Seems weird to pick a name that is so much a noble gas, and then not have the other names follow it. Is that because Nitrogen doesn't have parameters enough? Or Neon being a weird model name if it came to that? And why not Helium, instead of Hydrogen? Is that supposed to indicate rock-bottom pricing? What about the cheap cheap 3.1-flash-lite for pipelines then? Will they phase it out, and replace it with Hydrogen as the new lower price range? None of it makes sense!

Isn't that because Chrome is forced to use the same rendering engine on iOS? Whereas on macOS, Safari is total garbage across all of Google's sites and products compared to Chrome, for very obvious reasons.

Yeah, the sand looks more like a fine sequence of stop-motion frames than physics based sand.

The trade-off between speed and quality is everything though. I've found that semantic searching everything that might be relevant and dump the top N into the model is way faster at getting a workable result.

So this is a different clustering algorithm, of which there are many, maaaany?

That makes it... not very interesting, because it's not a drop-in replacement then, and potentially many other clustering methods would have worked in particular applications as well. Or does this one beat most of the others as well?

It's not a new kind of clustering. It solves exactly the same problem as k-means (minimize the mean squared distance to k centroids), returns the same thing (k centroids and an assignment), and has the same inputs (k and a number of epochs). The batch size and step size are set automatically, so there's nothing extra to tune. So it is a drop-in replacement for k-means specifically: in Faiss it's Emeans(d, k).train(x) where you'd call Kmeans(d, k).train(x). The code is in faiss at https://github.com/facebookresearch/faiss/tree/main/contrib/...

Okay I’ll just pretend to not use any of the gazillion tools and products that tries to coerce me into a deal against my informed consent. Surely I’ll beat the system.

People use AI whether they believe it or not. Think you found a hidden deal, and hidden cheap flight, leading you to a hidden matcha cafe with a hidden view? Think again! The booking company which is owned by a booking company which is owned by a booking company wanted you to find it.

I’ve found plenty of hidden things by just wandering around in the real world - I’d bet most of them weren’t pushed by AI.

Prompting the wrong things is just a skill issue.

I do it from time to time when I’ve run out of tokens at the end of the month. But even then I’ve created a script to pack multiple code files and their dependencies into a single txt, with a tree overview.

And of course you never change the test suite in the process… That would be bad developer practice, LLM or not.

Gemini 3.1 Pro can easily work with this effectively, if not a bit slower than an agent.

> But even then I’ve created a script to pack multiple code files and their dependencies into a single txt, with a tree overview.

Like:

    $ cat $(find project/ -name '*.py' -type f)

?

While admirable in vision this is just the masses tearing each other apart, with the feudal lord 1%-ers just looking on.

While tearing down Adobe you’ve probably paid Nvidia, Anthropic, and Microsoft in the process, while at the same time no amount of Claude will make a dent in any of these behemoth strongholds. Adobe, evil as they are, aren’t the ones controlling the world’s knowledge, flow of goods or compute.

This is almost comically similar to “the immigrants are taking our jobs and are eating our pets” and how in-fighting about this has enshitified the real world, while not having solved any single problem along the way.

It’s like overthrowing governments with machine guns and guerilla armies. Yes, the playfield has been leveled. Literally.