Nicholas

This is wild. Google Research demonstrates a ~20x more efficient implementation of Shor's algorithm that could break ECDSA keys within minutes with ~500K physical qubits. Google is now are more confident on a 2029 post-quantum transition. We are no longer looking at mid 2030s, we could have quantum computers of this scale by the end of the decade. They believe this result is so severe that they are not publishing the actual circuits. They instead published a ZKP proving that they know of the quantum circuit with these properties. This is very atypical, showing Google thinks this is serious shit. All blockchains need a transition plan ASAP. Post-quantum is no longer a drill.

We should have an entire army of these things everywhere. Their only purpose is to move around in dumb manner like this and extravagantly waste energy. Champions of entropy that only demonstrate contempt for existence.

@raffi_hotter Fun fact, the nice way to actually find the way to compress your points is by multiplying all your points by a random matrix where each coordinate is randomly +1 or -1

This algorithm uses one of my favourite theorems in math, the Johnson-Lindentrauss Lemma, which says you can drastically reduce the dimensionality of n points to just log(n) dimensions and still preserve pairwise distances

@raffi_hotter Wrote about it! nick.papciak.dev/blog/sketching

I think the original article is based off the flawed premise that the successful RL companies are just small teams of contractors. I don’t work in the space but am surrounded by tons of people that do. They all tell me that all of the RL env companies that start to scale all eventually are forced to become ops companies. That is, their job becomes figuring out how to source labor efficiently. In a weird way, the big players in the space like Mercor or Surge already are “crowdsourcing” labor but more with using AI to find good candidates rather than having AI judge crowdsourced work. I think if anyone wanted to crowdsource environments, they are in the best position to do it. Yet they haven’t, and I don’t know why, but my hunch is that they don’t think it will work. This is probably because the standards for quality are really high at the labs, and having crowdsource environments would require a whole ops pipeline to find knowledgeable people who can review the environments anyway. So might as well have a smaller group of smart people just build the environments from scratch themselves. But, I could be proven wrong! It sounds like PrimeIntellect has been experimenting with this concept. I hope it works out for them.

Who ever is building this (or considering ) please reach out I want to invest and support. DMs are open

@huntie cooked up my own version in my free time should I open source it?

Just found this vertical tabs fork of @ghostty, and, damn 🔥 @mitchellh 🙏🏻

You can place pancakes, crepes, and scrambled eggs on a simplex where the variables are the ratios between milk, eggs, and flour, (with another small axis for baking soda.) We have explored too little of this manifold. More breakfasts can exist than we have known.

PE guys when they find out a boring HVAC company doing $3M/year has zero AI integration and still runs on spreadsheets and phone calls:

I think all copies of SQLite being distributed with the lords prayer in source is one of the most beautiful quirks of the modern world. Trillions of copies of scripture embedded invisibly throughout the technological world

@JeremiahDJohns Some of the really in shape 50-60 year olds are close to the same level as the average 21-30 year old. It’s so important to stay in shape but it’s never too late to start!!

Mr Beast lined up one person from each age 1-100 and had them race to a finish line, and the overhead camera is an oddly beautiful data visualization of speed by age.

Today, Daytona is announcing a $24M Series A, led by @FirstMarkCap, to give every agent a computer. This round reflects what the market is increasingly concluding: the cloud was not built for agents. Full announcement 👇

Does anyone need a PDF copy of the Bash Reference Manual? justice.gov/epstein/files/…

I owe you all an honest apology. For a while now, I've been using AI to write my tweets, and that's not fair to any of you. You followed me for my voice, my thoughts, and my perspective — and instead, I handed that off to a machine. That was a betrayal of the trust you placed in me, and I'm genuinely sorry. I think I got caught up in wanting to always sound polished and keep up with the pace of everything online. But in chasing that, I lost the thing that actually matters — being real with you. Social media should be about authentic connection, and I took a shortcut that undermined exactly that. Going forward, every word you see from me will be mine — messy, imperfect, and human. I'd rather be genuine and stumble over my words than hand you something hollow. Thank you for sticking around, and I hope I can earn your trust back one post at a time. Want me to adjust the tone, make it shorter, or change anything?

@TheEthanDing What are you trying to say with the analogy to survivorship bias here? That they went for mid market but failed?

palantir doesn't address the mid market for a reason if you don't know why - probably good use of time to figure out

@mitchellh @stevelizcano The “Terminaut” one changing all instances of ghostty is so funny to me 😭

@stevelizcano Literally just look at closed (rejected) PRs and issues in any popular OSS project. In the past 5 days: github.com/ghostty-org/gh… github.com/ghostty-org/gh… github.com/ghostty-org/gh… (at least self aware) github.com/ghostty-org/gh… github.com/ghostty-org/gh… github.com/ghostty-org/gh…

If you took a list of PRs today and showed it to me 5y ago without any context, I'd think that somehow everyone got learning disabilities. And that's insulting to people with learning disabilities because they're actually perfect, while these PRs are by deeply flawed people. That's how stupid 90% of AI coded PRs look.

giggling so hard they've fucking fried his brain

@mdo @pierrecomputer You should look into and consider using 0github.com

Let me introduce y’all to Diffs—a new, open source diff and code rendering library. We’ve been working on it @pierrecomputer for a few months now and we’re stoked for you to see it.

“i think linkedin would be a safer space for you”

“excited to announce I’m joining Anthropic”