I just hit Google groups for the searches "postgres" (90,200), "mysql" (1,030,00), "postgres crashed database" (223) and "mysql crashed database" (1720). So that is slightly out of proportion.

However I look at that number, and I realize that there is going to be some self-selection bias. First of all since a lot of people believe that postgres versus mysql is partly a reliability issue, people who select mysql may be more willing to just restart after a crash. People who use postgres want better recoverability, and so are more likely to try to debug, have theoretical discussions, etc.

Based on some spot-checking of the searches, it appears that there is some validity to the theory. A lot of the MySQL messages are basically, "My database crashed." The postgres ones are discussions of transaction semantics, crashes in version 6 (old and rather unreliable) comparisons between MySQL and postgres, etc. Of the first 10 results, the postgres one only has 3 which seem to be a version of postgres 7.0 or better crashing, while all 10 of the MySQL messages are about MySQL crashing.

Cheers,
Ben