Congratulations on being in the top, er… (small number)% of nethack players!
I ascended a Tourist after successfully running the protection racket. Sadly I did it playing locally, not on a public server, so I don’t have a public record of this, and I’ve not had much time for playing nethack lately in order to try again on nao.
After the discussion of M:tG I’ve actually been trying to think of possible lessons on rationality that could be drawn from nethack; it’s an interesting case because it’s very difficult, but virtually every death is caused by pure player error.
virtually every death is caused by pure player error.
Wow, that is not my impression. Nethack inflicts random sudden death. Nethack works dramatically by repeatedly putting you in situations where you have to choose between certain death and possible death.
I’ve never read the source code, so my perspective is limited. Does your view of nethack as a rationality game consider reading the source code to be cheating, or to be a prerequisite for playing?
I’m undecided on whether to call optimal Nethack playing “rational” or “using a big lookup table”.
Wow, that is not my impression. Nethack inflicts random sudden death. Nethack works dramatically by repeatedly putting you in situations where you have to choose between certain death and possible death.
There are a few sudden deaths not caused by some degree of player error, all in the early game, but even those could be avoided most of the time with sufficiently paranoid play (that most people would find crushingly dull). Beyond annoying things like gnomes with Wands of Death or spike pits with lethal poison, most deaths are caused by reckless combat or failing to take appropriate countermeasures against certain monster types. Skilled nethack players have win ratios of over 60%, vs. people who play for years and never win once.
As evidence, on nethack.alt.org’s top deaths list, 10 of the 11 most common game ends are “killed by a (weak monster)”, all of which are likely to be player error, including “killed by a water moccasin”, which is all but guaranteed to be egregious player error. The 3rd most common is “killed by a wand”, which encompases one of the most common “unfair” deaths as well as a fair number of player error deaths. The 12th most common “death” is winning the game. After that, the 13th is again egregious player error, after which are piles more deaths by careless combat, with a few “killed by a *, while helpless” deaths that are again egregious player error.
Reading the source code isn’t cheating per se—it certainly doesn’t guarantee winning—but it’s not required, either. There are a handful of simple spoilers that help a great deal, but most of what you really NEED to know could fit on an index card. Among regular nethack players, almost all are “spoiled” to the hilt, but huge variance in success remains, because some people are better at making rational, methodical estimations of what they can safely do to advance the in-game goals.
Congratulations on being in the top, er… (small number)% of nethack players!
I ascended a Tourist after successfully running the protection racket. Sadly I did it playing locally, not on a public server, so I don’t have a public record of this, and I’ve not had much time for playing nethack lately in order to try again on nao.
After the discussion of M:tG I’ve actually been trying to think of possible lessons on rationality that could be drawn from nethack; it’s an interesting case because it’s very difficult, but virtually every death is caused by pure player error.
Wow, that is not my impression. Nethack inflicts random sudden death. Nethack works dramatically by repeatedly putting you in situations where you have to choose between certain death and possible death.
I’ve never read the source code, so my perspective is limited. Does your view of nethack as a rationality game consider reading the source code to be cheating, or to be a prerequisite for playing?
I’m undecided on whether to call optimal Nethack playing “rational” or “using a big lookup table”.
There are a few sudden deaths not caused by some degree of player error, all in the early game, but even those could be avoided most of the time with sufficiently paranoid play (that most people would find crushingly dull). Beyond annoying things like gnomes with Wands of Death or spike pits with lethal poison, most deaths are caused by reckless combat or failing to take appropriate countermeasures against certain monster types. Skilled nethack players have win ratios of over 60%, vs. people who play for years and never win once.
As evidence, on nethack.alt.org’s top deaths list, 10 of the 11 most common game ends are “killed by a (weak monster)”, all of which are likely to be player error, including “killed by a water moccasin”, which is all but guaranteed to be egregious player error. The 3rd most common is “killed by a wand”, which encompases one of the most common “unfair” deaths as well as a fair number of player error deaths. The 12th most common “death” is winning the game. After that, the 13th is again egregious player error, after which are piles more deaths by careless combat, with a few “killed by a *, while helpless” deaths that are again egregious player error.
Reading the source code isn’t cheating per se—it certainly doesn’t guarantee winning—but it’s not required, either. There are a handful of simple spoilers that help a great deal, but most of what you really NEED to know could fit on an index card. Among regular nethack players, almost all are “spoiled” to the hilt, but huge variance in success remains, because some people are better at making rational, methodical estimations of what they can safely do to advance the in-game goals.