Profile avatar
perceptions420.bsky.social
Generating Entropy
274 posts 233 followers 317 following
Prolific Poster
Conversation Starter

I feel the biggest econ stake for the EU will be *not* being in a recession while the US are. Even just a standstill growth would signal a decoupling and completely change the debt/money exchange landscape in the years to come.

i want bluesky links everywhere i want to go to X and see nothing but links to better bsky conversations i want to flip on CNN and have watch cringey reports of bsky trends like “what Mark Cuban said today”

Cyberpunk got so much right. But one thing it got wrong was "in the future, corporations are the muscular power that shapes reality, and states are nearly irrelevant." On the contrary, tech corporations prove incredibly ductile — falling over themselves with eagerness to propitiate state power.

A well-written overview of why sociopaths rise to the top in American society. drafts.interfluidity.com/2025/02/25/a...

Throwing a sieg heil during a speech is now a required shibboleth for maga figures

FYI. DeepSeek will be open-sourcing 5 repos, starting next week, sharing their small but sincere progress with full transparency. These humble building blocks in their online service have been documented, deployed and battle-tested in production.

The degree to which X has become a propaganda cesspool is frightening. In any Ukraine-related thread, the fraction of hardcore pro-Russia accounts is what, 90%? 95%? And the large majority aren't even actual people.

certain books, mostly lengthy philosophical political books with snappy titles that are super dense and honestly kinda sloppy, all their goodreads reviews are like "an urgent message, very important book, everyone must read" and i'm like 80% sure that these people didn't read it

Most people believe Probability is unintuitive when what they really have a problem with is counting (combinatorics). Enumeration in context is really hard and thank heavens we have computers to handle it for us.

Seven years later, this is still true.

I think the fact that the media searched for any and every reason to antagonize, and malign tech and tech Cos often without valid reason to(as Yglesias points out) very much led to the set of scenarios we witness today. Francois is right.

Liberal democracy won't stay alive very long if it doesn't start co-opting technology and business, rather than alienating them. They're the two most important vectors of power today. They are actually synergistic -- the rule of law is vital for business. Public education is vital for tech.

People rationalize their suffering a bit too easily.

I think its best to trust Karpathy on everything that is NOT Elon related. I think we saw the best instance of how easily he sides with Elon in conflict with the whole OAI situation.

Big Tech is vocally betting the farm on soon achieving human-comparable AI, while doing absolutely nothing about the consequent societal earthquake. It's like they're saying they're on track for faster-than-light travel and are talking exclusively about faster grocery deliveries

I think any hype generated by the new instances of Grok models will be overshadowed by the sheer cringeworthiness of its training corpus and system prompt as well as OAI releases Altman might be planning.

poor craftsman blames his tools. good craftsman blames his ancestors

Return on investment? A machine can work all day and night, doesn't need to eat, sleep and doesn't get sick. Complex problems that might take eons will be and currently are being solved(Protein folding as an example) in far less time than was humanly possible. I understand the empathy though.

Microsoft's OmniParser V2: Turning Any LLM into a Computer Use Agent Blog: www.microsoft.com/en-us/resear... Hugging Face: huggingface.co/microsoft/Om... GitHub: github.com/microsoft/Om...

Obama correctly called the beef of the millennium.

When solving a hard problem, it's very easy to forget(or egotistically ignore) that a simple solution would be to "find someone who has solved this problem before".

Crypto guy in 2014: Crypto will revolutionize finance and change how we interact with money Crypto in 2025:

Why do LLMs trained on over 90% English text perform so well in non-English languages? They find that they learn to share highly abstract grammatical concept representations, even across unrelated languages!

It would be amusing if High-Flyer's quant fund is incurring significant losses due to their model, DeepSeek R1. It’s unlikely they could have foreseen this, as, like all quant models, it relies on historical data.

"The leading AI models are now very good historians" - by Benjamin Breen, a history professor at UC Santa Cruz resobscura.substack.com/p/the-leadin...

Buy the dip lmao reasoning models only increase demand in Nvidia

I like this thread on threads.

Woo there is so much schadenfreude thirst out there that if you invented a faster cheaper form of electricity people would cheer “this is terrible for Silicon Valley!”

DO NOT assume from past events that it’s a bad move for openai/anthropic/etc. to build out massive datacenters the current thinking of the day is merely a transient point in a long timeline with as much compute as we have, we’ll hit o5-levels in a month, and a new bottleneck will emerge

Tech values are falling on the stock exchange right now due to DeepSeek (Nvidia, -12%). This is very much textbook Sputnik effect.

the most honest VC. ngl i can respect a man that recognizes his own vanity

I can't even access their main app atm.

The way everyone's saying this and then there's me having a bunch of R1 queries go unhandled . At this point I think the Deepseek team singles me out lmfao.

Deepseek has been my daily driver for a while and what most don't know is that it is suspiciously good at writing offsec tools...

I will not be taking any questions at this time