juancn

5 ms·

31.
by juancn 2mo ago
I don't get the legal contradiction. The search is supposed to be lawful without a warrant because you're not really in the US yet per-se, hence if you're not there, how deleting the data can be a felony?
32.
by juancn 2mo ago
Just set the following incantation: You must use ASD-STE100 Simplified Technical English (STE) when it doesn't detract from meaning.
33.
by juancn 2mo ago
Didn't OpenAI bought it?
34.
by juancn 2mo ago
Won't this end up with news orgs that don't have deals just being delisted? It's probably great for the 3 or 4 largest ones, but the rest?
35.
by juancn 2mo ago
It's freaking fun! That's basically it. What's common and everyday in one place is exotic in another. It lets you get something that's just plain fun, for little money and it feels like a shared experience with the local
36.
by juancn 2mo ago
As usual, it depends on the scale, but it's a sane default for 99% of use cases. Different use cases have different scalability limits in PG, when you get to them you need to deal with them. It would be perfect if it had somewhat trans
37.
by juancn 2mo ago
There are some nooks I still enjoy. Hacker News is one, I like Reddit, there's still some fun discussion going on. What I dislike is all the posturing and all the low effort crap.
38.
by juancn 2mo ago
One HUGE issue is that datacenters don't have an afterlife in the communities where they're located. They add to the Rust Belt once their lifetimes end, contrast this with old brick warehouses which were eventually repurposed as h
39.
by juancn 2mo ago
This is not new at all: https://www.cookingforengineers.com/ Look at the site design! It went viral on Digg! That's how old this is.
40.
by juancn 2mo ago
I would love to have some benchmarks for this on some reasonably practical scenario. In many common cases shift and masking can replace integer division (i.e. hash tables) and you avoid division altogether, and that probably has about the c
41.
by juancn 2mo ago
And if the problem is really really hard, you can throw an AI at it and hopefully get a probabilistic solution. (not necessarily an LLM, AI is a huge field)
42.
by juancn 2mo ago
Nobody owns Claude's output (nor that of any AI), they're in the public domain, they cannot be copyrighted. The author must be human, at least for now, in the US this precedent holds as far as I can tell: https://law.ju
43.
by juancn 2mo ago
Makes sense, there's some precedent that AI generated code is not copyrightable, it's essentially in the public domain for now. Oracle likes its licenses.
44.
by juancn 2mo ago
How does it smell though?
45.
by juancn 2mo ago
My guess is that except for simple one-offs or low risk situations, any recurrent use of AI will require some sort of expert to handle it, at least for a while. Liability plays a big role here, in the end a person needs to be made responsib
46.
by juancn 2mo ago
I still don buy it. The from-the-hip cameras typically have a waist-level viewfinder, which still is a good approximation of the optical path or you don't and you shoot completely blind. In that case any camera does the trick and so do
47.
by juancn 2mo ago
I think it's the cost (both tokens and initial setup and debugging the workflow). It's really easy to spend a gazillion tokens by mistake and the resulting workflows are still probabilistic and open for abuse or failure. Maybe it&
48.
by juancn 2mo ago
Optical viewfinder cameras see through a lens and force-align your eyes, so when you change the lens, what you see is different but highly aligned to what the sensor sees. Here, depending on where you place the thing you see a different thi
49.
by juancn 2mo ago
You lost me at the annoying trail on mouse movement. You advocate for good UI and immediately make some crap that makes it hard to read/use your site. It's kind of ironic. Please stop messing with: selection, mouse cursor, copy&#x
50.
by juancn 2mo ago
I really liked the ket notation. I was aprehensive at first, but it makes operations much more clear. I would have liked some refresher on some variables though (like d_k in quadratic attention).
51.
by juancn 2mo ago
They probably already scraped it.
52.
by juancn 2mo ago
I love when things work out in the end! (issue is fixed in the latest MacOS 26 and 27 betas)
53.
by juancn 2mo ago
To make ray tracing practical, you need to update the ray-tracing accelerating structures when you animate.
54.
by juancn 2mo ago
AI companies are a lot more like traditional manufacturing companies. They can only subsidize you so far, because they may not be even be covering their variable cost at this point. There's no software multiplier (build once — pay the
55.
by juancn 2mo ago
I dunno. It's sad in a romantic kinda way, because of the lost artifact, but the information is what makes the book valuable, not really the medium. The out of copyright books don't really need to be destroyed anyway for them to b
56.
by juancn 3mo ago
This is one of my favorite quotes from Brooks: "In practice, actual (as opposed to formal) authority is acquired from the very momentum of accomplishment." Was this AI paraphrased? From the article: "In practic
57.
by juancn 3mo ago
And that's a HUGE issue. SSH keys are easy to manage in comparison to PassKeys.
58.
by juancn 3mo ago
So? We have information that Fable was distilled from humans. If it works it works. Isn't that the argument? AI outputs are not copyrightable, so distillation is fair use. It may be a TOS violation, but that's a private matter. Ca
59.
by juancn 3mo ago
How? Where are they stored? Which password protects them? How do I back them up? How do you move them to another device? Can I print them out on paper as a last resort measure? Nobody fucking knows.
60.
by juancn 3mo ago
A password has the best recovery process. Period. Write it down somewhere. It's up to me, no third party. I had more trouble with a failed 2FA than with a password.

More ›