p-e-w

12 ms·

301.
by p-e-w 8mo ago
About Dry Grasses by Nuri Ceylan. Probably the best film I’ve seen in the past 10 years, which isn’t saying that much because the past 10 years have been among the worst in the history of film, but it’s still a very good movie.
302.
by p-e-w 8mo ago
> I can write you an unoptimised C compiler that emits assembly for $20k You may be willing to sell your work at that price, but that’s not the market rate, to put it very mildly. Even 10 times that would be seriously lowballing in the r
303.
by p-e-w 8mo ago
There is no definition of “predatory business model” that isn’t simply a reflection of the majority’s values, so there absolutely is a conflict between the two. Are churches a predatory business? If the answer is no, then why are sugar manu
304.
by p-e-w 9mo ago
> When someone is a certain amount smarter than you, distinguishing their plausible bullshit from their deep insights is really, really hard. Insights are “deep” not on their own merit, but because they reveal something profound about re
305.
by p-e-w 9mo ago
With fundamental rules, applicable to all situations, limiting what information courts can demand. There are things so private that they should be out of reach of the state regardless of what justification someone can come up with.
306.
by p-e-w 9mo ago
The only reliable way to protect rights is to limit power, and the only reliable way to protect fundamental rights is to limit power with absolute prohibitions. This was well understood in the decades following WW2, and many countries imple
307.
by p-e-w 9mo ago
What I don’t believe is that courts should have the power to force anything to happen just by signing a piece of paper.
308.
by p-e-w 9mo ago
Why? What is the rationale? Unless of course you subscribe to the idea that anything goes as long as a court decrees it, in which case there’s nothing to debate really.
309.
by p-e-w 9mo ago
> I feel like this wording isn't great when there are many impactful open source programmers who have explicitly stated that they don't want their code used to train these models That’s been the fate of many creators since the
310.
by p-e-w 9mo ago
They probably have a trillion emails with human labels, either from users directly applying them, or inferrable from actions like deleting. With that much data, even a simple Bayesian classifier should work pretty much perfectly.
311.
by p-e-w 9mo ago
Elections in most countries involve tens of thousands of volunteers for running ballot stations and counting votes. That is a feature, not a problem to be solved. It means that there are tens of thousands of eyes that can spot things going
312.
by p-e-w 9mo ago
The fact that all SETI endeavors haven’t really found anything is actually a very valuable result, because it constrains “they’re everywhere, we just haven’t been looking” arguments quite a bit. Even humanity’s (weak) radio emissions would
313.
by p-e-w 9mo ago
Being able to count using fingers is more valuable than having one more prime factor.
314.
by p-e-w 9mo ago
That’s where Bayesian reasoning comes into play, where there are prior assumptions (e.g., that engineered reality is strongly biased towards simple patterns) which make one of these hypotheses much more likely than the other.
315.
by p-e-w 9mo ago
> The Turkey was an LLM. It predicted the future based entirely on the distribution of the past. It had no "understanding" of the purpose of the farmer. But we already know that LLMs can do much better than that. See the famous
316.
by p-e-w 9mo ago
Every time this topic comes up people compare the LLM to a search engine of some kind. But as far as we know, the proof it wrote is original. Tao himself noted that it’s very different from the other proof (which was only found now). That’s
317.
by p-e-w 9mo ago
Projects like Wikipedia never have meaningful competition, because the social dynamics invariably converge to a single platform eating everything else.
318.
by p-e-w 9mo ago
Like this: https://github.com/nbp/holyjit
319.
by p-e-w 9mo ago
I know that that’s the plan, but I believe it when I see it. Mozilla invented entire language features for Rust based on Servo’s needs. It’s doubtful whether a language like Swift, which is used mostly for high-level UI code, has what it ta
320.
by p-e-w 9mo ago
Browsers have had sandboxing for well over a decade, and the 3-4 catastrophic vulnerabilities per year happen in spite of that. And most of them are in the browser code itself, not in dependencies. By far the biggest offender tends to be
321.
by p-e-w 9mo ago
It’s indeed rapidly progressing feature-wise, but I have yet to see an explanation for how they intend to manage security once market adoption happens. Ladybird is written in C++, which is memory-unsafe by default (unlike Rust, which is mem
322.
by p-e-w 9mo ago
I’m guessing the uncertainty about whether Servo will be meaningfully maintained 5 years from now is the main problem.
323.
by p-e-w 9mo ago
I remember a time when entire discussion threads were swiftly culled from HN based on the magnitude of their political content. These days, it’s pretty clear that the direction matters a lot more than the magnitude, and “flamebait” is only
324.
by p-e-w 9mo ago
I heard Däniken speak about 30 years ago, and exchanged a few words with him afterwards. He was a brilliant orator and came across as highly sophisticated. His arguments were contrived and I recognized that even as a child, but he was noth
325.
by p-e-w 9mo ago
Matrix published a proof of concept a few years ago of an alternative transport layer designed to work over connections with a few hundred bits per second, and massive packet loss, while still providing E2EE: https://matrix.org&#
326.
by p-e-w 9mo ago
I agree that garbage collection is fine and Go indeed has an amazing garbage collector. Unfortunately, it also has the worst type system of all mainstream languages created in the 21st century, so the benefits are rarely worth the drawbacks
327.
by p-e-w 9mo ago
Because between the 1970s and 1990s, Western nations decided that private operations should be the default for everything except where the law specifically requires state institutions, instead of the other way round. In many countries, esse
328.
by p-e-w 9mo ago
Wow, now that you pointed this out, that interpretation actually makes far more sense than the “correct” one.
329.
by p-e-w 9mo ago
> a house has a depreciation in the order of decades, whereas AI GPUs are a few years. That’s no longer true for datacenter-class GPUs. The A100 came out six years ago , and it still sells for $15k+, with no meaningful drop in the fores
330.
by p-e-w 9mo ago
What other uses are there? I only know the phrase as a synonym for “assuming the conclusion”, i.e., a type of circular reasoning.

More ›