Terminal UI—what will become of “apps” created by professional developers? I especially worry about standalone utility apps: if OpenClaw can create a virtual remote for my LG television (something I did) or give me a personalized report with voice every morning (another cron job I set up) that work exactly the way I want, why should I even bother going to the App Store to look for pre-built solutions made by someone else? </blockquote> Any given functionality? Really? Do you actually think you’ll eventually be able to ask Claude or ChatGPT to write a word processor for you? A photo editor? The next Zelda game? No, let me rephrase that; I mean, you can ask Claude or ChatGPT anything, and it’ll plunge ahead with the absolute confidence that comes from being a predictive text generator being sold as a robot brain in a jar. But do you think that what comes out is going to be good? Really good? Overcast</a> good? Halide</a> good? Scrivener</a> good? How long do you think it’d take you to even begin to approach any of those doing vibe coding? How much human interaction would be needed? How much actual programming and design knowledge would the creator need to arrive at anything remotely usable? Once again, the best an LLM can generate is by definition statistically median. This is not just a temporary limitation of current models; it’s a fundamental artifact of the vector mathematics that makes LLMs work. If you, the user, are patient and clever enough like Federico, you’ll be able to cajole better output. But cajoling an LLM into designing a large, consumer-focused application, from a clean internal architecture to all the nuances of good UX? I’m pretty skeptical. I know many people aren’t. I suspect those people are, over the next few years, going to prove me right in the most depressing way possible: flooding app stores with LLM-generated crap riddled with confusing UX choices, ugly design, memory leaks, and terrifying security holes. I don’t doubt that LLMs can become a primary interaction method we use to interact with certain apps for certain tasks, and yes, agentic AI could become a primary automation tool for those who care about that level of automation. But how many people is that? Viticci describes having set up a Zapier automation</a> that creates a new project in his to-do app for the next issue of his weekly newsletter by monitoring the newsletter’s RSS feed and creating the project after the current newsletter appears in it. Now, he’s replicated that in OpenClaw, which potentially saves money! (Accent on “potentially.”) But most people, faced with a similar problem, would have just created a repeating weekly reminder for “start new newsletter project” and called it good enough. This is a chimera nerds have been chasing for decades: surely, surely, the only reason the majority of computer users don’t write their own programs and automations is that it just isn’t easy enough to do it yet. When I say that I don’t think generative AI is going to finally crack that particular nut, it has nothing to do with AI: I don’t think there’s actually a nut there to crack. To most users, programming and automating is doing stuff, and they want to do work. (Or play the next Zelda game.) Again, I’m not rabidly anti-AI. But in the same way generative AI actively encourages us to ascribe powers to chatbots that they simply don’t have—a simulacra of absolute confidence at any given task is literally baked into LLMs—it encourages us to mistake doing stuff for doing work. Studies show that developers who use AI assistance consistently overestimate how fast it makes them, and I’m not surprised: it feels fast to use AI assistance. My own experiments using AI to draft technical documentation (at the behest of an all-in-on-AI employer) are great illustrations of that. The AI “writer” generated hundreds of lines of stuff that was mostly correct, most of it usable with a few tweaks here and there. But when I look at the diffs between the machine-generated first draft and my final copy, I wasn’t making “a few tweaks,” I was massively rewriting the whole thing. Can I swear the LLM didn’t help? No. If nothing else, it kickstarted the process. And I know that whether I’m coding, writing docs, or even writing fiction, I feel really damn productive when I’m refactoring or editing line after line instead of staring an empty, foreboding text window. But am I doing a lot of work, or am I just doing a lot of stuff? To support my writing, consider a tip on Ko-fi.com. Emacs and Shortcuts 2025-12-07T18:41:41+00:00 It’s a match made in…frankly, I’m not sure where. But I’ve been experimenting with using org-journal</a> as a replacement for Day One, and while it’s pretty good, there are two things that I immediately missed: The way Day One records my location and the weather for every entry.</li> Being able to quickly insert photos that I’ve taken into an entry.</li> </ol> Emacs can’t do that directly, but there are Shortcut actions that do everything I need there, right? So there’s got to be a way to get Emacs to talk to Shortcuts. Well, there is a Shortcuts package</a> for Emacs, but a bit of poking at it revealed that it seems to be just one way communication: it lets Emacs fire off a Shortcut, but doesn’t let the Shortcut return anything back to Emacs. Not so useful. But you know what? There’s a shortcuts</code> command line utility which can not only run a Shortcut, but get results returned to it. As an example of how to use it, here’s the photo picker Shortcut I created (click on it for a bigger image): </a> The Photo Picker Shortcut </figcaption> </figure> The first action lets you select photos. Then, it resizes each one and saves them as JPEG files to ~Documents/journal/images/</code>. Lastly—and this is the part that needs to get back to Emacs—it turns each filename into an Org-style link and outputs it. So, what does Emacs do to call this (and receive it?) It is an extremely simple bit of Lisp code: (defun wm-journal-photos () "Select photos, export, and insert as org links." (interactive) (insert (shell-command-to-string "shortcuts run \"Photos to Org link\" | cat")))</code></pre> The shell-command-to-string</code> function does what it says: runs a command and gets its output as a string. That string just gets insert</code>ed at point. The location and weather Shortcut is even simpler: it just calls the “Get weather at Current Location” action, then creates a text block with LOCATION and WEATHER as Org-mode properties. (Why, you might ask, does shortcuts run</code> have to be piped through cat</code> to work? You got me. It’s not an Emacs quirk, for the record, it’s an Apple quirk.) To support my writing, consider a tip on Ko-fi.com. Apple Wins Me Over Again, but It’s Getting Tougher 2025-11-21T00:00:00+00:00 Whenever I make a big purchase—a TV, a new computer, a car, an A/V receiver—I inevitably enter a long period of internal debate, arguments and counter-arguments, pros and cons. It doesn’t matter whether I’ve set money aside for this, how sure I am that yes, I do need/want/carnally desire the thing in question. Post-purchase, I can look forward to an immediate period of regret and self-doubt, although as with all things, it fades with time. For at least two years, I’ve pondered returning to a “laptop-first” computing approach. I haven’t done this since I went from a MacBook Pro to an iMac/MacBook Air combination in the late 2000s. At times, my iPad was my “real” portable computer, and I only took the laptop out with me when I absolutely needed to. Desktop/laptop works, as long as you sync all your working documents through Dropbox or iCloud or what have you. But it inevitably leaves you with some things on one machine and some on the other, and you end up with a lot of things in the cloud that you suspect you don’t need to keep there. And it also makes it hard to justify having more than one machine be “the big one”: if I have a desktop, that’s the big one, and getting a MacBook Pro just wouldn’t make sense. The MacBook Pro must become the desktop to justify itself. From summer 2025 on, I edged closer to buying a MacBook Pro with the M4 Pro CPU in it. When the M5 MacBook Pros dropped, they left me with a dilemma: wait for the M5 Pro CPU, or get a maxed-out M5 “non-Pro”? After long dithering, I settled on the latter. I knew I’d get a 14″, not a 16″ model, and the base M5 is still nearly twice as fast as my M1 Max Studio across the board in all things but GPU performance. I can “only” get 32GB of RAM, which is what the Studio has, but the truth is that I never come close to overloading that. Ditto for a mere 1TB hard drive. And, this frees the Studio to migrate into being a home server, which I haven’t had since my last Mac mini’s hard drive kicked the bucket. (Which also means the laptop can have even less on it!) I know the argument that the base MBP is “Pro in name only,” that you might as well get an Air, but if you really look at the specs, that’s not true. Besides, I love working at cafés and breweries where a nanotexture screen might really come in handy. But one aspect in my internal arguments, my point/counterpoints, was new this time around. For the first time in a quarter-century, I wondered whether my next computer should be a Mac. All that I wrote about last month in “If not the Mac, what?