81–90 of 94 posts
Is not using something for a single day a sign you never need it? I sometimes don't refer to SO for weeks, simply because I'm dealing with problems I can figure out without it. But sometimes I'm dealing with something I'm totally unfamiliar with, and docs and tutorials are badly written, and then some example on Stackoverflow can be a great help. And more importantly, sometimes I'm stuck on something really obscure.…
The amount I need to rely on the internet waxes and wanes as I work on different components of our software. Sometimes I have 20 tabs open, sometimes I know exactly how to get where I'm going and just need to spend time chopping the wood, other times I'm comfortable enough to check the source, sometimes I ask a coworker or maybe I'm frustratedly digging through the source because the only guy that knows how it works quit a year ago.
They're all different areas of skill and domain knowledge, when someone starts out they unsurprisingly have the most prior experience with googling.
Earlier quoted context omitted.
Nope. Thats really the point - if we find the best deterministic algorithm, it means we have found the one with the best deterministic running time. By loosening our criteria of running time to expected running time and allowing randomness, we can sometimes do better than deterministic algorithms.
It doesn't square well with my intuition - if we understand how allowing randomness improves expected running time, then surely we must be able to create a deterministic algorithm that is at least as good as the randomized one. Am I missing some critical piece of understanding here?
Iow, if you can never do X, then you cannot write any algorithm that may do X.
Earlier quoted context omitted.
Easy solution: 1. Look up the first 60 questions on StackOverflow. That amounts to 5 hours you would've spent on just the first one. 2. If you still have more questions, spend another 5 hours learning the ins and outs of the darn thing. Problem solved: now you're not sacrificing latency for throughput.
This is exactly the rent-or-buy problem! http://en.wikipedia.org/wiki/Ski_rental_problem You just described the best deterministic algorithm, but it turns out there are even better randomized algorithms, where you "flip a coin" when deciding to look up StackOverflow or keep learning (of course, that may or may not be practical :p).
Now you've finally matured into the kind of developer who can ask interesting questions - the type that are not easily answered by studying docs or source code. Sometimes because you're not asking the right questions, sometimes because your problem is obscure, sometimes because it's a little bit "subjective" - but above all, it should be interesting to other people who also know how to read docs and source.
So ... the type of question that gets rejected on StackOverflow.
I believe that site exists solely to punish me for doing google searches like "wiki something" to find a relevant related topic, and also to make me swear at google for removing the "block results from this site" feature from search, which, however temporarily, was awesome.
Earlier quoted context omitted.
Nope. Thats really the point - if we find the best deterministic algorithm, it means we have found the one with the best deterministic running time. By loosening our criteria of running time to expected running time and allowing randomness, we can sometimes do better than deterministic algorithms.
It doesn't square well with my intuition - if we understand how allowing randomness improves expected running time, then surely we must be able to create a deterministic algorithm that is at least as good as the randomized one. Am I missing some critical piece of understanding here?
if we pick the pivot to always be the first element, then any sorted array will cause worst-case running time. But by randomizing the pivot, we can do well (in expectation) for any input.
I learnt from electronics.stackexchange in the other direction: I applied my fragmentary knowledge to answering questions, as a hobby. A combination of googling, helping the questioner improve their question, and basic thinking has resulted in me getting about 10k rep and broadening my knowledge quite a bit. My top-voted answers are surprisingly general: http://electronics.stackexchange.com/questions/97277/when-ca...…
http://en.wikipedia.org/wiki/Integrated_circuit#Manufacturin...
The commentary about fab at home was not very interesting, either. The true answer is its very much like making a rocket engine, high performance, cheap, and safe all at the same time requires a nation-state, 2 outta 3 is going to take some engineering skill, only 1 achieved is actually pretty easy at home for an experimenter. Ditto the fab. Twenty people repeating that you're not going to make a ARM proc anytime soon is not terribly insightful, but mentioning the work of the "instruments of amplification" guys and insinuating that once you obtain double digit betas, the logical next step of on die bias networks for a simple class A amp or perhaps a RTL (R, not DTL or TTL) inverter...
SO is notoriously good at groupthink. Sometimes what you need is groupthink, so thats OK. Sometimes not.
I learnt from electronics.stackexchange in the other direction: I applied my fragmentary knowledge to answering questions, as a hobby. A combination of googling, helping the questioner improve their question, and basic thinking has resulted in me getting about 10k rep and broadening my knowledge quite a bit. My top-voted answers are surprisingly general: http://electronics.stackexchange.com/questions/97277/when-ca...…
The how to make an IC question is a good example of group-think on SO. The question was general and very wide open, but everyone jumped on "how to discourage someone from making a modern processor fab plant in their basement" bandwagon. Or not answering the question in a tangent, or going all snarky. The best 5 minute answer to the actual question is probably just to look it up in wikipedia as at http://en.wikipedia.…
EE.SO is very negative about "bad" questions, partly because it's small and can be run for the convenience of the dozen or so people who answer half the questions. I don't entirely approve but I see their point.