@dale Nobody’s locked in any rooms, though

@dale Then don't use that server, there are lots of servers. You have choices. Find one that's run by someone you agree with. It doesn't have to be yourself, but it does have to be someone - because of course servers don't spontaneously appear. And people who run servers have opinions.

If that's elitism, I don't know what to tell you? There's no ideal platonic network of free-floating nodes without people running them

@ubernostrum @dale To me, that just means folks are choosing to build a selective network. Let them defederate and build a pocket universe, you probably don't want to talk to them anyway.

#

Les Orchard

@benbrown I read the book, the book is good, then I bought the book. This seems like reverse order? But I did it anyway 🤷‍♂️

@benbrown Ahh, you're right, that's what I was missing!

@dale the very vocal folks behind the campaign can be loud and defederate, meanwhile a lot of other folks have not. They can’t make sure there’s no choice, there still is choice

And I guess pick your poison: devops experience paid by advertisers or devops experience with other motivations? That’s our story so far

@earthshine Who's got two tail lights? This guy 👉 👈

Sam Bankman-Fraud ruining big hair for all us dudes

Passport selfie; eye contact

@wohali it’s not exactly a fro, because I tried taming it for this passport photo last week, but behold my awful visage. Also i just got a haircut, so it’s relatively small again

re: Passport selfie; eye contact

@wohali But to be fair, it doesn't really fro when it's long, it kind of einsteins

random thoughts about Ted Nelson, AL, & ML

Kind of want a new Ted Nelson to write an "Machine Learning Lib / Dream Machines" subtitled "You can and must understand Machine Learning NOW"

So much of the same stuff applies like "cybercrud" and whatnot around how folks hide behind the Great and Mighty Computer and its Algorithms. And yet, computers didn't go away in the 70s.

And you can say "LLMs are just spicy autocomplete" - but also you should understand precisely the degree and flavor of the spiciness. Because this stuff is astonishingly spicy & capable even while also being based in part or whole on obvious fuckery.

The fuckery doesn't mean it's entirely a sham. I don't think it's going to flame out like crypto / web 3 stuff. I do think it's worth understanding fully and dragging into a more reciprocal & equitable shape. I don't entirely know how to do that.

en.wikipedia.org/wiki/Computer

re: pet health +

@socketwench I was once engaged in regularly syringe feeding a cat with a prodigious sinus infection on the floor of our shower for about a month, so I'm rooting for the Shelby over here

@leah pizza slices really need a Clone trait

@roadriverrail @futurebird Just saw a tiktok where someone said they woke up in the middle of the night to a cryptid / their personal sleep demon standing at the foot of their bed - and their only thought was "you don't scare me, I have to go to work in 5 hours, flit off"

AI/ML; LLM; retrieval augmented generation

So, I'm slowly working to catch up on like the last 10 years of AI/ML shenanigans. Maybe I'll blog about it.

The latest thing I'm scratching my head at is Retrieval Augmented Generation - or RAG. Near as I can tell, despite all the (probably LLM generated) words spilled about it, this is RAG:

You submit a detailed question to a program. The program performs a search through a pile of indexed documents using the text of that question. That turns up a bunch of roughly relevant documents.

The contents of those matching documents are then all just copypasta'd into a new submission with your original question appended to the end. (many optimizations glossed over here)

This is then run against an LLM, which comes up with an often surprisingly relevant response informed by all the copypasta'd documents.

Like, again, near as I can tell, the secret sauce is search-plus-copypasta. Seems so dumb yet it does interesting things?

re: random thoughts about Ted Nelson, AL, & ML

@tomjennings Yeah, that's the real cruddy part and why I don't think it can just be ignored as a whole

re: AI/ML; LLM; retrieval augmented generation

@datarama That's where I'm like LLMs are currently like oracles on psychedelic drugs. It's possible and maybe likely they come up with useful results, which is astonishing. But also sometimes they're garbage-in-garbage-out machines

@SuperSelena64 Oh, that's neat! Somehow I've never seen an N64 transfer pak before

layoffs; mental health

Processing giddy survivor's guilt by eating fistfuls of off-brand habanero doritos over the kitchen sink for lunch

re: layoffs; mental health

Shoving half a cheese brick into my maw while I enjoy a 1000 yard stare out the back window

re: layoffs; mental health

@mhoye just a little one, as a treat

@JoYo Those cat ears are the only cosmetic item I ever bought in Halo Infinite

re: layoffs; mental health

Me from decades' past would have been outside burning through a pack of smokes. But, I quit, so at least I've got some healthy choices going for me

@electron_greg At some point I'm going to have to get all my old computers in the basement simultaneously running this

@Jetengineweasel tell me the meeting at least had an agenda

@Jetengineweasel i am seeking my fainting couch at this moment

@waxy ohhh noooo, I've been resisting the NES-inspired keyboard but this one may be my undoing

re: AI/ML; LLM; retrieval augmented generation

@kellogh FWIW, along with stuff from @simon, I will also blame your Fossil effort for some of my catch-up, too. 😅

It's very much in the neighborhood of things I want to be doing and has led me down some rabbit holes with exciting terms like "k-means clustering" and "embeddings" and "llm summarization" as applied to a fediverse firehose.

@benbrown A floppy version didn't quite fit into my finances at the moment - but a) it's a real neat concept and b) I'm real happy to see you sold out! (hope you end up doing something like it again!)

re: AI/ML; LLM; retrieval augmented generation

@kellogh I've been kind of backburnering learning a bunch of this stuff for ages. But now it all looks like a neat set of legos to cram together

That thing where I didn’t realize how much I needed a new pair of walking shoes until a) I stepped in a puddle and my toes got wet from below and b) realize I feel an inch taller in a fresh set of the same shoes since the soles are meant to have treads

Like oh hey there are these grippy protrusions on the bottoms of the shoes that are really quite thick 🤔 Huh, what’ll they think of next on footwear?

3 body problem spoiler

@schlink Kinda the Red Wedding of 3 Body Problem 😅

2024/03/29