markhahn

28 ms·

391.
by markhahn 3y ago
Yet our cultures seem resistant to usage like "1.8 teracells". Is it too nerdy? Would it be more palatable to normies if each prefix had a cardinality suffix ("teraion" is not great...)
392.
by markhahn 3y ago
yeah, but a lot of them are slackers
393.
by markhahn 3y ago
huh? don't ML researchers use pytorch and tensorflow?
394.
by markhahn 3y ago
it's not just nvidia-specific, but nvidia-biased. the misuse of "core", pretending that CPUs are backwards, etc. every developer should know about more than nvidia's spin.
395.
by markhahn 3y ago
are there any documented cases of (intentionally) explosive pigs? might be handy for decommissioning pipelines that have outlived their political purpose, in places that make it difficult to establish who did it.
396.
by markhahn 3y ago
the problem is managing what you "scrape" off. such particles are ideal for causing strokes.
397.
by markhahn 3y ago
little-known?
398.
by markhahn 3y ago
Well duh: you use prioritised queues precisely when there is a capacity limit. In lots of cases, the facility is specifically mandated to achieve as close to full capacity as possible. Which is not to agree with the claim that latency queu
399.
by markhahn 3y ago
"nothing else" really? why would an institution pay them? in reality, faculty do more than publish, and are also judged by service, supervision, etc.
400.
by markhahn 3y ago
You're thinking of social-media-type "reputation". Instead, think of the goal being to associate measures of worth with the reviewers. If you're publicly rated by a bunch of worthwhile people, count yourself lucky.
401.
by markhahn 3y ago
more-open publishing does not necessarily do away with peer review. the really interesting part is that sometimes review benefits from anonymity and even blinding. doing that in a more-public publishing process would be interesting. (but
402.
by markhahn 3y ago
journal "monogamy" is the problem. modest proposal: always publish to arxiv, which assigns the permanent DOI. journals can be offered the article - one at a time if the authors desire, or in more of an auction format. when publi
403.
by markhahn 3y ago
I have to say, those are some ugly solenoids. Is it really that unusual/ocd to want them a little neater?
404.
by markhahn 3y ago
instead of sparc, I wonder if you were remembering the i860, which certainly ran NT/OS2.
405.
by markhahn 3y ago
and your devices probably don't cost $3 on aliexpress, right?
406.
by markhahn 3y ago
it's funny that when "discipline and rigor" is advocated for, the underlying premise is basically "pain builds character".
407.
by markhahn 3y ago
Search engines are factual; why should someone be allowed to override the facts? I'd be very happy if this were about correcting facts: about forcing the takedown of inaccurate content from websites. Google would then update its index
408.
by markhahn 3y ago
huh? wouldn't all the planes with AESA radar dispute this characterization?
409.
by markhahn 3y ago
what would it cost to really begin it's next phase (rebuilding, not what amounts to a mausoleum that runs some classes)?
410.
by markhahn 3y ago
hah - did you intentionally refer to ugliness there? I think one of her most cogent points is that theoreticians are often driven by "beauty", which is not a valid epistemological or metaphysical principle.
411.
by markhahn 3y ago
I guess I was waiting for a punchline related to the "boundaries" part. yes, this kind of error is multiplicative and thus has exponential effect, even though our intuition leans on additive effects. but causality propagates at c
412.
by markhahn 3y ago
unbelievable how pols never seem to get it. the purpose of making new laws is to further protect the public - not damage the public to make LEO jobs easier. sure, it sucks that they can't sit in their office and wiretap anyone, or mig
413.
by markhahn 3y ago
but that's just because manufacturers desperately hide the fact that their own firmaware is based on OSS, and that there are alternate stacks available. imagine instead that vendors had to acknowlege the structure of their firmware, an
414.
by markhahn 3y ago
Extraordinary eagerness for coverage (by a researcher) demands extraordinary caution.
415.
by markhahn 3y ago
Seems odd not to mention competitors such as AMD. Not as if Nvidia has any secret sauce. "Analysts" like this also seem to misunderstand Cuda as that such a sauce, when in fact AI wants higher-level toolkits (which are not at al
416.
by markhahn 3y ago
as a very different use case, I've been wondering about drones for de-mining. being able to hover efficiently would be a major win, and it seems like detecting mines isn't so much of a technical problem as a practical one...
417.
by markhahn 3y ago
well, you hope to stay at a safe distance, right? how difficult is it to deal with wind (or even precipitation)? having tight flight rules (eg to deal with wind) would be a problem in a lot of locations...
418.
by markhahn 3y ago
the basic premise is wrong: it's not a tradeoff, which would require someone to say "doing this correct and expected thing blows up because another correct and expected thing can go faster". the problem is mixing in incorrect
419.
by markhahn 3y ago
Really, the issue here is that security is never treated quantitatively. At least in my experience - are there examples of quantitative security? In some sense, the problem is that the goal is zero controls breaking, but of course, that als
420.
by markhahn 3y ago
jeez this is evergreen. I wonder if it's because it shows instances of profound stupidity among people we tend to credit as very smart, or if it's just fascination with the exotic and dangerous world of criticality. unfortunate th

More ›