[CraigRo]

First of all, nobody ever passed these exams with flying colors unless the examiners wanted that to happen or were having a bad day. Assume that there are three full professors testing you, and they have an average of 20 years each of post-graduate experience. That's 60 years of reading -- no way that a 2nd or 3rd year graduate student can keep up. Part of the process is humiliating the students a bit -- they did this for everyone.

They also had a language requirement, which was often fulfilled by memorizing the 500 most common words in a foreign language as well as 300-500 common math terms in that language; there was a library of prepared crib sheets for these. It was kind of a running joke in that it tested your ability to translate an article from German/Russian/French into English sufficiently well to explain it to a bored examiner, but most of the students kind of looked it as a kind of giant 'Wheel of Fortune' exercise in intelligently guessing between the revealed clues.

Second, some of the other graduate students basically went insane studying for this -- one locked himself in a room with a 30 day supply of microwavable meals, and another locked himself in his dorm and (having read MicroSerfs) only ate 2 dimensional food slipped under the door. A lot of this was immature students trying to out-do each other in commitment and probably didn't markedly improve pass rates.

Tao matured, but mostly because he got older -- there were finally reasonably smart people his age studying with him. He also used to hang out with John Conway a lot, and Conway was going through his own problems; he spent more time with the 'more conventionally normal' faculty and that probably rubbed off.

[tester756]

>Assume that there are three full professors testing you, and they have an average of 20 years each of post-graduate experience. That's 60 years of reading

who cares?

10 juniors with year of experience don't make it even close to decade of experience

[GTP]

Yes, what you're pointing out is that it doesn't scale linearly. But the point still stands.

[2-718-281-828]

that's just guessing. the point could also sit. what you mean to say is that it is still there. how is not relevant to your argument.

[CraigRo]

Typically exams are 2 full professors and maybe one asst prof. That asst prof will be 3-4 years in, plus 4-5 years of grad school + often a postdoc. And that asst prof will, on average, be better than the graduate student.