</a>” remains weighing on my mind. The sloppy direction of the Mac’s user interface, which used to be its crown jewel. The drive for ever more services revenue leading to ever more dubious in-system promotions from Apple itself. The obsession with control in the name of security leading to Vista-like permission prompts and, in edge cases, absurd hula dances to install supposedly “unsafe” software. The UI flubs can, and hopefully will, be turned around, though, and the other issues are more policy decisions than technical ones, and policies may change with new leadership. But they might not, and I have to reckon with that. Worse, sucking up to our current fascist administration is also a policy decision. I can’t disagree with John Gruber’s argument when he writes</a>, Faulting Cook for playing ball and kissing Trump’s ring (among other things) is like faulting a local business for paying into a protection racket—one in which the cops are complicit, and the whole scam exists at the behest of a crooked mayor. Moral rectitude can feel good, but not so much when the cops are burning your store to the ground. There is no authority to appeal to for help when the highest office in the country is running the protection racket. </blockquote> And yet, I can’t help believe that there’s a qualitative difference between reluctantly paying for “protection” and proactively funding the new ballroom. It’s the difference between wheeling to stay on the Don’s good side and wheeling to be the Don’s new best bud. Like Gruber, I don’t believe that Tim Cook is a MAGA believer; it stands against everything that Cook appears to espouse. But that’s the whole damn point. Apple has spent decades consciously chasing a demographic which leans left of center—positioning itself as the tech of choice for artists, creators, and even environmentalists. For them to now cozy up with the far right in the name of realpolitik is perilous. Just ask Tesla, which has been going down in flames like a Cybertruck hitting a tiny pothole. Now, again, I don’t think Apple is in danger of having most of their loyal customers stop being loyal, whether it’s over this or over Liquid Glass or over the way they treat developers or over their increasingly self-aggrandizing long-form commercials pretending to be keynotes. But they’re in danger of having their loyal customers (hello, hi) stop and think about new purchases in a way they wouldn’t have just a few years ago. That’s not a good sign. With me, they’ve won this round. For the design elements I care about—which include that annoyingly nebulous term, “fit and finish”—Apple’s hardware is still second to none. It’s hard to even get close in the PC world, and when you do, the PCs aren’t that much cheaper. And despite all the, uh, Tahoeness of macOS Tahoe, it remains more comfortable for me than Windows 11 or Linux with current versions of KDE and Gnome. It’s a mix of the solid UX foundations that have so far survived Alan Dye’s depredations, the apps that don’t have satisfying replacements on Windows or Linux, and the relative seamlessness of the Apple ecosystem. But the trend lines—both how surprisingly nice KDE is getting, and well, how everything Apple is getting—make me wonder if I’ll choose the same way in 2030. To support my writing, consider a tip on Ko-fi.com. The Emacs for Everything Obsession 2025-11-04T00:00:00+00:00 If you poke around the Emacs community, you’ll quickly find some—all right, loads of—users who want to do everything in Emacs. Not just writing, not just programming, not just publishing their blog. No. They want to read and write their email in Emacs. Read their RSS in Emacs. View their PDFs in Emacs. Manage their tasks and their calendars in Emacs. Browse the web in Emacs, even though as far as I can tell the options for doing so range from bad to bad. Before I got into Emacs myself, this struck me as profoundly weird. As both an old Unix user and an old Mac user, “the best tool for the job” has become deeply ingrained. Emacs is excellent at text manipulation and at leveraging its Lisp machine</a> heart for a remarkable level of extensibility, but surely that doesn’t make it the best tool for everything you can possibly wedge into it. Does it? As I started writing this, friend and former coworker JCS posted Living in the Browser vs. Living in Emacs</a> in his very Emacs-focused blog; he is one of those “Emacs for as much as possible” types, and explains why (and why it’s better than being so focused on the browser): When I’m in Safari there’s not much I can do to adapt it to my way of working. Sure, there are some UI adjustments and other settings to tweak […] but basically I have to work the way the browser authors thought I should. Even if I were using, say, Firefox, which is open source and theoretically open to user change, browsers are sufficiently complex that user modifications are not a realistic option. Emacs is different. Even putting aside its extensive user level configurability, it’s easy to modify the way any particular function works. You can make your desired changes, install the source in your init.el and those changes will be reflected whenever you load Emacs. Similarly, you can write entirely new functions in the same way. You don’t need to understand everything about Emacs to do this, only enough Elisp to express your desired behavior and a few simple rules to install it. </blockquote> I’ve also found this to be true—yet it also explains why there are tasks I’m not interested in stuffing into Emacs. As I’ve written before</a>, Emacs is manifestly not a Mac-assed Mac app</a>. It’s a native app, and it adopts more Mac conventions than your average Electron app does (it even works with the Services menu, mostly), but that’s not its forte. This isn’t a knock against Emacs; when it comes to text manipulation, it kicks all kinds of ass, and I do find myself wanting to move all those tasks inside it. I don’t want to just replace Panic Nova and BBEdit, I want to replace Ulysses</a> and Scrivener</a> and Plottr</a>. (Okay, technically I replaced Scrivener with Ulysses, and Plottr is a weird visual timeline app, but the point stands.) This isn’t trivial. I’ve long asserted that editing prose is fundamentally different than editing code. Ulysses understands the difference intrinsically; Emacs needs to be taught. But other code editors can’t be taught. For instance, I regularly use one of Ulysses’ most wonderful features, the ability to copy Markdown to the clipboard as rich text; I spent about an hour hacking that together in Emacs. I have that saved as a GitHub gist here</a>. It works with Org, too, and has functions for copying Org/Markdown as HTML and Org as Markdown. I’m still working on using Pandoc</a> to replicate Ulysses’s exporting features, but I’m sure it’s not only possible, but will probably end up being a little more featureful and flexible than I could do before. For tasks that aren’t editing, per se, but still text manipulation, Emacs also kicks ass. Sometimes its solutions aren’t pretty, but they’re always extremely functional. Magit</a> is as good a Git client as you’re going to find anywhere else. The built-in Ediff</a> is, as traditional for Emacs, saddled with a somewhat inscrutable UI, but it’s an amazing three-way merge tool (and Casual Ediff</a> gives it a more scrutable UI if you want it). Tower</a> and Kaleidoscope</a> are terrific Mac-assed apps for those purposes, but are they $195 per year better? Cue someone explaining how $195/year is trivial if you’re a Professional. Let me offer a well-considered rebuttal: oh, shut up. If you’re already using Emacs, the answer is no—and I know of people who use Emacs only for Magit/Ediff, just like I know of people who use it only for Org mode (the plain text outliner and task/agenda manager). But while reading mail, RSS reading, and PDF viewing are all about text, they’re not text editing, and to me that makes them fundamentally different. For one thing, I’m often doing them on the iPad or iPhone, where Emacs isn’t available; more importantly, though, I don’t want to read long articles sitting at my computer desk. I want to read them on a couch, on a tablet or e-reader, using something that’s typographically pleasant. To me, that’s Unread</a> or Instapaper</a>. As for email, obviously Emacs is a great mail composer, but it’s not a great mail viewer, and I do more email reading than writing. If you’re the sort who uses Org mode as not just your task manager and outliner but your DEVONthink</a>, it may well still be worth it for that integration. I’m not, in part because I need my task manager to be well-integrated with the whole Apple ecosystem rather than the Emacs ecosystem. And that’s worth keeping in mind: a lot of the arguments for “Emacs everywhere” I see are arguments for how well different Emacs packages tie together. It’s absolutely true, but I’m on a Mac, where the combination of Shortcuts, Automator, Services, and third-party utilities like Alfred and Keyboard Maestro can tie everything together. Org-capture</a> sounds great, but I have “Things</a> capture” everywhere. It’s trivial to turn things into Apple Notes or Tot</a> scratchpads. I use Org where it makes sense to me—:keeping track of tasks at work, where I am in front of my Windows PC 100% of the time. I’m experimenting with using it for story notes and plotting as an alternative to Plottr</a>, which I love in spirit, but find unbearably ugly design-wise. I suspect I’ve learned enough Org at this point, though, to confirm that for my personal task management, it won’t beat the Mac-assed system I already have in place. I’m wise enough to never say never about moving more things into Emacs; it’s possible I’ll eventually jump through the hoops necessary to set up Emacs as a mail client and see if I like it better than MailMate</a>. But I might try MailMaven</a>, too; one of the things I don’t like about MailMate is that even though it is a native app, it doesn’t feel very “Mac-like,” and I’m going to go out on a limb and guess Emacs isn’t going to improve on that. A lot of the other things Emacs can do are things that I don’t see much advantage for me except in specific contexts. “But you never have to use the mouse” isn’t a big draw for me. For one, I have Alfred</a>, so I can do a lot without using the mouse</s> trackpad if I want, and for another, sorry: I like using the trackpad. I’ve been using pointing devices for decades, I’m pretty good with them, and I think keyboard-only jockeys frequently undersell their virtues. Now, though? I’m going to run a spelling and grammar check on this blog post, preview it, publish it, and commit it to my Git repository, all here in Emacs. I may not do everything in this program—but I don’t find myself closing it very often, either. To support my writing, consider a tip on Ko-fi.com. Collaboration or Privacy: Pick One 2025-10-23T00:00:00+00:00 Matt Novak, writing at Gizmodo on October 22</a>: President Donald Trump is currently demolishing the East Wing of the White House in one of the more shocking acts of his second term. Trump plans to build an enormous $250 million ballroom in its place, a structure that will dwarf the existing White House. But who’s paying for it? As it turns out, a bunch of private companies and individuals. And, presumably, they’ll now expect to get a lot more access to Trump. Trump invited tech executives to dinner at the White House in early September, and CBS News reports that the president’s meeting was “interpreted” by at least one company as an appeal for donations. That kind of language suggests that Trump is again utilizing the mob-boss style of language that he’s become known for. Trump can’t explicitly say, “give me money or I’m going to make life hard for your company.” But he’s suggested