Cassandra is only carbon now

Sometimes I write intimate eschatologies or words about technology and math. Sometimes I make things by burning them with light or squeezing them through a small, hot tube. Sometimes I push water with a stick while sitting in a tiny boat.

Public Key

npub1lpkcu8qk89l4g2f8ggklfevfzfz62rdlauhw4yvzusgu7j6rch9sap7yxc

NIP-05 Address

xgranade@wandering-shop.mostr.pub

Profile Code

nprofile1qqs0smvwrstrjl659yn5yt05uky3y3d9pkl77th2jxpwgyw0fdputjcpz3mhxue69uhhyetvv9ujuerpd46hxtnfduqs6amnwvaz7tmwdaejumr0dswpd4e9

Publishing to

relay.damus.io

nos.lol

Author Public Key

npub1lpkcu8qk89l4g2f8ggklfevfzfz62rdlauhw4yvzusgu7j6rch9sap7yxc

Show more details

Last Notes

2026-03-27 16:39:17 UTC

I'm not going to link to it, it's disgusting as hell, but JKR has *again* gone after Imane Khelif, hours apart from her lauding the HBO reboot trailer.
Support of her show *necessarily* validates not only her transphobia, but also her using transphobic hate to justify racism.

2026-03-22 20:30:02 UTC

Emoji reactions in chat apps should allow you to react to them, nesting like LaTeX subscripts every time.

2026-03-22 19:26:40 UTC

I've criticized — a *lot* — Python and the Python ecosystem writ large for its insufficient opposition to and even outright adoption of AI.
But I want to make one thing absolutely clear as I do so: I don't recommend ditching Python, nor do I endorse or co-sign any calls to do so. I make the critique *because* I don't think we should ditch Python, but because Python is an amazing tool with a community that has repeatedly shown a willingness to do better tomorrow than yesterday.

2026-03-20 21:39:22 UTC

I hate CMake.
I also hate CMake slightly less than I do building C and/or C++ without CMake. Slightly.

2026-03-20 08:44:40 UTC

"Power users" is doing a lot of work in Google's blog post.
You don't need to be a "power user" to deserve to have access to computing without monopolistic interference.
https://chaos.social/@grote/116257002625921666

2026-03-20 05:31:56 UTC

A few thoughts on Astral / OpenAI, now that the emotions have sat for a bit.
First, let me start by noting that AI is an attack on open source, inherently, by necessity, and at a structural level. That argument is bigger than Astral, but the short version is that you cannot simultaneously expand the public commons and work towards it's enclosure; moreover, if the public commons do not stand for the public good, then it's not really a commons any more.

2026-03-20 03:12:40 UTC

Fred Meyer thoughts: why is it legal to use lethal force to protect corporate property? From a proportionate response theory of ethics, it's never justified to escalate from shoplifting to murder, but that's exactly the threat presented by having armed cosplay cops at every door

2026-03-18 21:32:31 UTC

- reply

@nprofile…k95v It's pretty terrible!

2026-03-04 20:12:28 UTC

- reply

@nprofile…6z3g Posted an unkind reply and deleted, sorry. I'm getting frustrated with the whole AI thing today, and I'm not being my best self. I should probably just step offline for a bit.
This is just so... frustrating.

2026-03-04 19:27:45 UTC

- reply

@nprofile…6z3g It looks like this was the PR?
https://github.com/chardet/chardet/pull/322
Even aside from the ethical and moral issues with LLMs, it doesn't seem optimal that a 15k line PR affecting almost a million dependent repos (if GitHub's count is to be believed) was up for three days before getting merged in.

2026-02-28 02:05:35 UTC

- reply

@nprofile…mv7s That is a very, very, *very* low bar.

2026-02-28 02:01:43 UTC

- reply

@nprofile…mv7s I don't mind them squabbling, no, but this isn't a case of the enemy of my enemy is my temporary ally, they're both enemies. Anthropic are not the good guys, they're not neutral parties, they're straight up bad guys.

