retchdog:~$ time brew install gcc
==> Installing dependencies for gcc: gmp, mpfr, libmpc, isl, cloog
[blah blah blah]
==> Installing gcc
==> Downloading https://downloads.sf.net/proje...
==> Pouring gcc-4.9.2_1.yosemite.bottle.tar.gz
==> Summary
real 3m51.241s
4 minutes, and that was just because my wifi sucks.
O_o
why is a glorified build manager using kernel-specific features?
once again, science and technology triumph where religion has shown mixed results at best!
the previously legendary properties of the River Ganges are now firmly established! at least until the resistance evolves.
this is one of those "Windows is better because it runs $foo"-type arguments. fair enough, but it doesn't fair so well when people use it against linux, for some reason.
anyway, i'll happily concede that everyone doing embedded development is better off not using a Mac or Mac OS X for many good reasons.
well, iterm 2 is just awesome, even though i usually don't bother using it because the vanilla Terminal.app does everything i need. what am i missing?
yeah but macports is an unstable cluster fuck.
homebrew has huge amounts of paranoid safety-checking and (so far) has not come crashing down in horribleness. it can even self-diagnose what's changed if something overwrites some of its installed files.
i haven't found anything i need that isn't in homebrew.
before that, i agree, it was pretty awful; macports sucks. however, you could build all of it yourself. i knew a very good developer who built a rock-solid GNU environment in OS X; took him a while though. i'll grant that linux is easier for getting up and running into dev without knowing/bothering with the setup.
so, the big problem with Mac OS is that you don't know the keyboard shortcuts. duly noted.
perhaps, but what can linux do that OS X cannot?
uh, i was referring to the general utility of the OS. so, what would inspire me to consider putting up with that trade-off? what do i get in exchange?
what on earth could make Linux a more useful OS than Mac OS X?
Windows i get for specialty applications like gaming and such, but Linux?
I also share them with people whose opinions I respect...
it's good that you've now explicitly admitted that you have contempt for your audience. this is something your critics have been saying for years. so, why not move on to a more suitable forum or, perhaps, a psychiatrist?
yeah, like how in Soviet Russia you could also submit a vote for anyone.
when i read the waffling "i don't have any evidence, but it's a good bet that i know everything anyway!" bullshit in the lede.
actually i was a bit vague with the order of the quantifiers. to clarify, there are some deck orderings where it will lose to itself and might lose to someone else (otherwise it would always win, which would be quite surprising). but at least if the deck is shuffled well, it cannot be strategized against except to break even at best.
No, no, no, no, NO! If you think that, you've completely missed the point.
There is no way to trick this strategy; it cannot be cajoled into a corner, because it knows all of them already. This strategy isn't optimal in the sense of being good, it's optimal in the sense that it cannot be tricked. Think instead of tic-tac-toe for a moment. Would you say that you could always beat someone at tic-tac-toe if you know their play-style? Of course not, because it's very easy to use a play-style wherein you can force a draw, always. Well, that's exactly what they've done here (with one important caveat, below). But what about randomness, you ask? That's why they need ~10^14 states! It doesn't care how the cards are shuffled, because it cannot be tricked on any ordering of the cards. It doesn't matter; the randomness of the deck just selects which one, but the strategy will work on whichever one is picked.
Now, the caveat comes in the form that the probability of winning is simply bounded below at 1/2. A slightly suboptimal player with 10x as much seed money will probably have an advantage just by being able to bankrupt this strategy. A mediocre player with 10000x as much money will also have a good chance of bankrupting this strategy before the guarantee kicks in. Think of having more money as a sort of "complexity measure"; this strategy forces an opponent to rely on having a greater investment just to have a chance, sort of like asymmetric crypto forces an opponent to have exponentially much compute power to read your messages.
This caveat is because they're sort of mashing up the techniques of deterministic game theory, with a probabilistic game. Nonetheless, the advantage is there. If you're playing a completely unknown adversary with equal funding, and you want to do your best, this is the way to go. End of story. Of course things get interesting in reality because no adversary is completely unknown; you can always try to Vizzini your way into the opponent's head but this is a dangerous move which can backfire drastically.
"Engineering without management is art." -- Jeff Johnson