exactly that several times since taking office in January. Trump had another dinner at the White House just last week, and it featured some of the biggest names in tech and crypto, including Apple, Google, Meta, and Microsoft, as well as Coinbase, Tether, and Ripple. </blockquote> I’ve seen defenses of Apple’s past kowtowing to the Trump administration; they boil down to, what else do you expect them to do? Are we expecting Apple to lead the resistance? They have a business to run, and this administration has made it clear that they’ll make life difficult for businesses that don’t play ball. I get it, and I have as much sympathy for a giant multinational megacorporation as I can possibly muster here. For all its missteps and foibles, I love Apple’s products, and—up until the last decade or so—loved the personality they conveyed. “Think Different” wasn’t just a great slogan; it expressed a genuine philosophy. But there’s a difference between “playing ball” and proactively offering to pay for a new baseball field. Apple could be doing the least of what they need to do to get by in this current moment; they’re actively choosing to go above and beyond. Giving obsequious gifts is a choice, not a necessity. Taking down ICEBlock due to an informal demand rather than a legal injunction is a choice, not a necessity. And funding the destruction of the East Wing of the White House is a choice, not a necessity. While funding the Epstein Ballroom isn’t as egregious in a First Amendment sense as removing ICEBlock was, there’s a way in which it feels more troubling, because it can’t be taken as anything but servile submission. You can’t credibly defend that with what else do you expect them to do. So far, every time this administration has said “jump,” Apple has not only said “how high,” they’ve offered to throw a few more jumps in for good measure. Apple, in the Tim Cook era, has built much of their reputation on privacy, and from all appearances, they’ve genuinely meant it. Their platforms do offer a higher degree of privacy than you can get by default with Android or Windows. The Trump administration, by contrast, has built their reputation on going after all enemies, real or imagined, with all the tools they can dream up—chief among them being the rapid expansion of the surveillance state</a>. They’ve gone out of their way to monitor constitutionally protected speech</a>, to let ICE use facial recognition to track immigrants and tie their identities to “derogatory information”</a>, and to break down barriers between data silos so they can profile you without court oversight</a>. It doesn’t take much imagination to think of ways this administration might pressure Apple (or Google or Microsoft) to give up their users, to quietly—or even openly—build in the kind of back doors that Apple has, in the past, resisted</a>. Even a year ago, if you’d have asked me whether Apple would keep resisting that, I’d have answered “yes” without hesitation. Now, though? Even before this ballroom fiasco, we recognized that the question was no longer “will Apple cooperate,” that it had become “is there a line Apple won’t cross.” But every time you say “yes sir, yes sir, three bags full, sir” to someone like Trump, the more he’s going to ask for next time—and the less power you’re going to have to say no. Maybe there still is a line Apple won’t cross, but each new collaboration pushes that line farther into the distance. To an authoritarian regime, there can be no such thing as a right to privacy against the government. They require, they demand, the right to surveil their citizens in any and every way possible. And a company whose reputation is built on a promise of privacy cannot collaborate with such a government and keep that promise. I’m no longer confident that, at the end of the day, Apple has the courage to not cross that line. To support my writing, consider a tip on Ko-fi.com. If Not the Mac, What? 2025-10-15T10:56:16+00:00 Back in the dark ages (the 1990s), I was an inveterate operating system switcher. At various points, I ran DR-DOS with DESQView/X (what Windows 3.1 should have been), FreeBSD, several distributions of Linux, Windows 2000, QNX Photon, and BeOS. With the exception of QNX, I didn’t just run them, I used them. Used BeOS? Full-time? For more than a year, very productively—but that’s another article. (Just say I wish some of the folks who’ve written or podcasted about BeOS in the last few years had mayyyyybe spoken with some of us who were there at the time.) BeOS is, ironically, what brought me to the Mac: I’d used Macs at college and loved them, but they were expensive and, by the mid-90s, falling behind the times technology-wise. And I liked Unix. BeOS was a kinda-Unix with a kinda-Mac UI, much better than what Linux and BSD had at the time, but arguably torpedoed itself through impatience. So when Mac OS X burst onto the scene, I was blown away. Full Unix in all its nerdery, combined with an Apple-designed desktop environment and the full complement of Mac software! For desktop use, I’ve been here since the days of the tangerine iBook. </a> I did not have the matching mouse. </figcaption> </figure> Even so, I kept running Linux on servers for years, and I dabbled with Linux desktops in VMs. As for Windows—well. I didn’t spend a lot of time with Windows, serious productive time, until my current job shipped me a laptop with Windows 11 Pro. It’s, you know, fine. There are a lot of UX decisions Microsoft makes that I still don’t like (the ribbon interface has made Office near-incomprehensible). And if you think Apple’s attempts to get you to try “Apple Intelligence” are annoying, boy howdy you have not seen annoying until you’ve seen Microsoft Copilot. And there’s all manner of little things that the Mac still does better, tiny nicks and cuts that add up. Trackpad acceleration curves. The relative fiddliness of grabbing window borders on Win11 compared to macOS. Accessing alternate characters the Mac way versus the Windows way. Okay. I’ve done my best not to be that guy who complains about new things—I really do like the look of High Sierra, and don’t pine for the days of Snow Leopard, let alone the pinstripe plastic aesthetic of early OS X. But macOS Tahoe marks the first time I’ve ever sat down in front of an Apple-designed UI and thought: this just looks bad. It’s not “Liquid Glass” that’s the issue, per se: it’s how half-assed the Mac implementation of it is. I already disliked the design direction they’ve been going in the last few years, hiding the incredibly useful proxy icon from title bars, making all the keyboard shortcuts in menus grayed out so they look disabled, and the like. Tahoe, though, makes the Mac feel like—well, like some of those early Linux desktop environments. You know what I mean even if you hate that I said it, don’t you? UI widgets just look too big, with weirdly low information density. There’s drop shadows under buttons for that First Day With Photoshop feeling. Sometimes icons or text labels don’t appear to be quite lined up correctly. Nothing quite goes together. It lacks fit and finish. Combine that with other directions Apple’s been taking in the latter part of the Tim Cook era, and I confess to a growing uneasiness. The focus on service revenue has brought us promotions in System Settings, which feels junky. Apple’s control-freak nature encroaches on the Mac a little more with each successive OS version; I can still do everything I want, yes, but more and more “edge cases” are becoming wrestling matches. They’ve never been great at fixing long-standing bugs, but it feels like it’s been getting worse with each passing year. SwiftUI seems woefully behind where it should be after six years, given that it’s supposed to be the future of UI development on all Apple platforms. Tahoe’s design choices have led to another round of “see, iOS and macOS are merging!” from various corners; while I remain extremely skeptical of that, it’s undeniable that their UX design is drawing closer together, largely to the Mac’s detriment. I do think it’s possible that Apple is working on an entirely new operating system to potentially replace both macOS and iOS, but that’s another article.</aside> And, of course, we haven’t gotten into Tim Cook’s decision to kowtow to the authoritarian-in-chief. Let’s just say that I think that it’s not just a bad decision from an ethical/moral standpoint, it’s a bad decision from a tactical standpoint. So I won’t lie: I’ve found myself thinking, over the past year and change, if I ever did leave the Mac, where would I go? Windows? Aesthetically, Windows 11 may beat macOS Tahoe, which shocks me. In the majority of other respects, though, it’s at best a wash—and for all of modern Apple’s missteps, I’d hardly give Microsoft higher marks in the ethics department. And, on a practical level, I want my damn Unix. Enter Asahi Linux</h2> This is the Linux distribution for Apple Silicon. So far it only supports the M1 and M2 processors, and the completeness of what it supports varies from model to model, but the M1 Air is pretty well supported. While the text-based installer isn’t the friendliest-looking by 2025 standards, it’s straightforward, running with minimal user intervention. After a much shorter time than I expected, my Mac rebooted into Linux. And…everything just worked. I didn’t expect it to be as fiddly as it was a quarter century ago, but it really is at a Mac/PC level of “just get going now.” (Like both macOS and Windows, there are bits and bobs where that fails, but not at an unusually high level.) But: I think it actually looks good? </a> </figure> It took me a bit to figure out how to make the file icons in the details view reasonably-sized, but hey, unlike Windows, at least it has that view. (Windows has a list view, but still no disclosure triangles to look in subfolders, because who would want that, apparently.) Since I still haven’t migrated away from 1Password, all my passwords are right there; since I recently switched from Panic Nova to Emacs, my blog editing/publishing system worked with almost no fiddling. And KDE Plasma’s UI chrome is…pleasantly subdued. Linux still has its own set of little nicks and cuts when you compare it to the Mac, but it’s nothing like it once was, and it’s definitely more in tune with my way of doing things than Windows is. So it’s hard not to start thinking, at this point, “if I really wanted to try ‘living’ in Linux for a while, what would I need to do?” I’m back on my Mac as I’m writing this section, looking at what I use