currymj

31 ms·

391.
by currymj 6y ago
the Doomsday Argument seems plausible, but assuming you are selected uniformly at random from all possible humans results in a hidden assumption that the future can have an effect on the past, which doesn't seem reasonable. I realize t
392.
by currymj 6y ago
oh cool, good to know!
393.
by currymj 6y ago
Back in the roaring '20s Charles Ponzi came up with a way to theoretically make a profit trading postage stamps, and raised a lot of money from investors, although in fact he never actually bought the stamps, and instead invented the P
394.
by currymj 6y ago
as people are saying, the typical software engineering advice simply wouldn't work in a research context. one exception is the most basic stuff - people should use version control, do light unit testing, and explicitly track dependenci
395.
by currymj 6y ago
this is very cool. another interesting open-source project for imperfect-information game solving is DeepMind's OpenSpiel: https://github.com/deepmind/open_spiel However as far as poker goes, I believe they only p
396.
by currymj 6y ago
right, but why is that the case? I honestly don’t know but it seems like some kind of different attitude towards technical work could be one explanation.
397.
by currymj 6y ago
you always hear this about the UK. It seems like generally speaking, science and engineering lacks prestige over there compared to being a lawyer or banker? Possibly due to the “two cultures” problem combined with the English class system.
398.
by currymj 6y ago
JFK certainly doesn't have a subway stop at the airport. You have to get off the subway, get a separate AirTrain ticket, and ride the AirTrain to the airport.
399.
by currymj 6y ago
there are lots of applications that benefit from GPU acceleration, where even a single small-ish GPU is useful. Even outside of deep learning -- DL tensor libraries can be useful just for GPU numerical operations + autograd. Managing moving
400.
by currymj 6y ago
if you go to the "documentary" section on Netflix or another streaming service, there are dozens and dozens of really low-quality documentaries. Some are on conspiracy topics (UFOs, ancient aliens) or stupid and sensationalist. Th
401.
by currymj 6y ago
are there pictures of the finished product from the outside? how long until I can buy this?
402.
by currymj 6y ago
doubt it's going to happen this time. the current crop of AI research is producing a lot of real results in all kinds of fields. voice transcription, recommender systems, translation, object detection -- all cases where neural networks
403.
by currymj 6y ago
a fun thing that pedantic nerds can appreciate about football is the obscure, old-fashioned strategies that remain technically legal. in the early 20th century it was common for a player to take the ball and drop-kick it themselves to score
404.
by currymj 6y ago
i don't know if it's the type system... Julia has a very rich type system. It's not exactly popular (yet?) but it's certainly very usable and good for numerical work and ML. but they started out with the goal of building
405.
by currymj 6y ago
I guess the thing with these movies is the high level narrative is interesting, and at the low level, all the individual historical events discussed are interesting. But the stuff in the middle just isn't there. It strongly feels like
406.
by currymj 6y ago
this is a little too mean, but still very funny and gets at the point quite clearly.
407.
by currymj 6y ago
I’ve watched a couple of Adam Curtis’s films. He is a very skilled artist and they are entertaining, but I find them frustrating. They are smarter than conspiracy theories but they still follow a conspiracy logic. Finding patterns where the
408.
by currymj 6y ago
is there a high quality introduction to Python type checking for people who already know Python? maybe something very opinionated about how you're supposed to use the features, sort of a "Hitchhiker's Guide to Type Checking&q
409.
by currymj 6y ago
the monk in question used a lot of pseudonyms.
410.
by currymj 6y ago
this book describes the linear programming formulation for finding the optimal value function. cool to see, I think this formulation is underrated by the CS community compared to other communities also trying to solve MDPs. however I wish i
411.
by currymj 6y ago
it's hard for me to learn something if i don't have something I need to do with the knowledge. it's not that I can't sit down and do it, but things won't stick in my brain. so I would say, try to find something you
412.
by currymj 6y ago
i see what you’re saying in the broader case — i can imagine it would be worth listening to someone who had reckoned with the evil they did — but that’s not how people sold DFW. The pitch was “here’s a book of good advice from a man who was
413.
by currymj 6y ago
truly talented writer, not really the kind of guy who should be held up as a moral teacher. had a long-term pattern of sleeping with his students. committed serious domestic violence against one of his partners. this wasn't well-known
414.
by currymj 6y ago
i tend to think that the approach here described as "incompetent" isn't necessarily bad as long as you maintain honesty about it. for instance, it might be more effective to just donate money but there's nothing wrong wi
415.
by currymj 6y ago
I suspect for both peer review and CVEs, and probably some similar situations I'm not thinking of, it's not just a misconception, it's often more like wishful thinking. People really want there to be a way of telling what&#x2
416.
by currymj 6y ago
this sounds similar to problems with peer review in academia. it mostly works fine as a guardrail to enforce scholarly norms. however many institutions want to outsource responsibility for their own high-stakes decisions to the peer review
417.
by currymj 6y ago
TD-Gammon from 1992 was an early success of neural network based reinforcement learning. Descendants of the technique remain standard today in RL.
418.
by currymj 6y ago
a big conceptual point in RL is the focus on the Bellman equation. value of a state equals immediate reward plus discounted future value. if you know the value of every state, just always move to pick the highest value. well known methods l
419.
by currymj 6y ago
this one is interesting because they basically invented the “pay what you want” box for selling digital media. tons of people copy this now, it’s not even notable. but at the time, the idea that they would let you type in $0 was shocking.
420.
by currymj 6y ago
I believe the main reason for including ODE solvers is basically to do Bayesian parameter estimation of ODEs from data. likewise as far as I know, linear regression and mixture models are both done in a Bayesian style (a hierarchical model

More ›