I was somewhat disappointed to find a lack of Magic: the Gathering players on LessWrong when I asked about it in the off-topic thread. You see, competitive Magic is one of the best, most demanding rationality battlefields that I know about. Furthermore, Magic is discussed extensively on the Internet, and many articles in which people try to explain how to become a better Magic player are, essentially, describing how to become more rational: how to better learn from experience, make judgments from noisy data, and (yes) overcome biases that interfere with one's ability to make better decisions.
Because people here don't play Magic, I can't simply link to those articles and say, "Here. Go read." I have to put everything into context, because Magic jargon has become its own language, distinct from English. Think I'm kidding? I was able to follow match coverage written in French using nothing but my knowledge of Magic-ese and what I remembered from my high school Spanish classes. Instead of simply linking, in order to give you the full effect, I'd have to undertake a project equivalent to translating a work in a foreign language.
So it is with great trepidation that I give you, untranslated, one of the "classics" of Magic literature.
Stuck In The Middle With Bruce by John F. Rizzo.
Now, John "Friggin'" Rizzo isn't one of the great Magic players. Far from it. He is, however, one of the great Magic writers, to the extent that the adjective "great" can be applied to someone who writes about Magic. His bizarre stream-of-consciousness writing style, personal stories, and strongly held opinions have made him a legend in the Magic community. "Stuck in the Middle with Bruce" is his most famous work, as incomprehensible as it may be to those who don't speak our language (and even to those that do).
So, why am I choosing to direct you to this particular piece of writing? Well, although Rizzo doesn't know much about winning, he knows an awful lot about what causes people to lose, and that's the topic of this particular piece - people's need to lose.
Does Bruce whisper into your ear, too?
I'm going to take a stab at explaining/translating some of the examples from that article. The first is about "mana screw". In Magic: the Gathering, each player has a deck of 60 cards, of which about 2/5 are "land" or "mana", and the rest of which are "spells". Each turn, players draw one new card, put down one land card if they have it, and play spells whose total cost is less than or equal to the number of lands they have. Costs are typically distributed in a bell curve centered at about 3.5. If a player has too few lands relative to the cost of his spells, he can't play them; this is called "mana screw". If he has too many lands, he won't have spells to play with them; this is called "mana flood". Lands and spells also have color; to play a spell, some number of the lands used must be the correct color or colors.
At the start of the game, each player draws 7 cards, looks at them, and decides to either keep them or mulligan, which means he puts them back, reshuffles, and draws a new hand with only 6 cards. (If the 6 card hand is also bad, he can do it again, getting one less card each time.) If the initial 7 cards are worse than average (too few, too many, or the wrong color lands), then the player chooses between a small certain loss (one less card) and an uncertain large risk (you might not draw the lands you need, and be mana screwed). In practice, most players are strongly biased towards keeping hands they shouldn't, which means accepting the uncertain large risk over the certain small loss.
The second issue is "netdecking". Players can either choose the cards in their deck themselves, or get a deck from the internet, usually by looking at a recent tournament and copying the winner's deck, or if they're really serious about it, guessing which decks they're likely to face, choosing a pool of candidate decks, looking at win/loss statistics, and choosing the deck which gives the best chance. Copying a deck that's known to win is much more effective, but making a deck yourself is more fun. As a convenient side-effect, making your own deck gives an excuse for losing.
The third issue, which the article alludes to but doesn't tackle directly, is that MtG involves a large number of easy decisions, with a small number of hard decisions mixed in but not clearly labeled. Players who think too long about all of their decisions are chided for stalling; on the other hand, players who fail to slow down and think will often lose the game because of doing so. Most players can't tell the difference between a hard decision that requires thought, and an easy decision that can be made quickly; instead, they make all decisions quickly until they know they're on the verge of losing; then they switch to thinking carefully about every decision, but at that point it's usually too late.
There's one thing in the article that even I had to use some Google-fu on, although it's not really significant. "The Ron" is Theron Martin, who was suspended for five years for cheating - his DCI rating had been artificially inflated because someone had been sending the DCI bogus tournament results in which he won games that never happened. Theron Martin claimed that he was innocent because he didn't know it was happening.