day-to-day. The basics would be easy enough to fill in: a mail client, a Discord client, a Telegram client, a graphics editor, a music player. The big problem would be all the things that I do that are tied to the whole Apple ecosystem, applications that seamlessly sync between all my devices. Things</a>, my task manager, is everywhere, as is my weird math notebook program Soulver</a>, as is my creative writing program Ulysses</a>, as are Apple Notes and Tot</a>. The list goes on. While Apple critics see this as nefarious lock-in, it’s arguably Apple at its best: it really is great that all this stuff works together pretty seamlessly. As long as I actually have a Mac running macOS, this isn’t an insurmountable problem. If I really wanted to move completely off the Mac, it becomes a pretty huge one. Lots of people who don’t use Macs have iPhones, sure, but they work together tremendously well. I’d need another cloud storage provider other than iCloud. I could technically keep Apple Music, but I’d probably better off switching to Tidal or Spotify. (There are some arguments for switching to one of those even while staying on the Mac, because they work better with other non-Apple gear I’m already using, but so far I’ve remained stubborn.) I’d lose Vellum</a>, the amazing ebook (and even print book) creation program. I’d lose Acorn</a>, a photo editor which—in my personal opinion—blows the doors off anything I could get under Linux. (I’m willing to be convinced otherwise, but Acorn does a lot with a brilliant user experience, something GIMP</a> has never been accused of. We won’t even get into the cringe name they refuse to change.) Having enumerated all that, though: this thought exercise has revealed that the software isn’t the real blocker for me. Linux is good, but PCs still (mostly) suck</h2> In the past couple of years I’ve had two Windows laptops in the house. One was what I guess you’d call a low-end gaming laptop from Dell, borrowed from a friend when I was playing around with VR. The other is the work one, a fairly high-end Lenovo 15″ ThinkPad from 2023—the kind of model people often point to as the PC equivalent of a MacBook Pro. Dear reader, it is not. It has, to be clear, a much better GPU. It has two USB-A ports to go along with its two Thunderbolt USB-C ports, while the MBP has no USB-A ports and three Thunderbolt USB-C ports. They both have an HDMI port; the ThinkPad does not have the SDXC card slot the MBP does, but it does have a TrackPoint mouse, if that’s your thing. But in nearly all other respects, an MBP from 2023 kicks the ThinkPad’s ass twelve ways to Tuesday. For a start, the ThinkPad is plastic. Just like macOS Tahoe, ironically enough, the fit and finish is questionable. When the lid’s open, the gap between the lid’s bottom and the body of the laptop—a gap that doesn’t exist on an Apple laptop at all—isn’t perfectly even. The trackpad is, as always, not quite as nice. And it’s hardly fair to talk about the difference between the screens, given that a 14″ MBP’s screen is a 3024×1964 micro-LED display with a 120Hz refresh rate. Also, the Lenovo’s fan runs audibly even when it’s not using its separate, fancier Nvidia GPU; when it really gets going, it sounds like a hovercraft. You kind of expect it to lift off the desk. I could say a lot of the same things about my M1 MacBook Air in comparison. The M1, introduced in 2019, is only about 10% slower in single core performance than the 13th generation Core i7 from 2023. (It’s about 50% slower in multi-core, but it’s eight cores versus 14.) The Air just has a standard LED screen, but it’s brighter, crisper, and much sharper, with 2560×1600 resolution compared to the ThinkPad’s 1920×1200—almost twice the number of pixels in a smaller display. And the Air is fanless. Yes, this means that if you push it really hard it’s going to throttle down faster, but you don’t buy an Air because you want to constantly push it to its limit. I’m sure that there are Windows PCs that are closer to a MacBook in meaningful ways; the Microsoft Surface line is solid, and even Lenovo makes snazzier machines than this ThinkPad. But I wouldn’t want to run Windows, I’d want to run Linux, and I’d want to run Linux well. None of this “it’s pretty good except you only get half the battery life, the webcam doesn’t work, and if you have this particular WiFi chip getting online requires entering these 28 commands at the bootloader and sacrificing a goat” stuff. And, I’m also sure some readers are frothing at the mouth at my assertion that Apple makes good hardware at all, let alone some of the best hardware. What about the butterfly keyboard? What about heat dissipation? What about the poor DIY repairability? The butterfly keyboard is long dead, along with a few other bad design decisions dating back to the late Ive era; the heat issues are negligible for people who aren’t animating feature films on their laptops. Wouldn’t I prefer a cheaper and more repairable laptop? Sure, but not at the cost of living with an ugly, loud, dim plastic black slab from which no spark of joy can escape. So…?</h2> Unsurprisingly, I’m staying with macOS. I have a quarter-century of inertia behind me at this point, and I remain unconvinced that anyone else makes hardware this damn good. But I’m considering disentangling what I can of my computing life from dependence on Apple’s ecosystem. This will be a pain in the ass in all sorts of ways, and I don’t know how complete it will be. Do I want to move off Apple Music? Off iCloud Drive? Right now all of those feel like horrible ideas. But they could work. I could adjust to Tidal. I could move back to iA Writer, and be able to use Emacs for heavy lifting as a bonus. (iA Writer just operates on folders of Markdown files, rather than Ulysses’ iCloud-specific system.) I don’t want to move back to Dropbox, but it’s there, and it works well, and it’s cross-platform. Yet I want to drag my feet. Maybe next year’s macOS (I’m hoping for “macOS Yuba City”) will be substantially refined. Maybe there won’t be any more roadblocks thrown up to dissuade users from doing radical things like installing “unapproved” applications. Maybe Tim Cook will pull his head out of Donald Trump’s ass. These are all big maybes, though, and we’re living through an epoch where “free software, free as in freedom” may stop sounding like a pithy, slightly annoying slogan and sound like an essential, slightly annoying battle cry. So. I still expect, at this point, to buy an M5 MacBook Pro when the higher-end processors come out. I still expect, at this point, that if the weird writers’ brainstorming app I’m working on ever comes together, it’ll be a SwiftUI app. But I’m keeping my M1 MacBook Air booting into Asahi Linux by default. Just for practice. To support my writing, consider a tip on Ko-fi.com. Seriously, Markdown 2025-08-20T10:51:05+00:00 Every so often, there’s a surge of “here’s why no one can do serious writing in Markdown” articles. They don’t come on any obvious schedule, or with any obvious trigger; usually, somebody complains about something Markdown can’t do that {reStructured Text, AsciiDoc, Org Mode, DocBook, DITA, LaTeX, Pollen, Typst, a Selectric Composer you found at a garage sale} can, it gets picked up on Hacker News, and a few hundred nerds chime in to explain it’s because Markdown {has no formal specification, isn’t extensible, was just designed for blog posts, ran over their childhood dog}. When this happens, I look over from the technical writing I’m doing in Markdown, or possibly the next novel I’m writing with it, and nod thoughtfully. I should, I presume, prepare my resignation letter now, apologizing for not being a serious technical writer, while gently explaining to my current and past clients and employers the error of their ways. Yes, the same people on Hacker News a decade ago were holding up RethinkDB’s documentation as terrific, but if they only knew I’d been writing in Markdown, it would suddenly become unserious. The entire Microsoft Learning Portal? Away with it! And folks I haven’t worked with better be informed, too. Yes, Stripe’s docs are considered a gold standard, but they are written in Markdown! Get the Shame Nun in here, stat! All right, let’s be slightly more serious. Yes, there are countless different Markdown processors out there, and yes, all of them handle edge cases differently. But where they “fail” 99% of the time are edge cases, and if you follow reasonably good practices—the sort of nitpicky things that MarkdownLint</a> whines about—you’re pretty unlikely to run into trouble. I’ve been doing this for a long time, and I’ve never had to learn a “new” style of Markdown or unlearn bad habits from previous work. I know I can’t speak for everyone’s experience. But if you write Markdown like this on the regular: # My Great Document Who needs blank lines after headings; this works _most_ of the time! This next line will surely become the horizontal rule tag I intend. ----- - Bullet 1 - Bullet 2 - Bullet 3, which is not connected to the link below despite the lack of spacing [A [cool website](http://bar.com)](http://foo.com)! I'd really like to **emphasize *this and also** this* right now.</code></pre> And somehow you get away with it until you switch Markdown processors, I’m sorry, but the processor is not the issue. If you write crappy Org Mode markup, you’re going to get crappy Org Mode output. “Yes, but it’ll be the same crappy output no matter the processor! That’s the problem!” Yeah, but is it? Most other lightweight markup systems only have one or two implementations. If you’re using Org Mode, you are almost certainly using the actual Org Mode in Emacs; if you’re using AsciiDoc, you’re probably using AsciiDoctor to process it. AsciiDoc’s formal specification is still a work in progress, and there’s no formal spec for Org Mode’s markup, as far as I can tell. If there were only one or two Markdown processors out there, this probably wouldn’t be an ongoing debate. Of course, CommonMark</a> does give Markdown a formal spec, and in practice most Markdown processors are compatible with it in all but the edgiest of edge cases.