2026-02-28 01:12:16 UTC

People cannot be trusted to use "megabyte" as a word. Too many people use it to refer to 1024 and not to the correct SI meaning of "mega."
This can be solved by introducing "kidi," "medi," "gidi," and so forth meaning "kilo (decimal)," "mega (decimal)" and so forth. These new prefixes are completely synonymous with kilo, mega, giga, and so forth, but have the advantage that sloppy developers from the 80s haven't misused them yet.

2026-02-27 23:58:41 UTC

I like Erin's reporting a lot, but I really need to hammer this point: Anthropic did consent to participating in the mass surveillance of Americans. They just didn't do so on terms that Hesgeth liked.
Pitch this as good-guy Anthropic standing up against bad-guy Hesgeth at extreme peril.
https://fed.brid.gy/r/https://bsky.app/profile/did:plc:m65ifh7vn5zdgs7izcmht4gy/post/3mfuvvrf2pk25

2026-02-26 01:40:33 UTC

Car culture means that if you can exclude someone from having access to a car, you can exclude them from huge swaths of public life.
I read the terrifying news out of Kansas in that light.

2026-02-25 00:59:09 UTC

Ever since @nprofile…qa6s pointed me to it, the term "semantic ablation" has been stuck in my brain, not the least of which because it builds on the near tautology that human communication is high-entropy by nature.
https://www.theregister.com/2026/02/16/semantic_ablation_ai_writing/

2026-02-20 22:08:55 UTC

No, opposing LLMs isn't "purity culture." I've seen this now from quite a few different people, and I disagree vehemently. It is good, actually, to have moral principles and hold to them, even when people with more money than you find said principles annoying.

2026-02-02 21:47:24 UTC

- reply

@nprofile…3qha I offer solidarity in the form of "wait, *how* many quantum computing people are in those files?"

2026-02-01 11:38:27 UTC

- reply

@nprofile…vnlz By the time someone gets to needing hacks like changing $IFS at runtime, it should be obvious that something has gone *wildly* wrong with how small do-one-thing-well kinds of programs are glued together, and yet we're still having to deal with fucking Bash as some universal glue in 2026.

2026-01-15 08:13:36 UTC

Just got approved for Codeberg CI!

2025-12-30 20:42:12 UTC

- reply

I did not give permission to whomever is painting hamburgers out of edible wax to try and trick me by making me think that hamburger is what I will get when I fork over $15 at a drive through. I did not give permission to anyone trying to language-hack my brain Snow Crash style until I buy their product, sign up for their service, give them my money.

2025-12-30 20:39:28 UTC

If you play D&D or read a book or watch a movie or play a video game, you're giving consent to have your feelings manipulated. That consent is revocable — you can stop playing, put the book down, turn the movie off. There's real and serious issues with how that consent is negotiated and communicated, whether your consent is informed, and the like. I don't want to gloss over those, only note that there is some form of revocable and conditional consent offered.

2025-12-15 21:20:38 UTC

- reply

I get it, Instagram is where the followers are, do what you need to do. But then please, *please* do something else, too. Don't make it so that people have to give Zuckerberg all their personal data just to even see what cool things you're up to!

2025-12-15 21:19:23 UTC

I am begging, begging, begging cool people making cool things to please have any presence on the web other than Instagram.

2025-12-15 19:40:35 UTC

- reply

Rather than, you know, this world, where we continually choose to keep COVID going by not masking, not vaccinating, not doing in-office work, and not ventilating.

2025-12-15 19:38:25 UTC

It's always weird how so many movies and TV shows seem to be set in an alternate universe where either COVID never happened, or it did and we chose to end it.

2025-11-30 05:43:22 UTC

- reply

@nprofile…f4xf flow?

2025-11-26 08:50:38 UTC

- reply

@nprofile…4cck That's incredibly scary, not just directly, but also in the context of queer erasure, manufactured moral panics about CRT, and every other thing the right wing wants to censor out of existence. OpenAI has given them an immensely powerful tool for doing just that.

