Y
HN Search
Hacker News Search
new
|
comments
|
top
|
jobs
advael
searching PlanetScale…
1.
▲
2.
▲
3.
▲
4.
▲
5.
▲
6.
▲
23 ms
·
361.
▲
by
advael
2y ago
That's silly. The police were doing a considerable amount of violent suppression of protests in Portland. People even started calling the mayor "tear gas Ted" The protestors themselves mostly did property crimes though. The &
362.
▲
by
advael
2y ago
I agree with you, and kind of even agree that crowdstrike may not directly be at fault. But my point is that this competency is bled out of hospitals by external forces, primarily two: distant administration from companies that buy and mana
363.
▲
by
advael
2y ago
Who in the org do you expect to have that competency, and do you think hospitals aren't keeping crucial things like credentials or software that gates access to things in the cloud when literally everyone in the world is encouraged to
364.
▲
by
advael
2y ago
I think a good remedy would be to completely remove "normal procedure" as a defense against liability. Our legal standard should defend people who break protocols if they know they will result in harm, and prosecute people who don
365.
▲
by
advael
2y ago
The unfortunate fact of the matter is that centralizing IT systems around large corporate products, including the on-prem software and any cloud services, necessarily means less local control of what can go wrong and how it can be mitigated
366.
▲
by
advael
2y ago
I basically agree with most of what you've said here, but I'm not talking about humans in general, I'm talking about epistemic authorities. As I said, I think the intuition that leads people to the conclusion that animals thi
367.
▲
by
advael
2y ago
Between this and Boeing it's chilling how dangerous it can be to whistleblow against these kinds of cartels I think this site kind of undermines itself by saying stuff like "Windows kills more people than COVID (albeit over a long
368.
▲
by
advael
2y ago
That wasn't always true, and changes in that direction were made to a lot of government agencies, doing things like making them pretend their budget is a business and that they need revenue streams is nonsensical and doesn't work,
369.
▲
by
advael
2y ago
It really isn't. I am making a claim about a pervasive bias in scientific institutions I view to be course-correcting. I'll acknowledge plenty of scientific results I dislike as being true. Like I really dislike that every room-te
370.
▲
by
advael
2y ago
What's even sadder is that this is said in an economic and regulatory environment that has gradually winnowed away all the examples of businesses that made the argument even the slightest bit compelling if you squinted
371.
▲
by
advael
2y ago
Sure, but the answer they gave to this reporter was the same usual corporate garbage response that included "we need analytics to market our products" (???) I think it's fucked up that any agency is "marketing products&q
372.
▲
by
advael
2y ago
Thanks for the reading recommendations, that stuff sounds fascinating, and I'll admit that my reference to RCTs was perhaps an overly mean dig at overcorrection for methodological rigor, a tic I likely developed from my exposure to the
373.
▲
by
advael
2y ago
Dam, the angry backlash mindset shows up like clockwork, complete with my favorite triad of its hallmarks - vague references to "philosophers", invocation of numbers with the specific intent of saying buckwild unquantifiable shit,
374.
▲
by
advael
2y ago
The pages load responsively, there are no apparent bugs in the code, the organization is coherent and sensical, and the writing is concise and conveys both information and the unique voice and personality of the author This is a better-made
375.
▲
by
advael
2y ago
Love to hear stuff like this, both because it's interesting in its own right, and because the fact that it gets published and taken seriously gives me hope that we're finally getting our heads out of our collective asses with rega
376.
▲
by
advael
2y ago
Nothing if you do it right. Indexing also goes wrong if you do it wrong. In both cases, they make sense as abstractions, but are less safe than certain other abstractions I see lots of good code in the linux kernel that uses explicit indice
377.
▲
by
advael
2y ago
Yea, my bad I guess. I tend to think people mostly get that biometrics are, well, mostly immutable and that not being able to switch them up in response to a suspected breach is a huge inherent weakness. So the only defense I really get of
378.
▲
by
advael
2y ago
Sure, it's doable, but it was in this case the source of the issue, along with some obfuscation of where and when those particular sections of code were being executed through shenanigans with encapsulation and inheritance (I respect p
379.
▲
by
advael
2y ago
This comment's tone is presumptuous, judgmental, and reactionary. Based on the small sample of textual prose in this completely unrelated context, I have to assume that the author's whole interpersonal vibe and decision-making pro
380.
▲
by
advael
2y ago
I mean yea, if we assign the most status to bullshitters, of course they're gonna think dumb shit like "the infinite bullshit generator is already superhuman intelligence"
381.
▲
by
advael
2y ago
Nah. Many people who have been writing C++ poorly for decades never learn to do it better. This is less about "even people who are great are bad" and more about "no amount of experience guarantees that you get great". Th
382.
▲
by
advael
2y ago
I think the most common way to try to insult someone right now is to call them a "con artist", and this can sometimes get very silly, but the core kernel underlying that reality is that we've built a world by and for people w
383.
▲
by
advael
2y ago
Okay, but what does that mean? AI is a search process. Do you mean you want the AI to formulate queries? Test hypotheses? Okay. How? What does that mean? What we know how to do is to mathematically define objective functions and tune agains
384.
▲
by
advael
2y ago
Also, one of the major classes of problem that gets solved and we view as "progress" in machine learning is framing problems. Like we couldn't define "draw a good picture" in a way we could actually assess well, GAN
385.
▲
by
advael
2y ago
I agree with this tip. Works great for anyone who can autodidact, and if you're good at finding and vetting resources, autodidacting got easier with the internet, and has only gotten a little harder with the proliferation of nonsense o
386.
▲
by
advael
2y ago
Even a password can be changed if there's a compromise. Biometrics are bad because they can be imitated, but not changed. A breach is permanent
387.
▲
by
advael
2y ago
What in the world Most "diffusion models" use similar VAE to transformer backbone architectures. Diffusion isn't an architecture, it's a problem framing As for the rest of this, I'm torn between liking the poetry of
388.
▲
by
advael
2y ago
Not pretending a GUID constitutes a security measure in the first place? It's just not the right tool for the job in any sense
389.
▲
by
advael
2y ago
I mean it seems simpler to just buy a cybertruck if that's what you want
390.
▲
by
advael
2y ago
I mean SSNs are the worst possible authentication mechanism and yet we still have to freak out every time they're leaked. Security practices are so utterly backwards everywhere that it's quite apparent no one powerful is incentivi
More ›