</aside> Here’s my theory: what’s at the root of these complaints is that people want Markdown to be something it isn’t. It’s a markup syntax that represents a subset of commonly-used HTML. You can plug it into a content management system or a static site generator, but it isn’t a CMS or an SSG any more than HTML itself is. AsciiDoc and reStructured Text, on the other hand, are markup languages closely tied to specific documentation generators, AsciiDoctor and Sphinx. Org Mode is a plain-text structured editing system for outlining, task and agenda management, literate programming, and publishing to multiple output formats. They’re complete solutions for specific problems. If you have that specific problem and you want a complete solution, you don’t want Markdown. So: why do all these companies use it anyway? Why is your favorite documentation so likely to be written with Markdown behind the scenes? Well, I’m going to go out on a limb here: it’s because it’s a markup syntax that represents a subset of commonly-used HTML, rather than a complete solution. Markdown doesn’t care what content management system you’re using, or if you’re using one at all. How your site is constructed, whether you’re using a template system or a CSS preprocessor or crazy-ass client-side rendering, is not in Markdown’s domain. Use the CMS you’re already using—there’s probably a plug-in. Need extensibility? That’s your site generator’s issue. Write your own damn shortcodes. “Wait,” somebody’s surely saying. “How is that an advantage?” It’s an advantage because you have all of HTML and CSS at your disposal. If you know how to use them, you get exactly what you want, full stop. That’s not a goal of AsciiDoc or reStructured Text or Org Mode, or much of anything else I’ve seen proposed as better alternatives. (It is a goal of Dean Allen’s nearly forgotten Markdown predecessor, Textile, but like Markdown, Textile isn’t in and of itself a site generator.) I’m not positive that, say, Stripe’s documentation site couldn’t be reproduced in rST, but I think it’d be really damn hard, and a lot of that has nothing to do with the capability of the markup system. It has to do with the flexibility of the output. To be sure, you can integrate non-Markdown markup into another system. GitHub renders README files with a variety of lightweight markup styles; some SSGs and CMSs have plugins that let them process rST or Org markup or what have you; you can always set up a workflow that lets you export from your favorite format to HTML or (ahem) Markdown transparently. That’s great if you’re more familiar with another markup style and want to keep using it, but that’s not an argument for superiority, it’s an argument for familiarity. When you focus purely on the markup semantics, it’s hard to argue that any other system is “better” than Markdown. (I find text that uses Markdown markup to generally be more readable in its unprocessed form than most other lightweight systems, and that’s important to me, but it may not be important to everyone.) I’m not saying “give up and just use Markdown.” Being familiar with another markup system is a powerful argument for staying with it, for a start, and there are other reasons to use something else that have nothing to do with markup. There are specific kinds of technical documentation—usually in contexts where a lot of sections are shared between manuals, revised independently, and assembled programmatically—where AsciiDoc is your only practical alternative to full-bore SGML/XML. And as a newly anointed Emacs nerd, I find myself using Org Mode daily at work, and that includes writing extensive notes using its markup system, not Markdown. I’m using it primarily for its task management capabilities, though, because that’s the kind of stuff Org was designed for. But you know what else Emacs is excellent at editing? The lightweight markup I’m actually using when I write developer documentation for one of the most popular games in the world, published by one of the biggest software companies in the world? Here’s a hint: not reStructured Text. To support my writing, consider a tip on Ko-fi.com. It's the Heat and the Humidity; Also, the Fascism 2025-08-05T00:00:00+00:00 “It’s not the heat, it’s the humidity” isn’t as much an aphorism as a punchline, as people point out—often with an unwarranted hint of smugness, as if nobody else had made this observation before—that 95° F is too damn hot no matter how humid it is. And, I mean, sure, but imagine these two scenarios: Tampa: The high today will be 95° F. The low will be 75° F and the relative humidity will be 70%, for a peak dew point of 74° F and a peak heat index of 123° F.</li> Sacramento: The high today will be 95° F. The low will be 59° F and the relative humidity will be 40%, for a peak dew point of 67° F and a peak heat index of 99° F.</li> </ul> For those of you who do Celsius, Tampa is 24° C–35° C, dew point 23° C, heat index 50.5° C; Sacramento is 15° C–35° C, dew point 19.5° C, heat index 37° C.</aside> Is the hottest point too damn high in both cities? Obviously. But in Sacramento—my favorite underrated California metro area—it only sucks for the afternoon hours; the morning is pleasant, the evening is warm, and the humidity stays basically comfortable. Here in Tampa Bay, you wake up to warm, muggy air that turns into a sauna before noon. A sauna with a blistering UV index. In theory, I knew all this when I moved back to Florida. I grew up here! I understood what I was signing up for, right? Again, I mean, sure, in theory. But it feels worse this time around. In part, that’s simple reality: it really is hotter now than it was a quarter-century ago. Summer gets here earlier and leaves later. The storm season lasts a little longer and gets a little more violent. And, some of it’s subjective. I didn’t travel at all, not even a staycation, from July 2023 through June 2024, the longest stretch I’ve gone without a single night somewhere other than home in decades. I’m not actually in Tampa proper—I’m in the rural exurbs, so I can’t even go out for lunch breaks except to a handful of places. The net result has been a level of stir-craziness I haven’t felt in years. (It hasn’t helped that I gave up my “remote work Thursdays” since I moved back, where I picked up my work laptop and relocated to some entirely new place for the day.) Even so, my first two summers back in Florida—2023 and 2024—didn’t engender the kind of restlessness I’ve been wrestling with this summer, the lack of focus, the sense of frustrated wheel-spinning. Last summer was a summer of burnout</a> for me; this is a summer of…well, whatever the nervous energy version of ennui is. But, I know this year it’s not just the heat or the humidity. It’s the sense of helplessness. Fascism is here, a third of the country is greeting it with thunderous applause, and neither the legislative nor judicial branches seem motivated to provide either check or balance. The industry I fell into thirty years ago is in the midst of another bubble, and its leaders are more insistent than ever that it is not a bubble, it is in fact either the greatest invention since fire or an existential threat, while they’re also either providing direct support for American fascism or just sucking up to it. The Venn diagram of “techbros who have gone MAGA” and “techbros preaching the gospel of AI” is more or less a perfect circle, and this is something we, as a society, need to face in a way we’ve been doing our damnedest not to. I am much closer to traditional retirement age now than not; when I moved from Florida to California in 2002, I figured I would be out there for most or all of my working-for-someone-else career. Even as I fell in love with northern California and the Pacific Northwest at the same time Florida became just a little more hot, a little more humid, and a lot more politically antagonistic, I carried that around in the back of my mind. By 2020, though, I couldn’t. There are still places in Florida that, taken in isolation, I’d be happy living in—some neighborhoods around Orlando, some towns in Pinellas County—but the politics aren’t getting better any time soon. Also, despite what our mini-Trump governor claims, Florida is no longer cheap: housing costs here have climbed so fast in the last five years that Sacramento, Tacoma (a suburb of Seattle), and even Portland, Oregon are now cheaper to rent in than either Orlando or Tampa. That’s based on a spot check of rents in April 2025. Just before I actually did move back to Florida, I wrote, “it’s hard not to daydream of moving back even though I haven’t left yet”; I’m still daydreaming of it. I can’t—won’t—leave my mother behind, but even setting aside the Herculean logistics of moving an 80-year-old with chronic health problems (and a cat) cross-country, the places I want to live aren’t places she’s attracted to. On top of that, the house here is completely paid for, on an acre and a half of wooded land that has, so far, proven resistant to flooding. For all the mounting problems with Florida, those are compelling reasons to stay. For now. So that leaves me in a holding pattern, the one I’ve been in since July 2022. I don’t know what breaks it. The uncomfortable spectre with me for all that time has been that I am, in a real sense, here to take care of my mother until she dies, that considerations of where to move to next—where to retire to—are, for the indefinite future, moot. Increasingly, though, the heat and humidity are getting to her, too, and the still-rural community we’re in is getting a lot of house farm development happening. In a year or two, we’ll have the traffic of suburbia while still having essentially nothing urban—or even suburban—around us, the worst of both worlds. Add in two bad hurricane seasons in a row with a third (and fourth and…) likely, and we may hit a practical limit together. I don’t know what happens then, where we’ll end up, what it will mean for me when she does pass. Will I be somewhere I want to stay, in a home I want to keep? But for now, we persist here, together, and watch to see what the next year brings. To support my writing, consider a tip on Ko-fi.com. Dr. Neckbeard, or How I Learned to Stop Worrying and Love Emacs 2025-06-04T09:44:20+00:00 So, you might know that I’m a text editor junkie. At a job in the 1990s, I used a great Windows-only editor called HomeSite</a> for web development. When I moved to the Mac in 2001, I wrote to Allaire, the developers, and asked if there was a Mac version in the works. They wrote back and explained they’d written it in Delphi, a Windows-only development environment, so probably not. But, they added, had I tried BBEdit</a>? So I tried it, and loved it—and got the “always something more” itch. I moved from BBEdit to TextMate</a> when the latter hit version 1.5 and stopped sucking. After giving up waiting for the TextMate version 2</a> unicorn, I tried Sublime Text</a>, which was fine, but never really