2025-11-15 20:42:55 UTC

- reply

@nprofile…f4xf Is that how those old hacks for hooking up portable CD players to in-car stereos over short-range FM transmission worked? Makes a lot of sense how you could use the same technique for silent raves, yeah!

2025-11-10 20:28:59 UTC

It's kind of amazing that a Republican-controlled presidency, Senate, House, and SCOTUS managed to cause a shutdown, then Trump used the shutdown to tear down about a quarter of the literal White House, proceeded to starve literal children, and yet somehow Schumer managed to lose *on the optics*.

2025-11-07 03:48:14 UTC

My search skills are failing, but trying to find an essay I read a few years ago about understanding gen-alpha humor through the lens of historical artistic movements embracing surrealism in response to fascism. If anyone happens to have a link, I'd very much so appreciate it!

2025-11-05 19:02:54 UTC

- reply

@nprofile…xl7t Solitare wasn't included so Windows users had a way to fuck off from work, it was designed to train people how to use a mouse.

2025-11-01 19:53:53 UTC

I've made this point as a joke before, but I want to be serious about it, too. I really think Home Assistant has become a kind of personal- or family-size SCADA platform, more than specifically a, well, *home* thing.

2025-10-30 22:50:52 UTC

- reply

@nprofile…6yhc Taking license activation servers offline ought to be illegal, or at the very least, doing so should constitute allowing people to reverse engineer and circumvent license action steps.

2025-10-25 20:54:19 UTC

- reply

@nprofile…mrzh Are those CADs for assembling things with Legos or for printing your own bricks?

2025-10-07 18:08:51 UTC

Weird question, but does anyone have the Win95/98-era icon for *.cab files handy? I've tried, but have had poor luck trying to find that very specific icon.

2025-05-15 20:28:34 UTC

Huh, I wonder why this Firefox extension hasn't been updated in the better part of a decade.
*checks README on GitHub*
"This repo dedicated to the living memory of its original author."
Well, fuck.

2025-05-13 20:40:37 UTC

"...we should be skeptical of techniques which seek to secure things against ourselves."
—shattering "passkeys" as a word, @nprofile…g7cz
https://irenes.space/leaves/2025-05-10-shattering-passkeys

2025-05-05 20:48:32 UTC

- reply

Was Hitler like that, and history as told by his sympathizers just... normalized that away? Is this unique to Trump, that we're supposed to just ignore him talking about windmills and showerheads and Alcatraz and whatever else at completely random intervals, with zero reference to what he's ostensibly supposed to be talking about or indeed, with zero reference to any context whatsoever?

2025-05-05 07:28:26 UTC

There's not a lot of daylight between this kind of religious ecstatic fervor for AIs and the NYT article that concluded that, because ChatGPT output a string containing "I am alive," ChatGPT must be sentient.
https://merveilles.town/@lrhodes/114450638623375146

2025-04-24 01:16:05 UTC

If you depend on Discord, I recommend building an exit plan for in case the new CEO goes all OKR-brain on it and makes it unusable or even more hostile than it is already.
That might not happen! But if it does, the best time to prepare is *before* it happens.
https://finance.yahoo.com/news/discord-names-humam-sakhnini-ceo-161500361.html

2025-03-31 20:58:38 UTC

I can't help but think of visibility as a kind of social contract.
If you want to see us, then you need to be able to be there for us when the people who want us to die also see us.
We get to be visible for our own sake, of course, but if you encourage visibility, much less demand it, please be prepared to offer the support that comes along with your side of that contract.

2025-03-25 06:51:46 UTC

If you run a program that's deterministic, then by definition, your experiences with that program will be typical of its expected behavior. There's some important nuance there about making sure that the input you provide is the input you mean to provide, hence reproducible builds, containerization, etc.
But still, that's by contrast with programs whose behavior is inherently stochastic. There, it's easy to fool yourself into believing the program "works" when you just got lucky.

