Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
paulgb
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
43 ms
·
691.
▲
by
paulgb
4y ago
Python has become a hugely critical language for science, AI/ML, finance, etc. and the current state of tooling has lagged the language's importance. I can think of lots of ways a company that solves that problem could add adjacen
692.
▲
by
paulgb
4y ago
> What is the point of making every half decent Developer tool a whole startup? I mean, paying people who work on it, for one?
693.
▲
by
paulgb
4y ago
I wouldn’t consider this to be “misusing” a graphics API in the same sense that tensor operations on top of WebGL are. WebGPU intentionally provides compute shaders with the goal of enabling non-graphical pipelines. > [WebGPU use cases i
694.
▲
by
paulgb
4y ago
Buttondown is great! It seems that being indie allows you to just ship a good product that does what it says on the box, instead of trying to squeeze every last bit of juice out by building a bloated product that is a jack of all trades but
695.
▲
by
paulgb
4y ago
Eh, there’s a reason we have a system where the accused has the right to make their case in court. An arrest is a strong signal but by design has a lower burden of proof than it takes to punish someone within the system. Given that being pu
696.
▲
by
paulgb
4y ago
The typical use case is getting around prompt size limits for information retrieval tasks over large corpora. If you want to ask a question of a large corpus, you first use nearest-neighbor search in embedding space to retrieve snippets lik
697.
▲
by
paulgb
4y ago
Expecting this trend to continue forever means expecting the value of Bitcoin to double every four years. It’s basically the wheat and chessboard problem: possible the first few times, then quickly impossible https://en.m.wikiped
698.
▲
by
paulgb
4y ago
> Bitcoin has been mined for over a decade and never came even close. The problem with using past performance of the Bitcoin network as an indicator of future performance is that the exponential nature of halvings means that each decade
699.
▲
by
paulgb
4y ago
> If mining hardware becomes inefficient why would using it to "attack the network" be profitable? Can you say more about what attack you're thinking of? The idea is that under normal day-to-day circumstances, a big deterr
700.
▲
by
paulgb
4y ago
Something I've always considered a curious decision about Bitcoin is the decision for halvings to occur as discrete events rather than as an exponential decay occurring on each block. Having them be discrete events is probably easier f
701.
▲
by
paulgb
4y ago
You're right that the article doesn't address that counter-argument, but the suggested means of profiting from the attack is immune from it: > A successful 51% attack, indeed even a credible threat of such an attack succeeding,
702.
▲
by
paulgb
4y ago
I'd add Tailscale and Fly.io to the category, too. Well-written technical blogs are truly the way to HN's heart.
703.
▲
Why MD5('240610708') is equal to MD5('QNKCDZO')?
(stackoverflow.com)
81 points
by
paulgb
4y ago
|
80 comments
704.
▲
by
paulgb
4y ago
I think two things are happening. One is the “things aint built like they used to be” bias — buildings from the past that are physically well built are the ones that don’t get torn down, so we associate the past with the sturdy stone buildi
705.
▲
Show HN: Botsh gives GPT a containerized scratch space to apt-install any tool
(github.com)
1 points
by
paulgb
4y ago
|
0 comments
706.
▲
by
paulgb
4y ago
Nomic are also the ones behind gpt4all. Such a powerhouse team. https://github.com/nomic-ai/gpt4all
707.
▲
by
paulgb
4y ago
He means “mistake” as in “unforced error which will backfire on them”, not “accident”.
708.
▲
by
paulgb
4y ago
> All stories are about the same two junior recruiters who were hired just before hiring slowdown/pause. And also, based on posts on tik tok where people regularly make up or exaggerate stories for social clout.
709.
▲
by
paulgb
4y ago
Yes, that still holds. Nobody is forcing Twitter to be a free speech platform. It was Elon's claim that it would become one under him, and he garnered a lot of support on the basis of that idea, only to repeatedly make fools of people
710.
▲
by
paulgb
4y ago
You're right, thanks. I've edited my comment.
711.
▲
by
paulgb
4y ago
I'm so tired of tech's lumping of all journalism into one bucket. It comes off as knee-jerk cynicism to absolve ourselves of the work of having to develop the media literacy to tell good journalism from bad. In the case of the pos
712.
▲
by
paulgb
4y ago
It appears that you can't even "like" a post containing the word "substack", which has the effect of suppressing discussion of the fact that substack is shadow-banned on twitter. Edit: tedunangst correctly points ou
713.
▲
by
paulgb
4y ago
Because it’s a way to buy algorithmic preference, the blue check has already subconsciously taken on the same association in my brain that “promoted” at the bottom of a tweet has. If more notable accounts drop the check, and more grifters b
714.
▲
by
paulgb
4y ago
> It's a really funny hill for them to die on and it will look horrible when they cave. “They” being twitter, or NYT? I don’t understand the game Elon is playing with them; why would they even want to be verified now that it doesn’t
715.
▲
by
paulgb
4y ago
> The whole screwup stems from the bizarre misconception that the blue checkmark was some mark of prestige, rather than just "this person has a public-facing job and we've verified their identity". The misconception was co
716.
▲
by
paulgb
4y ago
I've always loved the simplicity and flexibility of CGI. To check my understanding: since CGI just takes a raw request over stdin and returns a response over stdout, would a WCGI wasm module be compatible with WAGI[1] and vice-versa? [
717.
▲
Chroma raises $18M seed round
(trychroma.com)
8 points
by
paulgb
4y ago
|
0 comments
718.
▲
by
paulgb
4y ago
It will be interesting to see the performance differential. Tensorflow.js provides a benchmark tool[1]. When I ran them[2] on an M1 MacBook Pro, WebGPU (in Chrome Canary) was usually 2x as fast as WebGL on large models, sometimes 3x. [1] h
719.
▲
by
paulgb
4y ago
This would essentially mean that every computation would have to run as slow as the slowest supported hardware. It would completely undermine the entire point of supporting hardware acceleration. I’m sympathetic to the privacy concerns but
720.
▲
by
paulgb
4y ago
The dates don’t check out; waxy confirmed it goes back to at least 2019 but that story is from 2021.
More ›