grabbed me. From there, I’ve been all over the map. Back to BBEdit</a>. In an on-again, off-again affair with Panic Nova</a>. Trying and failing to love Visual Studio Code, despite a certain level of grudging admiration. On the creative writing side, I’ve been through Markdown-specific editors from iA Writer</a> to MultiMarkdown Composer</a>, finally settling on Ulysses</a> for fiction where I’ve mostly stayed despite continuing to play around with Obsidian</a>. And then, there’s the editor I kept trying every so often and giving up on: Emacs. If you don’t know Emacs, it’s a legendary text editor, dating back to the 1970s</a>. What makes it legendary, other than being five decades old and still in active development? A lot of editors have some level of extensibility: BBEdit can be scripted in AppleScript, Code and Nova in JavaScript, and all of them can run shell scripts. But Emacs isn’t just scriptable—it’s an application platform, one specifically built around textual data. It’s not a pretty program, but install Markdown Mode</a> and Emacs becomes a better Markdown editor than whatever you’re using now. (Unless you’re using Emacs.) Magit</a> makes it as good a Git client as the $99 Sublime Merge</a>. And there’s whatever the hell Org Mode</a> is, sort of an outliner that’s also a GTD-like system and a literate programming</a> environment and possibly a floor wax and dessert topping. The problem, though, isn’t just that Emacs is spiritually opposed to being a Mac-assed Mac app (it’s actively insufferable about it</a>), it’s that it has a terrible “out of the box” experience. It’s not just the alien keyboard shortcuts, or the eclectic terminology. (A window is not a window, it is a frame, while a split pane is a window; you do not cut, copy and paste, you kill, save, and yank; etc.) The real issue is that it’s a monumental exercise to merely match, let alone surpass, what you get after ten minutes of extension browsing in VS Code or Nova. Even so, I tried to make the switch, every so often. I wanted to like Emacs. But it always defeated me. My experience always ended with some variant of “it’s been three days and I still can’t figure out basic code folding or word wrap the way every other editor does it and don’t tell me I need to learn Lisp to solve this, motherfucker, I have WORK to do.” A few years ago, “starter kits” became more popular, configurations that you install along with Emacs to have it set up right out of the box. They didn’t help. They might have made things worse. I’ll come back to that. So why keep trying? I have tools that work, right? Multiple things, no less. Why try to get into an infamously difficult editor whose past is inextricably tied to an infamous creep</a>? There are a few reasons. The first is the most straightforward: functionality. As much as I love BBEdit, there are things I do fairly routinely—like edit templates in my static website—it’s not as good for as other editors are, and its extension system isn’t powerful enough to let third parties add new ones. Worse, even out of the box, Emacs clobbers BBEdit in terms of its package management UX. That’s worth repeating: Emacs has a better user experience than BBEdit, a native Mac app, when it comes to package management. Nova has great package management UX, and that’s a major reason its extension ecosystem is far bigger than BBEdit’s even though it’s only been around five years. Yet Nova feels like a bit of an also-ran at this point. Visible development has slowed; many of its third-party extensions, including critical ones like JavaScript/TypeScript, have been effectively abandoned. What about Visual Studio Code? At first, it felt like what Emacs might have been like if it had started thirty years later. But the more you dig into it, the more it’s clear it’s open source the way Google products are: more, you know, open-ish. If you’re not using the official build you don’t get access to everything, critical extensions are closed source, and there’s an increasing sense its real purpose is to lock you into an ecosystem. And that leads into a second reason. I don’t think of myself as particularly anti-commercial—hell, I sell books, I have an idea for software that I would, in fact, like to sell—but there is a definite advantage when something you like, even rely on, is developed by a worldwide community of nerds rather than a corporation. Big companies enshittify; small companies go out of business, or close because the owner wants to retire (see Script Debugger</a>, and even my favorite line of shampoo and shaving products, Nancy Boy), or (ahem) become much more successful as a game publisher than a Mac tool developer. The third reason, though, is more abstract. By approaching it with the right mindset, building up Emacs is—mostly—fun. And Emacs itself is “always something more.” The nerdy itch stops being “do I like this new program that does something cool I haven’t seen before” and starts being “can I do this cool thing I haven’t seen before in Emacs.” The truly annoying thing about Emacs, you see, that it really is as powerful as its proselytizers tell you it is. I couldn’t help but notice that Emacs got a fire lit under its neckbeard a few years back. It has modern features like language server and Tree-sitter support, and—sorry to harp on this, BBEdit—actual package management. Now it includes some popular third-party packages like Org Mode, too. So a couple of months ago, I decided to give Emacs one more try. First I tried a very minimal starter kit, Crafted Emacs</a>. For reasons I don’t remember now, this ran aground yet again. But rather than give up, I restarted from a blank slate, working through the book Mastering Emacs</a> (a purchase I made a few years back in a previous failed attempt to migrate) and building a configuration file up one step at a time, not adding any package or setting without understanding what it was there for. I borrowed some bits from Crafted Emacs, tweaking when I wanted to. Improve the completions and search. Turn on the language server. Add a few languages I regularly use. Bring in TextMate-style snippets. And…this time, it’s kind of working. Almost every time I’d reach for Nova—which mostly relates to blog stuff these days, but also some PHP coding—I reach for Emacs instead. There are some things it already does better, and a few things that are a little clunkier (if there’s an Emacs equivalent to Nova’s beautiful task system</a>, I haven’t found it yet), but not much I miss. For BBEdit, it’s harder to say; there are some ways in which it’s the farthest behind the other editors, but other ways in which it’s neck-and-neck with Emacs as a Swiss army knife for text with a better UX. I have years of little tweaks built up, too, that I use on a regular basis. They’re duplicable in Emacs, but it’d take a while. YASnippet</a> is probably best in breed, though, and Emacs’s keyboard macros might let me build up equivalents to BBEdit’s text factories. I’m using Emacs for technical writing now, too, because my current contract job requires me to use a Windows laptop. In a way, I’ve come full circle. I doubt I’ll ever be all Emacs, all the time, even if this sticks (we’ll check back in a year). Ulysses has lovely ways for organizing and exporting long-form text that I’d be hard-pressed to hack up equivalents to, and of course it works on the iPad—a major feature when I go out on day trips with just the iPad in tow. And while I’m starting to use Org Mode for my work tasks, Things</a> will stay with me as my personal task manager for the same go-everywhere reason. I have a few unorganized thoughts to wrap up. First and foremost, I’ve come to believe that starter kits are a bad idea for new users, which is a problem to the extent they’re presented as just what somebody needs when they’re migrating from a more turnkey editor. As a newbie, I have no idea what the hundred or more packages being pulled onto my system do, I don’t know whether a given error being thrown up is important or not, and when something goes wrong—which becomes ever more likely the more dependencies there are—I don’t know how to fix it. And it’s quite possible that even if I find an apparent solution online, it’s a solution that won’t work with that starter kit. I suspect starter kits are more useful for people who’ve gotten their Emacs sea legs under them already, but I’m already not sure whether I’d want to bother switching. Second, while there’s a lot of great Emacs tutorials out there, there’s surprisingly little that steps you through building a modern setup from scratch, particularly if you’re looking for explanations of why you want to embark your corfu with vertico. I’m emphasizing “modern” because the closest I’ve found, the aptly named “Emacs from Scratch</a>,” hasn’t been fully updated since 2020, and a fair number of convenience packages—things that make Emacs feel more modern—now can be replaced by built-in functionality, or newer, slimmer packages that take advantage of new features. Also, Emacs from Scratch is a video tutorial, which is fine, but rarely my thing. I don’t know if I’m the guy to write an “Emacs Configuration for the Moderately Patient” tutorial, but if I actually stick with it, we’ll see. Lastly, I don’t think Emacs will ever be particularly approachable to folks who aren’t willing to hack around with their editors in a more involved way. And that’s okay. I admire the work being put into modernizing it, but the very idea of an extensible, scriptable text editor is, well, pretty nerdy. Either you’re going to get really into the notion, or you’re just, well, not. And even in that realm, Emacs is unabashedly uber-nerdy. MacStories’ Federico Viticci, famed for championing scriptable editors like Obsidian and Editorial, would be all over Emacs if it wasn’t, well, Emacs. I get it. But the thing is, I think I also get Emacs. And once you get Emacs, there’s probably no going back. To support my writing, consider a tip on Ko-fi.com. Maybe the Waiting Is Actually the Easiest Part 2025-05-12T15:45:00+00:00 I’ve written before how there are two opposing views of large language models and generative AI, with little to no space between them. In one view, they’re already an essential tool and anyone who isn’t wholeheartedly embracing their use in day-to-day work might as well be insisting on keeping their job as a telegraph operator. In the other, they’re a maddening case of the emperor’s new clothes—costs are only going up, the frequency of hallucinations is increasing rather than decreasing</a>, and the sooner