2025-03-25 03:40:24 UTC

If you are an AI proponent, please consider: what would falsify your claim that LLMs are useful?
To be sure, the burden of proof is on those claiming a new technology is useful and can be applied ethically, but even just for a moment, consider what it would take to convince you *personally* that your claim is inconsistent with observed reality?
If there is no such thing, if you are always able to shift to a better future LLM to avoid criticism, then that might mean your claim isn't empirical.

2025-03-25 02:22:34 UTC

- reply

That comparison goes even deeper, extending into political imperatives as well. With apologies to @skiles.bsky.social@bsky.brid.gy, the cosmology underpinning AI hype has fully merged with that underpinning creationism.
https://bsky.app/profile/skiles.bsky.social/post/3lkwcffxrzk2u

2025-03-25 02:17:46 UTC

- reply

I've made the comparison between AI hype and creationism before, namely in that both require ascribing agency *without evidence* to what is ultimately an unthinking stochastic process. But the comparison between rhetorical styles and techniques is important as well, right down to the use of citation cliques to manufacture the appearance of academic consensus around a strongly non-consensus position.

2025-03-25 02:15:05 UTC

LLM shilling really strikes me as a form of Gish gallop. If you point out that they don't work as advertised, suddenly the conversation is about how they could work one day. If you point out that there's no theoretical or empirical basis for that expectation, suddenly the conversation is about "democratizing" coding skills. If you point out that there's no evidence LLMs actually accomplish that goal, suddenly the conversation is about AGI or something.

2025-03-20 21:26:24 UTC

Apparently my one book and most of my academic papers are in the libgen database, all under my deadname.
So, yeah. Seems like a lot of what LLMs regurgitate about quantum computers will be trained on misgendering me.
Fucking wonderful.

2025-03-19 10:23:12 UTC

Men will literally act like LLMs are alive instead of going to therapy.

2025-03-14 19:40:01 UTC

I forget where I first saw this analogy, but describing LLMs as a broken clock just makes so much sense. You cannot scale up or improve LLMs to address fundamental limitations with the technology any more than you can make a hyperscale broken-clock data center to more accurately tell time.

2025-03-11 04:43:41 UTC

- reply

OK. It doesn't work in Vanadium, so it's worse than just Chrome-likes only, it seems like it needs to be specifically Chrome, maybe?

2025-03-11 04:35:41 UTC

- reply

Oh, lovely, it doesn't work on non-Chromium browsers. This keeps getting more and more fun.

2025-03-11 04:34:29 UTC

- reply

"Please remain on this screen. Exiting this screen will require you to start over."
Oh, that's a code smell of the highest order. Jesus.

2025-03-11 04:22:23 UTC

- reply

And that's even pre-Musk bullshit, just garden variety fuckups from collusion in the technocratic sphere.

2025-03-11 04:21:33 UTC

I *love* that in order to log in to a government service like the IRS, I need to accept some random tech company's ToS.

2025-03-10 16:37:27 UTC

Planning on hosting a conference in the US? Great, there's lots of people here who are effectively stuck and who can't leave; it's awesome to have events that we can go to.
Just know that international attendees will need to face the possibility of indefinite and extremely cruel incarceration in order to join your event.
https://mastodon.ar.al/@aral/114138219554678749

2025-03-04 01:48:12 UTC

Is there a BitTorrent-style alternative to `docker pull`? Should there be?
https://feed.hella.cheap/@bob/statuses/01JNFDBDD10BAQ6Q22QA51PH4B

2025-02-16 01:00:54 UTC

- reply

@nprofile…6m2n I just wish more projects could agree on whether help is short-form -h only, long-form --help, both, or something weird like -?.

