As for me, I'm enjoying working on my little social warrior projects ;)

I'd like to hope my thread earlier was worth people's time in its full detail, and people are responding like it is, so that's reassuring. But I think if it needed a tl;dr, I figured out a good one. Half an hour later, of course, as these things tend to go!

If you don't consider open source software to be activism, you will write open source software that doesn't do activism.

(not everything works this simply, sad to say. If you don't consider open source software to be political, you won't end up with apolitical software, it'll just be software with *bad* politics)

That's about all the brain I had left for today. Thanks to @401matthall for the inciting conversation that led me to this train of thought, and to @ireneista and @onepict and @futurebird and my little sibling and so many others I could never exhaustively name for being good spiritual influences on me. I consider myself very fortunate to be in that caliber of company, and I'm not shy to say so.

So you literally make better software if you think like this from day one, you actively use your creativity and imagination to scheme how you can make other people's lives better as the end goal that makes any of this matter. It's a head start. And you don't end up pulling a Linus Torvalds later, or even a Tridge.

I do think software as activism is higher effort up front. But it pays off pretty much continuously from there, often reducing your effort while increasing your positive impact. It's very mutual. One of my big things I think about lately is how to make software that requires little to no maintenance - in the ideal case, permanent - so that I can spread myself to a wider scope long-term. I can't do that if I'm firefighting all the time! Which - you see now *how* it makes you better, as a worked example, right?

What we're seeing in LOSS (Libertarian Open Source Software) is that we built an empire on that luck, which seems impossible until you realize all the survivor's bias. Most projects were useful only to the author, or nobody at all. But there were so goddamned many of them, the barrier to entry was so low, that some of them happened to be aligned with other people, and then you end up with projects becoming popular, and eventually critical infrastructure, where the duty of care downstream is given belated importance by someone who never fostered it as a skill, and who is also worn down by a vocal minority of users bringing their entitlement to the ticket tracker.

And it all feels so different than making a cool tool for your mom, right?

If you don't care about other people, you can't be good at this. Not in a way that really matters. Maybe you scratch your own itch and some other people are able to reuse your code. That feels more like luck than talent - their needs just happened to be similar enough to yours that you didn't have to respect them, they were able to coast on the respect you had for yourself.

In the long run, that's really fragile. You play that game long enough, you lose, because it's gambling. And you don't have to gamble, or at least, you can stack your odds.

You can talk to the stakeholders and understand, not just the stuff they think to say, but the nooks and crannies they wouldn't without exploration. You can study the technology and deeply understand what it can do and where its limitations are. Tilt the table. You owe the house nothing.

That brushes up against something I've talked about before. Humans have messy language because that's actually the more competent tool for navigating a messy world, the ambiguity is fit-for-purpose. Machines operate in a much stricter logical paradigm where nothing is ambiguous, because that's what you need for reliably stacking abstractions. Programming is the work of translating between these languages which do not have the same raw concepts available to them - a translation that can never be perfect because neither side is perfectly expressible in the other, so you need human judgment about what makes for the best translation, like keeping the heart of a poem intact while picking words that rhyme in the target language.

And, in that thread, I talked about how LLMs suck by delegating this exact part to a machine which cannot be "aligned" in the way that a human programmer can. But a human doesn't automatically have that alignment either. It's a skill. It's an ethos. It's a responsibility.

It would be impossible to list all the people I consider to be good technical influences on me, but because of a conversation earlier, I had to think about it and realize that the people I've found most useful in shaping my technical taste are the people who have had the most *spiritual* impact on me.

Which... I dunno about that phrasing, I'm not a very woo person. But it's the same thing I often call "politics" or "values." It's very much using the moral engine part of the brain.

I think... everything else flows out from putting yourself in the shoes of the people and machines impacted by your work. You need to have the human best interest at heart, and a respect for the machine's time and energy too. Because what is the work except mediating between those two parties, right?

"Come on, old girl. One last run, for old times' sake. Let's show them what you've got left in you." I say as I gently pat my mugful of used coffee grounds from this morning like a grey-muzzled mare.

Context: https://mastodon.social/@deech/117468103327297109

Anyway I hope they like it 👉👈

I don't even go to lobste.rs but I made them a meme out of the wellspring of generosity in my heart

The best reason to do any home pickling in the morning is that you can have the windows open to vent the heat and smells out of the house.

The second best reason is it gives you an excuse to tell your partner "rise and brine!" and make a stupid smug face at her until she hits you with a pillow.

Honestly if I reimplement the underlying code someday, instead of just wrapping a real libc (total "someday" project but defensible to do eventually), I could really make progress on one of my minor dreams - a post-libc world. In particular, dropping a boat anchor on locales from orbit.

And since that statement always brings a peanut gallery - localization is good, very important actually. C locales, though, are the "congratulations, buddy, that's the worst anyone's ever done it" meme of localization.

Man, if I really wanted to advertise other projects to rely on `prn-sys`, which it's structured to support (but still a bit too feature-light), I should make the logo a wrapped condom and add a tagline "you've seen the standard for libc - don't use it without protection."

Censorship resistance: good usually, but old news. Everybody already understands the value of that.

Sponsorship resistance: conceptually novel, as yet unappreciated, yet clearly a vital thing to explore to ensure software sustainability...

The "ugh, who moved the scissors and forgot to put them back? It better not have been me" to "godDAMMIT, me" pipeline

So let's actually look at the API endpoint documentation, which does actually render on Dillo, but requires scrolling through the entire left sidebar nav before any content, because "hey, we're Mastodon. We can't give TOO MUCH of a shit." Reminds me of the guy complaining about his burrito being organized like a pack of lifesavers, but I'm digressing, just... look.

https://docs.joinmastodon.org/methods/bookmarks/#get

Now you'll notice that it has these `min_id`, `max_id`, and even `since_id` params, which are reflected in the `madonctl` CLI. Like, this maps to using the `--max-id $sid` options on the command line.

You will also notice this, not really explained, PC_LOAD_LETTER-ass notice: "Internal parameter. Use HTTP Link header for pagination." Okay Zelda, but what do you mean by that?

2/n

Okay. So I think I finally understand why Madon/ctl can no longer paginate through favorites. Minor gripes about the Go codebase aside, it's not their fault. It's upstream Mastodon. And it's really stupid. And the fix would be so invasive and interface-breaking that I think it will be the thing that makes me bring API access in-house into the #Moonshine single-file script.

So. Storytime, I guess.

1/n

I mentioned this in a private conversation a few days ago, in a bit of a tough love intervention, but I think I need to say it here too.

Allowing technology to be the territory of fascists, is complying in advance.

(this also needs to be softened a bit with - god I know we're all tired and burnt the fuck out and not operating with our Best Brains Ever right now, so I'm sympathetic to the impulse, I just think we should be resisting actively where we can muster the strength to do so)

Also since my head is full of thoughts today in ways that keep interfering with code authorship, here's another thing.

I intend to be curator and primary author of an extremely expansive stdlib, where the goal is to treat non-stdlib dependencies as an antipattern. I intend to be lead dev on a package manager and an operating system. And I am an abject critic of managing FOSS projects by committee, preferring a visible leadership model, for reasons that are like the flipside of KYC: you should, for trust reasons, know who makes your upstream deps. I am also, if the rest of this post (or my writing in general) didn't give it away, a strong personality that won't be everyone's cup of tea (which is fine).

That could, if unexamined, be a pretty bad recipe for making myself an ethical single-point-of-failure. So, I examine.

1/n