we admit that OpenAI is the next Enron, the sooner we can all get on with real work. Opposing viewpoints about new technologies is hardly unheard of, and it’s more difficult than we like to believe to not only spot winners and losers, but even to make generalizations. Most new technologies fail and world-changing technologies are hard to spot aren’t mutually exclusive. Are they both true? Kinda? Maybe? Just because boosters say a technology is world-changing doesn’t mean that it is, but that doesn’t make nay-sayers automatically correct. What’s weird about generative AI, though, is the seeming rise of a third position—one that accepts most of the criticisms of AI while nonetheless taking it as a given that we should be incorporating it into our workflows wherever possible. Call it, perhaps, the “AI triumphalist” position: it doesn’t matter whether the ethical concerns, or even the practical business considerations, are ever addressed—it’s the future, like it or not, baby. A surprising number of people whose opinions I (usually) enjoy reading/listening to seem to have adopted this: they acknowledge all the problems, then go on to use ChatGPT for coding, incorporate it into their shortcuts, and so on. So far none of them have used the image generation tools, as far as I know, but it’s quite possible that’s only because these folks are mostly in the Apple-centric space, and Apple’s “Image Playground” is a dumpster fire even by generative AI standards. I can’t help but wonder if the proof by assertion fallacy</a> is in part at work here. When we keep hearing anything repeated over and over, eventually we can’t help but wonder if we’re missing something vital. Every tech company is racing to AI all the things, from our code editors to our image editors, our search engines, even our keyboards and mice. Surely, surely, there must be something to this all. When we try it ourselves, it’s promising, isn’t it? The responses to our natural language search queries are just what we want, confidently presented, and correct like nine out of ten times. The message summaries only occasionally screw up in laughable and/or horrific ways. The header images we’re creating for our blog articles might not be what was in our head, but they’re interesting and colorful and you have to look closely to see the mistakes, and hey, we wouldn’t have paid a real artist to do this in the first place. That medical diagnosis definitely sounds authoritative, and look, what are the chances those wild mushrooms are that poisonous, really. So when I hear pundits and analysts look at what Apple is (and isn’t) doing in the AI space, I often see it just taken as a given that Apple has to do this. They have to shove AI desperately in all the things, because they’re way behind the curve. The delays in bringing the magical Star Trek-level Siri they pretended they were just about to ship might be existential. The quality isn’t the issue—this is the future, goddammit, and they have to be there yesterday. The AI triumphalist position is the dominant one. And yet, there’s still no indication that hallucinations are a solvable problem. There’s no obvious path to profitability for OpenAI (even their paid plans cost them more money to run than they make), and there’s virtually no “AI industry” without OpenAI. What if the triumphalists are wrong? I’m not (necessarily) suggesting this is a binary, that there will be no value in generative AI, let alone value in other machine learning that, for right now, is unfortunately lumped in with it. Even in the anti-triumphalist case, there will be use cases for large language models. But in this “worst” case, people realize that more often than not, the time they have to spend checking their amazing robot assistant’s work reduces or even eliminates the subjective benefits. And companies quietly start backing away. Google rolls out an “improved” summary for results that applies machine learning in different, non-bullshitting ways. The Copilot brand sticks around, but it gets quietly de-emphasized. Companies whose business model is “repackage somebody else’s LLM” disappear, unless they’ve found a genuine niche. Companies whose business model is monetizing LLMs, like OpenAI and Anthropic, get bought or, in some cases, just fold. (Since LLMs as they exist today are largely fungible, building a moat around such a business is incredibly difficult.) In this scenario, Apple’s mistake won’t prove to be ignoring AI for as long as they did, it will prove to be not continuing to ignore it. Adding a few carefully focused features here and there, like ML-assisted image cleanup and better text prediction, sure. But there are just so many other more pressing issues—from declining software reliability and concerningly poor UX decisions of late to what on God’s green earth they’re doing with the Vision Pro—they should be prioritizing. So could Apple’s foot-dragging end up being accidentally genius? Maybe. They’ve always pushed “machine learning” as their brand, and if it’s in their best interest to stop talking about LLMs, they’ll do it in a heartbeat. While Apple rarely kills popular or even cult hit software and services with the wild abandon that Google does, stuff that never took off quietly gets murdered with no fanfare—Image Playground could easily go the way of Ping and Music Memos. I suspect WWDC is going to be a big tell this year. The rumors that have people buzzing with excitement, dread, or skepticism aren’t rumors about new “Apple Intelligence” features—not even rumors like “they’ll finally ship what they announced at the last WWDC.” I don’t subscribe to the idea that the rumored new huge UI changes across the line are an evil plot to distract us from Apple Intelligence’s failures, but it’s surely not lost on anyone in Cupertino that a year spent talking about the bold decision to change the UI font to Zapfino or make application icons non-Euclidean or whatever is a year not spent talking about Siri. And that might just mean that if an LLM-backed Siri doesn’t ship until mid-2026, it’ll be entering a market which no longer breathlessly expects generative AI to cure cancer, replace all software developers, and animate Oscar-winning movies over a lunch break. That might just be a better market for all concerned—including Apple. To support my writing, consider a tip on Ko-fi.com. The Robots Have Come for Your Em Dashes 2025-04-14T11:11:00+00:00 Did you know em dashes are a sign of AI writing? Since when? Well, that’s what all the LinkedIn influencers say! Humans don’t use them in writing, right? They use hyphens. Of course we use em dashes—they’re basic punctuation marks! We use em dashes as a pause for emphasis—like I did in the first sentence in this paragraph—or to set off a quasi-parenthetical phrase, like I just did in this sentence. A hyphen is used to form compound words, or to break words by syllable at the end of lines. They’re not the same. Nobody better tell the influencers about en dashes; they’ll spontaneously combust. So where’d this idea come from? You got me, but I’d guess it stems from the nature of AI-generated text. It always shows a certain level of polish: no spelling or grammatical mistakes, complete sentences, a clear organization and structure. LLMs get this right in their “first drafts” because the majority of text they were trained on is itself polished—and it turns out that this level of polish covers the correct use of em dashes. Is there a better way to detect AI-generated text than looking for em dashes, then, smart guy? All generative AI is, by literal definition, statistically median, and this isn’t something that can be overcome by creative prompt engineering. Flair and personality in human writing comes from individualistic, quirky choices born of a writer’s likes, dislikes, and life experiences. LLMs don’t have any of those. So LLM-generated text tends to read like pleasantly anodyne business memos, or book reports by a precocious ninth-grader. So if the text you’re playing “robot or not” with is a business memo, a high school paper, or a blog post by the kind of LinkedIn influencer who thinks em dashes are a sign of AI writing, it might be a challenge. In other circumstances, it shouldn’t be. But it apparently is! The problem—and, dear reader, there is no gentle way to put this—a lot of ostensibly smart people neither write particularly well, nor read enough to get a sense of what good writing is like. Most of the writing they deal with—memos, emails, text messages, Amazon and Yelp reviews—is slapdash, unpolished. So if they come across anything in those contexts that looks refined yet relatively generic, they suspect ChatGPT. Apparently, we’ve so come to expect a lack of writing style from humans—and I mean style in the sense of Strunk & White, not in the sense of actual style—that we’re suspicious of it when we see it. If you don’t want your writing to be mistaken for AI, what should you do? Use “fuck” more often. To support my writing, consider a tip on Ko-fi.com. What Makes an App Feel “Right” on the Mac? 2025-03-31T10:45:00+00:00 On Mastodon recently, I wrote: As I play around with more code editors: Zed is pretty neat, although it pushes “AI code assist” to a mildly annoying degree—and despite technically being a native app, it’s no more Mac-like than Neovim is. </blockquote> I got an interesting response from Fred McCann: What do we even consider to be a native Mac app? Cocoa/AppKit? SwiftUI? Catalyst/UIKit? Apple ships apps that don’t feel like Mac apps. I guess I’m already accustomed to dev tools being weird and cross-platform. </blockquote> So I’ve been thinking about that. The UI toolkits are obvious answers, but I’m not sure that they’re completely satisfying. For instance, I love MarsEdit</a> in theory—enough that I bought version 4 and the upgrade to version 5—but in practice, it’s never really stuck for me. Granted, in no small part that’s because I keep doing things like redesign my website using Zola, a static site generator that MarsEdit doesn’t work with. But even back when I was using Tumblr and, later, Micro.blog, it turned out that being a Mac-assed Mac app wasn’t enough for me—the aesthetics mattered. I could never make MarsEdit look as good to my eyes, be as pleasing a long-form writing environment, as I could iA Writer, Ulysses, or my old stalwart, BBEdit. What’s startling to me, though, is that another writing environment I’ve come to really like is the one I’m using now: Obsidian. Which is not, in any sense, native. So it’s possible that the right question—at least for me—isn’t “is this app using a native UI toolkit,” it’s “is this app a good Mac citizen.” In other words, does it embrace long-standing Mac conventions? The standard menus (Application Name, File, Edit, View, Window, Help) should be present, in that order, with the expected submenu items.