currymj

92 ms·

931.
by currymj 9y ago
part of the reason this may be salient now is that there are SO MANY undergrads taking CS courses. classes that might have had 30 students in the past now have 300. you can’t hope to grade projects for all those students, even if you put th
932.
by currymj 9y ago
the reason the pay is so low is that it’s part of a PhD, usually, so the TAs tolerate it. i don’t think you could hire someone to just be a TA full time at those wages. you can’t admit more PhD students, necessarily, because to graduate the
933.
by currymj 9y ago
this wasn't a law, though, it was an executive order, which is just a decision by the president on how to run the executive branch. in this case, it only applied to companies that had contracts with the federal government for more than
934.
by currymj 9y ago
definitely not recently but I think MATLAB, Mathematica, S, Stata, and lots of other commercial scientific/numerical packages could qualify. it may not be possible any more, though.
935.
by currymj 9y ago
other than like, the children of international royalty and whatnot, you just can't get into an Ivy League school without working really hard one way or another. rich kids might go to less-prestigious private schools, but they're n
936.
by currymj 9y ago
this isn’t sustainable though, as people with smaller Erdös numbers will die out and larger ones will be too common. so we should probably allow the smaller Erdös numbers to be inherited through primogeniture, to make sure we still have an
937.
by currymj 9y ago
Lou Pai famously sold his Enron stock in a divorce settlement, after his wife left him because of an affair with a stripper. He got away with $280 million and no insider trading charges. He and the stripper are now married and spend their t
938.
by currymj 9y ago
no, finally, for once, this is actually a real application of the blockchain! specifically any that has support for zero-knowledge proofs so that transactions are actually anonymous. you could have a bunch of people buy lottery tickets by s
939.
by currymj 9y ago
HFT has a lot of impacts and may be good or bad overall, or socially useless, or whatever. but it undeniably does result in a better deal for a normal person trying to make small investments -- they've replaced the old market makers bu
940.
by currymj 9y ago
right, which is bad because it results in an extremely dull game. it’s not inherent to the rules of baseball, though. the easier it is to hit for power the more this is true. imagine if we averaged 0.5 home runs per game instead of the curr
941.
by currymj 9y ago
in some other languages it's basically called that. "Informatik" in German. there's still the vestigial title in some department names in the US, "Department of Computer Science and Informatics". I agree it wou
942.
by currymj 9y ago
i realize the entire speech is a troll, but it successfully made me very angry. i'm not a physicist, and have no interest in physics, yet I do need mathematics. so it's always been irritating to me how much introductory or "a
943.
by currymj 9y ago
the type-checking catches lots of memory errors (irrelevant in a GC language like Python, but wouldn't be caught by the compiler in C or C++), but it also catches data races when writing concurrent code, which is definitely something t
944.
by currymj 9y ago
in 2015 i thought cryptocurrencies would probably appreciate so i bought a reasonable amount that was within my ordinary-consumer budget and my appetite for risk. i did okay. if i wanted to do the same thing now but short, i probably could,
945.
by currymj 9y ago
I would like to do this but I don't have several million dollars lying around to get in on the futures market. And I'm sure as hell not going to keep a margin account at any cryptocurrency exchange.
946.
by currymj 9y ago
i don't think so. the author here also helped co-write the "deep learning is alchemy" talk that was somewhat controversial at NIPS. i think this is especially important if you purely want to do applications. we have a bag of
947.
by currymj 9y ago
i don't know about "useless". scientific computing, for instance -- you may have a weird set of dependencies for some code that you want to deploy a copy of on 200 slightly heterogeneous nodes, once, and hopefully never again
948.
by currymj 9y ago
agree with the other posters, could you guys please publish something on what you're doing?
949.
by currymj 9y ago
this mainly has to do with the incentives of journalists in an attention-based industry, I think.
950.
by currymj 9y ago
at the same time, it's possible for the website to do absolutely everything right and still lose a lot of information. one reasonable solution is just to say tough luck, it's still your fault, and then every company dealing with s
951.
by currymj 9y ago
you're lucky to have a life where you get to do that. but do you have subordinates? people with a lot of autonomy are harmed less by long hours and stress. i mean quite literally, if everybody is working 70-100 hours under intense pres
952.
by currymj 9y ago
people often mention forcing refactoring/rewrites as a benefit of unit testing, that you end up with code that is actually better in testing-unrelated ways. this is true in my experience. i wonder how true this is of formal verificatio
953.
by currymj 9y ago
i thought the best part of the article was towards the end, where she talks about all the different ways that gymnasts have of redefining success. i think it may be healthy to avoid doing that. like, don't say "well, my company fa
954.
by currymj 9y ago
i'm optimistic AMD GPUs will be usable for general purpose computing soon. ROCm seems to be coming along nicely, and they have various deep learning frameworks functioning to various degrees. in particular, I think there's buy-in
955.
by currymj 9y ago
If you are interested in speech recognition, Mozilla just released a big open corpus of speech data and some trained models. Visual Basic is a fine programming language with very weird syntax. Hell, Microsoft has put out .NET bindings for C
956.
by currymj 9y ago
Not sure it was confusion, but rather intended as an example of how "high end" features spread to the bottom of the market quickly, such that in a few years nearly all new cars may be internet-connected.
957.
by currymj 9y ago
the idea that these are processes and you can sample your trajectories doesn’t get enough emphasis, at least in presentations related to machine learning. i mean it’s of course acknowledged but i wish i had seen more plots like these when t
958.
by currymj 9y ago
only extremely basic stuff, nearly done with Deutsch-Josza now. But again the alternatives are basically just representing your qubits as huge vectors and gates as huge matrices and just doing lots of multiplication, which is an extremely c
959.
by currymj 9y ago
Microsoft's Q# is interesting. It's framed in a way that seems weirdly premature, but it is probably the nicest platform to learn about and play with quantum algorithms. Better than multiplying huge matrices in Octave, for sure.
960.
by currymj 9y ago
it is a really interesting language, in the sense of having many unique and unusual features.

More ›