2025-02-11 23:23:54 UTC

I used to see the AI bubble and trans rights as distinct issues. I no longer do. The fascist movement in tech has truly metastasized, as evidenced by Elon Musk's personal coup, his endless supply of techbro supporters, tech companies' eagerness to axe DEI programs once Trump gave them an excuse, erasure of queer lives from tech products, etc.
To the extent that AI marketing is an attempt to enclose and commodify culture, and thus to concentrate political power, I see it as a kind of fascism.

2025-02-07 05:09:32 UTC

No messaging app on a phone can be more secure than the phone itself, so why bother with Signal?
This is why threat modelling is so important: there are absolutely a *lot* of threats that Signal protects you against, even on a phone that leaks too much data to Apple and Google.
It is both true that Signal is very worth it, *and* that to get the most out of it across the most threat models, your phone itself should be as secured as you reasonably can get it.
https://kind.social/@EnfysBook/113958146467673520

2024-03-26 23:20:10 UTC

- reply

@npub1pfe…e2dm Yeah, I didn't get how running things through a split--tunnel VPN broke HTTPS unless they were doing some fuckery with root CAs, but then why wouldn't the Snap app catch that? Definitely something odd...

2024-03-26 22:14:00 UTC

- reply

@npub1pfe…e2dm Also a great example as to why e2e encryption is needed, not just HTTPS.

2024-03-02 20:13:29 UTC

If I could magically correct one misconception about quantum physics, it would be that Schrodinger's cat isn't intended as an explanation of quantum mechanics, but was him mocking it. Oh, and that he was a prolific child abuser.

2023-09-04 18:20:02 UTC

- reply

@npub14cw…hfwc You don't make anything safer by adding arbitrary extra footguns that make it easier to make mistakes. A safe power tool is one that fails open, not that fails in ways that cause surprising and unanticipated behavior.

2023-09-04 18:14:47 UTC

When a language is difficult to use correctly, it forces programmers on to support networks and into communities that often fail marginalized people; the "git gud" ethos of most C++ developers is testament enough to that. I would argue that by necessity, an inclusive language is also an easy language to use, because difficulty is a kind of gatekeeping.
I hope I haven't confused necessary and sufficient.

2023-08-26 01:51:04 UTC

Is there a word for when you tap the wrong thing while scrolling and now YouTube thinks you're obsessed with whatever it was?

2023-08-16 02:45:09 UTC

- reply

@npub1pfe…e2dm The same could be said for implicit and/or dynamic typing still being typing...

2023-08-15 08:33:32 UTC

So, this is dystopian as fuck.
https://seattle.eater.com/2023/8/14/23832221/downtown-spirits-liquor-store-just-walk-out-tech-amazon
https://stockroom.wandering.shop/media_attachments/files/110/892/705/225/198/866/original/66ce5a00a87fadcd.jpg

2023-07-15 23:06:31 UTC

- reply

@npub1pfe…e2dm At least it's SemVer: Web 2.0 truly did break Web 1.0.

2023-07-15 22:41:28 UTC

I swear, recent discourse has me about this close to writing a tech-manifesto and calling it "Web 1.9." Like I feel like I genuinely want a pre--Web 2.0 kind of experience but with the benefits of many of the technologies that generally fall under the Web 2.0 umbrella.
At its best, that's kind of what the fediverse feels like to me: a transposition of phpBB / webrings / Yahoo!-style directories to the modern world?

2023-07-05 03:58:51 UTC

- reply

@npub1utv…87lr I don't know yet, sorry... still try to find a new one. That all happened a few days a go when I tried to open a new card with them, so it's all pretty fresh. I loved the hell out of my AmEx cards, to, so I'm pretty bummed about it.

2023-07-05 03:52:40 UTC

- reply

@npub1utv…87lr At least one trouble for the customer end of things, as i found out in a sequence of three very insulting and transphobic calls, is that AmEx seems to consider "has changed name at some point" to be the same as "deadbeat."
In several hours on the phone with them, I could not convince them that not only do I have a long credit history, I have a long credit history *with AmEx*.