</li> Settings should come up when you select Settings… or hit ⌘,. Settings should appear in a window, not a tab (and definitely not a text file that opens in a tab).</li> Everything in the Services menu should work as expected, including when it’s invoked from the context menu.</li> Any text field should work the way it does everywhere else in the system. It should respect the subset of Emacs key bindings that all Cocoa text fields do, it should respect new key bindings you’ve put in the system key bindings folder, text replacements you’ve defined system-wide should work, smart quotes should be made smart if you have that turned on globally (and left stupid if you do not), and so on. Text editors that have their own key binding system get a partial pass here.</li> The general “fit and finish” needs to be Mac-like. Icons and symbols, even if they don’t come from the SF Symbols</a> library, shouldn’t look like they came from Windows, GNOME, Mac OS 8, or Mars; the same should be true of UI widgets and form elements. The UI font should be San Francisco, or at the very least something similar.</li> </ul> If we hold things to this list, programs like Nova, MarsEdit, and Apple Pages—canonical Mac-assed Mac apps—all do unsurprisingly smashingly. But Obsidian, the Electron-based program I’m writing in right now, does shockingly well, too. Microsoft’s Visual Studio Code doesn’t do quite as well (most notably, it opens its settings “window” as an editor tab), but it does better than Sublime Text (which opens a text file for settings), and much better than the banana crazypants menu and icon design of the cross-platform e-book management program Calibre. On the flip side, being 100% native is no guarantee. The long-running story plotting app Dramatica Story Expert</a> was native, yet such a poor Mac citizen that it not only violated several of those bullet points for decades, its fonts somehow, incredibly, never rendered at retina resolution. And modern apps written using Apple’s own UIKit-based Catalyst</a> frequently break the Services menu or text fields or both. Also, there’s the insanity of the new Settings app on the Mac, which replaces checkboxes with toggle switches, hides functionality behind the circled-“i” icon that traditionally means “information”, and groups preferences less by well-defined category than by, I don’t know, astrological signs or something. Don’t worry, though! The search function is (checks notes) absolute bollocks. What about web apps, you might ask? I’ve been thinking about Federico Viticci’s recent post about the iPad’s “Sweet Solution”</a> with respect to “feel”: In working with my iPad Pro over the past few months, I’ve realized something that might have seemed absurd just a few years ago: some of the best apps I’m using – the ones with truly desktop-class layouts and experiences – aren’t native iPad apps. They’re web apps. </blockquote> This frustrates me because he makes a good case in that article—and for me, at least, that’s a problem. While it’s possible to make apps that are relatively good Mac and iOS/iPadOS citizens using web technologies, that’s not the same thing as making a web app, an application that runs entirely in the context of a web page. I’m not against web apps—Google Maps and Google Mail are both canonical examples of good ones (setting aside for the moment one’s feelings about Google as a political/corporate entity). But at least for me, document-based web apps are usually a bridge too far. Unless multi-user collaboration is essential, give me Apple Pages and Numbers over Google Docs and Sheets any day and twice on Tuesday. So, I’m not sure where this ultimately leaves me. I’m likely to keep using Obsidian and keep failing to use MarsEdit. I’m likely to keep rooting for Panic Nova over Visual Studio Code (I’m using Nova to update the website, and run preview and deployment tasks). Yet I’d take Code over Sublime Text</a>—although I might just take a well-configured NeoVim</a> over either. It’s not like it looks any more out of place on the Mac, and hey, you can’t kvetch about it having incomplete Vim emulation. To support my writing, consider a tip on Ko-fi.com. Having to Think About Essential Computerness 2025-03-24T11:35:00+00:00 Years ago, a friend of mine who also uses Macs told me a story about why another mutual Mac-using artist friend chose the Mac over the PC some three decades ago. His technophile friends—of which he had many; this was Silicon Valley in the 1990s—often showed him what they were doing with their computers. The Mac users said, See what I did with my computer? But the PC users said, See what I figured out how to get my computer to do? And so he got a Mac. This has always been crucial subtext not for Mac lovers as much as for Mac haters. The specifics change over the years, but even the nerdiest Mac fans tend to appreciate that, by and large, they don’t have to think about the essential “computerness” of their Macs unless they choose to, while even the most casual PC fans—not users, but power users—tend to be inherently suspicious of any computer that doesn’t make its users think about that. See, every PC fan knows at least one less tech-savvy PC user in their family or their workplace whose system is regularly FUBARed because they don’t understand things that are, to the power user, very basic shit. And then there comes the goddamn Mac, saying, “Hey, relax, you shouldn’t need to understand that. Look, I didn’t even have a command line for the first fifteen years of my existence! Have some kombucha. Kick back.” And it drives them absolutely bananas. Okay. I suspect the power users I know (and don’t know) are already seething at me, but look. I ran Windows 2000 on my home PC and used Windows in various flavors at work. I installed Linux back when getting a GUI running on it required you to find the dot clock values your video card supported to enter them by hand in the XF86Config</code> file. I’ve edited binary files by hand in hexadecimal on a TRS-80. What I’m saying is, I have walked the walk. And, man, after OS X came on the scene, the most amazing thing happened: now Macs were Unix workstations—and still Macs. To this day, I still don’t think I can get PC users to really understand how flexible and extensible the Mac became under the hood once it became the successor to NextStep. (I have found that PC power users tend to judge customization on “can I reskin the windows” and Mac power users tend to judge it on “can I automate it”; I’m pretty firmly in the latter camp.) I can do everything on the Mac that I could back when I was running Linux and FreeBSD. But it’s still a Mac. This isn’t about being a walled garden—the Mac isn’t one, and I’m tired of uninformed or cynical arguments to the contrary—it’s about not finding myself trying to debug mysterious problems after the most recent major system update, or just because it’s been a year or two since I did a complete reinstall of the system. “Things will always eventually get weird and slow, and you should do a clean reinstall” has been advice for going on forty years in the Windows world. But a disquieting thing has been happening over the last few years—some might say as long as the last decade, but I think I’d put it right around macOS “Big Sur” in 2020. That’s when the OS version number finally moved from 10.x to 11, the UI was redesigned in small, subtle ways (not all for the better), and Mac hardware started transitioning from Intel CPUs to Apple Silicon ARM-based CPUs. More and more often, I have to think about my Mac’s essential computerness now. Because sometimes things that were working just stop. Or don’t work right at all. For the first year and a half, the Studio Display itself occasionally had to be rebooted—the monitor had to be rebooted, because it runs goddamn iOS. For the last few months, I’ve been getting weird glitches with USB audio on occasion, including my external USB speakers just not working until they’re disconnected and reconnected. Lately, I’m frequently chided that I should have remembered to eject the backup drive from my laptop before disconnecting it, but there is no backup drive: it’s backing up to another Mac over the LAN. Most recently, in Big Sur, some of Apple’s new fancy screensavers and dynamic wallpapers just stop being dynamic until you kill their process to restart them. And, like I said, I can walk the walk. I created a little Shortcut called “Kick wallpaper” that lets me restart that process easy-peasy. I was proud of that for the thirty seconds it took me to remember that I wouldn’t have to have written that little Shortcut if it wasn’t for Apple’s bug, one that they haven’t fixed in any Big Sur update. What’s worse, sometimes the system just gets weird and slow. An application doesn’t launch at all, or some background task that should be running clearly isn’t. Desktop widgets stop updating, and since widgets have been the first thing in decades that’s got me finally embracing a task manager—that’s a future post—that widget better be current, or my task system falls apart. Hazel starts reporting weird errors, and after debugging, I realize Spotlight indexing has stopped. Why? Who knows. Will rebuilding the index fix the problem? Maybe. Would anyone but a power user even be able to debug this? Doubtful. Maybe they could just log out and log back in again, but I’m not sure most Windows or Mac users would think of that. They’d just reboot and hope. Or do a clean reinstall. I’m not going to pretend I’m going to switch to Linux any time soon, let alone Windows. There’s too much I’d miss. While I could find rough equivalents to Mac-only applications like BBEdit, Nova, Acorn, Ulysses, and Marked, there’s a coherence that I just don’t see anywhere else, still, after all these years. The Mac is, for me, still a better platform. But I can’t help but wonder if modern Apple’s obvious inattention to software quality—there’s no kinder way to put it—is letting an important differentiator, potentially the most important one, slowly vanish. Now, once in a while, I sit down in front of my Mac and realize I’m not thinking what can I do with my computer. I’m thinking what can I figure out how to get my computer to do. To support my writing, consider a tip on Ko-fi.com.