Notices by Cassandra Granade 🏳️‍⚧️ (xgranade@wandering.shop)

By way of agreement, while I'm not a professional programmer any more, I have been one for much of my life, and in the year 2024, I'm not even curious about AI.

In addition to all the obvious reasons and moral issues, AI doesn't offer any meaningful abstractions over any problem domains. Abstractions are nice because they allow you to manage complexity: they leak, they can be undone and peeled back to see that complexity again.

AI offers a lie: hard things are easy.

https://hachyderm.io/@sunfish/113521920268793303

I love making hard things easy, using abstractions to automate complex tasks. But that's not what AI is. I can't peel back that complexity, I can't use the output of an AI to generalize to unanticipated examples of complexity. The boilerplate code emitted by an LLM will always be boilerplate instead of a code generator that abstracts out what makes that code boilerplate in the first place.

"Disabled by default on enterprise PCs."

It's worth examining why tech corporations and tech press seem to agree that it's OK if personal privacy is obliterated, so long as corporate secrets are sacrosanct.

https://cyberplace.social/@GossiTheDog/113517702610785853

I never thought I'd say this, but I miss Win32-style UIs. Electron is nice for what it's good for, but it does mean I have to relearn basic UI vocabulary in every single application. I lost the link to it, but there was a great article I saw a little while ago cataloguing how much even check-box vs radio-button vocabulary has completely broken down.

Oh, fuck me, this is bad. Like subprime mortgages kinda bad.

https://pivot-to-ai.com/2024/11/19/pumping-the-ai-bubble-a-data-center-funding-craze-with-novel-types-of-debt-structures/

I'm sad that so many folks leaving Twitter are heading to Bluesky. It really feels like we've learned nothing about centralized corporate social media. (And yes, I consider it to be centralized as long as blocking bsky.app doesn't just take down a plurality of the network, but the whole bloody thing.)

Part of what makes that as sad for me is that I know turns into people who will pressure me to leave the fediverse. That that kind of pressure is considered ok.

It makes me sad for the same reason it makes me sad that it's OK for folks to pressure me into using WhatsApp or Discord, but not for me to ask for folks to use Signal or Matrix. That it's OK for folks to tell me I need to use Word or Windows, but not for me to ask for patience with using LibreOffice on Linux. That I get made fun of for Zoom having bugs on Linux, or for suggesting Jitsi when we hit our hour- long limit.

@noracodes I get that's not the main point of your thread, but also, it is weird to me how much the bulk of the nominal left just crumbles the moment that questions about tech come up.

@noracodes "base decisions ... on ideology"

Then what in the fuck else is an ideology for? If you give up on the idea that ideology can guide behavior and decisions, you've given up on every power your ideals could ever have held.

That's a fucking *weird* thing for progressives to do, especially right now.

@noracodes

> "You're just [doing thing] because you're ideologically committed to being independent from corporate control" is not the sick burn some people seem to think it is.

That, on a t-shirt, made by a unionized shop (or even better, worker-owned co-op).

@bstacey

This is why I have zero trust in Bluesky. It's proceeding exactly like every other VC-backed corporate-funded social media platform I've ever used, starting with overt good intentions, a lot of promises to users, and absolutely no technical infrastructure to help ensure that those promises are kept.

The main difference is that Bluesky could actually build a decentralized network instead of a centralized platform. To the best of my knowledge, they haven't done so yet.

https://hachyderm.io/@molly0xfff/113414584052666033

I should add that there's another piece to this as well. While I've seen no shortage of centralized platforms fail, and am thus personally weary and frankly too tired to try and kick that football again, I *also* think there's progressive value in not having centralized ownership over human communication.

Right now, the best way I know how to achieve that is with decentralized protocols and infrastructure.

https://www.xgranade.com/a-progressive-case-for-the-fediverse/

That doesn't solve everything, not nearly. But I think it's an important progressive value, and I do think progressive organizing and activism in general needs to be more proactive about resisting centralized *ownership*.

Musk and Zuckerberg have shown the harm that centralized ownership can do to other progressive values, to be sure.

Again, it's not the complete story, but I do think it's a *part*.

Update on my ongoing struggle to bring some semblence of reality to the prompt engineering article on Wikipedia:

I finally found a peer-reviewed article supporting the wild claims made on that page! But it was peer reviewed by a conference on prompt engineering with a PC comprised almost entirely of AI vendors.

So yeah.

That's not automatically invalid, of course. Specialist fields can and should have their own venues, but it definitely raises a lot of flags in such a new field when a conference is mostly run by vendors.

The knowledge basis in this case is mostly written by individuals with financial stake in positive results and peer-reviewed by colleagues with similar incentives.

That's not wonderful.

@aud My whole thesis was basically on how to rigorously evaluate the quality of control in quantum systems... I hear you.

No shit, if you train a stochastic parrot on biased data, you get biased output? Say it ain't so!

On a somewhat related note, my efforts to help improve Wikipedia's coverage of LLMs is again stymied by the problem that few people write scholarly articles explaining that the moon is not made of green cheese.

The few that do are absolute fucking *heroes*.

Said effort is also not helped by the truly absurd amount of spam hyping up AI, and the incredible SEO efforts thrown behind said spam.

If you search for "what is prompt engineering", for instance, not one fact-based source shows up in the first three pages of DDG for me (searching in a private window). It's *all* AI hype by either LLM vendors, startups that want to sell you "libraries" of prompts, or courseware that takes the above credulously.