fate 1.1: * fate is now also a GraphQL client 🤯 * SSR & Cache hydration contributed by @RelevantElement You can now write fate views on the client and fate seamlessly turns them into real GraphQL queries. fate.technology/guide/graphq…

Want to learn about the new Solid 2.0 primitives? I got you! brenelz.com/posts/learning-s…

AI is massively increasing the quantity of all digital products/information. Papers, books, music, etc. It's all being buried. Great, or otherwise. Our current methods of sifting, mostly involve relying on the opinions of others, in some way. The obvious solution is: More AI. However first, people actually have to trust the opinions of their AI, and care about their suggestions. Will this happen? Idk. Maybe groups you trust/have a pre-existing relationship with, will operate ai curation services? Is this the future of marketing?

I predict: The first major ai lab with a SOTA diffusion language model, will attempt to keep it under wraps, and stream the tokens in order.

Nobody is impressed by a resume. If you have to provide one, you've already lost. For even the most accomplished people, after you're presented with their resume, they'll seem slightly less impressive because if it. Theres just something low status about selling yourself this way.

What is art? Besides justsynthetic data generation?

I believe that any future AGI will natively require the concept(s) of ontology, and objects built-in to it. It must possess a desire/instinct to identify objects, and use them. Objects arise from tracking causality. How one thing affects another. The more abstract/higher level the object, less work one would have to do the effectively predict the future, it's more efficient, not having to do the bookkeeping of lower-level objects. Also objects give the models points of interaction, where the can intervene, and experiment. If you don't have objects, you can't experiment, play. Learn. I think current models don't have robust ontological instincts. And this is highlighted by their performance in arc-agi-3. Which very clearly requires an instinct to discover objects, and how they relate. @arcprize

Effect systems effortlessly fulfill the Thielian criterion of 10x supremacy. All they needed was a mainstream language; and one miraculously and unwittingly became a viable host vis-à-vis the illogical aim of typing the Untypable JavaScript. Now we wait 🧘

Consumers wont get a tariff refund. But in the end, we paid much of the cost. Wild.

I think labs training to solve arc-agi-3 , has the potential to lead to much worse emergent misalignment. if it has to divinate/invent/pursue goals from the context alone.

Every day, I see incredible pieces of software shared on X for the first time. It's unfortunate that most will not be utilized to their full potential.

Solid skills github.com/doeixd/solid-skil…

The Taj Mahal without the towers is mid.

Serving Super-Intelligence seems like a bad business to be in. Inevitably, your users will use the super intelligence to build their own, cheaper super intelligence. Right?

I think this is the future.

Here ya go! Feel free to check for yourself. I could've done something wrong. But I don't think that TypeScript is inferior. Or that the code is significantly more complex. Or that the user experience is any worse.

Tada! one-billion-checkboxes-space…