alecco

6 ms·

121.
by alecco 5mo ago
> [...] better idea to directly fund the expansion of capacity [...] > > However, that the hyperscalers and AI companies aren't doing this says a lot about their true beliefs about how much future demand AI will have. With wha
122.
by alecco 5mo ago
This is very personal to me. I took that flight several times. Always went through strong turbulence. "At 32 years old, Bonin was the least experienced pilot on the flight crew. When the aircraft's pitot tubes froze and iced over
123.
by alecco 5mo ago
The drama would move to how to make the prompts "fair". Spoiler: "fair" is code word for we are all the same .
124.
by alecco 5mo ago
> For you, it's time to tilt to bond Do you know anything about the state of bond markets?
125.
by alecco 5mo ago
I've seen many co-ops fail. They never address the long-tail nature of productivity, perhaps to not offend people or because most people can't admit it or even understand it. Popular loudmouths take power. Quiet competent people l
126.
by alecco 5mo ago
The model weights , sure he can "steal" them. But do what then? Serve it on the side? Even distillation would be obvious. And I wouldn't be surprised Anthropic is running the smaller inference models, keeping the base large
127.
by alecco 5mo ago
The EU created a € 94 billion fund for exactly this a couple of years ago. And then they put in charge a lawyer who is the daughter of a powerful Spanish media mogul. I tried registering a couple of years ago, but it required a female co-
128.
by alecco 5mo ago
For specific use cases, absolutely, a harness and other techniques help (this is literally what I'm working on). But GP was talking about general use. Over-promising is a very stupid thing. Nobody will value the intermediate steps. Nob
129.
by alecco 5mo ago
US labs got tough on "adversarial" distillation [1]. I suspect that's one of several reasons why Chinese big labs are lagging again. [0] US AI firms team up in bid to counter Chinese 'distillation' (Apr 7) https:&#
130.
by alecco 5mo ago
> Local modals are 6 months to 18 months behind frontier. I wish this was true but it is not. And I am working on open source models so if anything, I would have a bias towards agreeing with you. Frontier closed models (GPT/Claude)
131.
by alecco 5mo ago
This deserves its own blog post and HN submission. Since SSDs have been 2x to 4x prices people are now more likely to buy used and could get burned. BTW thank you for raising this.
132.
by alecco 5mo ago
> never really bought into the efficiency claims Macros, dude. It's a bit annoying to become disciplined in text structure/regularity to have macros work flawlessly, but it's a bliss afterwards. And then... recursive macro
133.
by alecco 5mo ago
Poe's Law
134.
by alecco 5mo ago
This is engagement bait. I almost fell for it.
135.
by alecco 5mo ago
> directly to PTX Weird. There's a recent NVIDIA MLIR that is quite good and fast. Or they could target the even easier and more recent/fashionable tile IR [1] used by CuTile [2] (a little bit higher level but significantly eas
136.
by alecco 5mo ago
Today's Wired is a shadow of its former self. I don't know what happened. But it's sad.
137.
by alecco 5mo ago
You can run nuclear (or gas or whatever) at a minimum to keep the network stable. Read my comments. Solar and wind are very unstable and unless you have the latest generation hardware (costing hundreds of millions) you need some stable sour
138.
by alecco 5mo ago
The government refused to protect nuclear plants when prices went negative for a long time, so of course they were turned off because the companies were losing money. That was the obvious plan all along. And no more blackouts because now th
139.
by alecco 5mo ago
> Damian Cortinas, chair of ENTSO-E’s board, told the Financial Times that “the issue is not about renewables” but about the grid’s ability to manage “fast voltage variations” that can destabilise the system. Unusual oscillations trigger
140.
by alecco 5mo ago
AI (even if pseudo-AI) is already a huge productivity multiplier and becoming better... but due to demand also becoming more expensive. And if things keep going this way, only corporations with deep pockets and the top 1% will be able to af
141.
by alecco 5mo ago
It's even in the movie. Zuckerberg was inspired by Hot or Not and used his friend's chess ranking algo for Facemash (comparing college girls). And that inspired him in part for making the Facebook.
142.
by alecco 5mo ago
Some day I should write about a very similar experience at ~7yo with my slightly older cool cousins showing me things as shocking (or more) out of their (MD) father's huge library (hush-hush). Things like high resolution pictures of ex
143.
by alecco 5mo ago
The viral chart is about Codex 86.1M installs (up from 5.8M) in the week of 27/5-3/5. https://substackcdn.com/image/fetch/$s_!20oI!,f_auto,q_auto:...
144.
It Was a Good Quarter for "Other Income" [A16Z Charts on AI Growth](a16z.news)
2 points by alecco 5mo ago|1 comments
145.
Anthropic weighs deal for near $1T valuation as revenue surges(ft.com)
12 points by alecco 5mo ago|1 comments
146.
by alecco 5mo ago
https://archive.ph/WpFnG
147.
by alecco 5mo ago
> Torvalds said that disclosing the bug itself was enough, without the pursuant circus that followed when a major problem has been discovered. [1] So it's not surprising Dirtyfrag was disclosed by a fix in the Linux kernel. [2] [1]
148.
Copy fail 2: electric boogaloo [with actual timeline Dirtyfrag embargo break](afflicted.sh)
1 points by alecco 5mo ago|0 comments
149.
by alecco 5mo ago
People are blaming the guy who wrote the exploit for breaking the embargo but it was actually broken in Linux by publishing a fix [1]: > on 2026-05-05 Steffen Klassert pushed f4c50a4034 to netdev/net.git with Cc: stable@vger.kernel.
150.
by alecco 5mo ago
People are blaming the wrong guy for breaking the embargo but via this blog post [1]: > on 2026-05-05 Steffen Klassert pushed f4c50a4034 to netdev/net.git with Cc: stable@vger.kernel.org. Once the fix is out it's usual for rese

More ›