Music theory is a tool for analyzing music and, optionally, helping make it. Treating analysis and composition as inverse operations is going to mislead you. There are useful analyses in music theory that would be completely alien to the person whose work it's based on.

Don't most of those use AWS?

apparently it shows lot of reports for AWS on downdetector ..

>> "ChatGPT Chat gets a fresh filesystem for each chat session. These cannot be accessed from any other session."

That's what the Library is for. You can use GitHub to keep it outside ChatGPT. GitHub also provides a Chat-based option for sites.

OpenRouter->Stripe (~$7B)

Hugging Face->Nvidia ($13B)

Keep an eye on Exa. With local models and niche harnesses proliferating, they're going to need a way to search without everyone having to run their own index.

Google is an obvious buyer if they could get it past regulators. A Doubleclick-level addition of the agentic stuff they're bad at to the indexing they're good at.

Nvidia is the obvious less conflicted #2

Secret third option: Cloudflare. They're already the glue for so much of this stuff.

On the one hand, SV-speak has trained me to read acquisitions as an end of good stuff. On the other, if this ATProto experiment really is something different, we need to be willing to consider that old patterns don't hold and expect something better.

I like the optimistic track.

one fundamental issue in atproto is the missing permissions layer, the proposal is insufficient for many use cases, graze was shut out of their main financial path by Bluesky (paid creator posts, Bluesky cut off talks)

tl;dr - people want control and privacy, not an open anonymous read all the data in the world, foundation for their online social activities, and this needs to be core to a protocol

I've also been holding off from jumping into building on AT proto for exactly that "no one wants all their data to be as public as possible" reason, but private data is now in alpha which is a meaningful step towards solving that: https://atproto.com/blog/atproto-spaces-alpha

its not a meaningful step, its a bare minimum (ACL) token acceptence of reality, I don't get the sense they actually believe in private data on social

its designed in such a minimal way that the app devs will have to implement on top for their features and create a centralization point

this is why I believe it needs to be baked into the core, part of the vision and ethos from the start

Dolly Parton seemed nice on that episode of The Orville.

It's nice when using large sample libraries so you can stream them off the drive without as much caching in memory. Local LLM workflows probably also benefit from being able to run models too big for RAM off the drive.

I bet this is a lot of what is driving it on the DC side, and just in general bringing disk closer to RAM as far more conventional database type workloads.

ChatGPT Work and Codex use that. Normal chat has a different, unspecified limit.

Yep. I'm actually using ChatGPT integration via exe.dev to write code. It uses my subscription through an API and runs down the usage limit in a similar way.

The part I don't understand is how usage is "charged" against this limit.

Sol has held stuff for a while to do the same sort of hazard checks I assume Fable is doing, but it always releases them. I think that's the better way to handle it rather than preventing me from seeing how far I can get generating schematics to use in Minecraft. Currently: a mostly normal voxel house.

OCR was slow and unreliable and was for a very long time.

"slow" - wasn't like they were pressed for time. It took them almost 20years to even start the investigation! Also not certain you needed that much reliability compared to what was available to match one pic to a quality scan to get a reasonable number of candidates