Dan Finlay

@danfinlay.com

Just futzing around

I don't know how anyone ever gets away with claiming they're the first to do anything. It carries a huge assumption of omniscience. In practice, being the first to do something often means you just did something for the first time, which is a more juvenile but still commendable state.

I remember the first time a friend ever used the phrase "load bearing" with me. We were at a house party, and someone was like "wow, it would be great if this wall wasn't here", and he said "you could just take it out as long as it isn't load-bearing", and in that moment I learned a little bit.

There’s an improv saying that I usually attribute to James Thomas Bailey of ComedySportz LA: “An improviser’s knowledge is a mile wide and an inch deep”. Maybe that background contributes to why this very combinatorial era has been fun to me.

Citizen Platano 🇵🇷@daniloc.xyz · 2w ago

going to be an emotional crisis for lots of *specialists* folks decided to go all in on narrow depth and the rewards from that are going to be under steady threat a diversified portfolio of skills and interests, meanwhile, enjoys steady returns under this paradigm compositional problem solving

The other side of the coin is that sometimes you are aware of a security issue that you need to urgently patch. I’m not convinced that a publishing delay system will be a clear win for npm security. Does anyone have the data on both sides of the issue?

Paul Jensen@paulbjensen.bsky.social · 3w ago

@socket.dev notified of the compromised Astro package (7.1.0) within the GitHub dependabot request, so didn’t merge that PR. I think they proved their value there, and given the package was published only 9 hours ago, that is more proof of the need to delay npm package updates.

I like Fable. I don’t like the games Anthropic plays to try to lock its users in: Constantly moving trial dates, subscriptions that don’t apply to custom tools. These are not what a mature stack will look like. I hope their financials don’t depend on these kinds of shenanigans & they get over it.

You’d think that Jevon’s paradox would saturate our demand for paradoxes, but in fact we see only a growing hunger for paradox.

Modern social media is basically a distributed reply all pocalypse. Each replier foolishly believes they can put the madness to rest by extending the thread one more post.

Inference clubs definitely deserve to be a thing. A little tricky to rally 100 people who are all willing to trust a host. That host has to be going above and beyond to build community credibility, or have other ways of networking a larger group..

eva (^_^)/@eva.computer · 2mo ago

if it costs like $350,000 for 8 Blackwells to locally run an open model with the capabilities that are equivalent to Opus, then i wonder if we will see something like small private clubs for inference like 100 people putting in $3500 for access to a private-ish AI run by a coop or club

As a guy who once turned a duck-sewage-green pond clear, I'd just share that my mental model is that you're going to have plants as long as you have unspent nitrogen in the water. If I were them, I'd throw in a bunch of those nitrogen-unfixing enzymes.

BildBild

Don't you see? When you use AI models, you *take the plagiarism back*. "Good artists copy, great artists steal" - Picasso Be the thief, not the thefted.

I had an RIAA crossed out bumper sticker in high school. It was cool to be proud of piracy, ie cool to flaunt your ability to share and consume data freely. The ethics are probably like "if you're an underdog, nobody will begrudge you for it, but if you're the only one making money, then UBI."

Thorne 🌸@ens0.me · 2mo ago

IT'S SO WEIRD after spending my formative years watching people rightly criticize organizations like the RIAA for suing teenagers for downloading MP3s

Btw, "Iroh bluesky PDS" is exactly the type of idea where any technical user is like "yes, I'd like the devs to do that", but traditionally would fall onto a dev team that is already underwater keeping up with the basics to be considering adapting to new shiny stacks.

Dan Finlay@danfinlay.com · 2mo ago

Iroh is one of those "everyone should just adapt their network stacks to support that as soon as possible" moments (iroh bluesky PDS?), and AI is going to make the phase shift happen faster than ever.

Finally looked up a painting guide for the TNG type II phaser model I bought at a convention when I was in like 3rd grade (my dad was awesome). I did NOT expect my "speed paint" to end up in the guide for allowing those grooves to pop in the handle.

before, my childhood pass: Sharpied the black areas (honestly not terrible, but colored the peaks black instead of the valleys are on the show), white-out on the buttons, with some gold paint-marker on the thumb button (long worn off). I'd never painted the body before, it was cast in a resin that was close enough for my child self who had no paint kit.I went back and forth between speckling dove gray and aluminum. In this picture, I think it's a bit too aluminum. I went back and forth until I thought it had a bit of a "cast aluminum" texture.Here after I did a black wash on the grip and painted the top buttons. Still need to black the front.

Iroh is one of those "everyone should just adapt their network stacks to support that as soon as possible" moments (iroh bluesky PDS?), and AI is going to make the phase shift happen faster than ever.