If there are too many false-positives, it just becomes an annoyance. If the false-positives are too high, it can have the opposite effect, that code detected to have bugs is more likely to not have a bug. The false-positive max is around 5% I think. Anything above that - it would be more effective showing warnings at random locations.
If I read it correctly, that 10% was only for the static checks flagged in code reviews. They have some checks that are 0%.
Which is to say there are some checks that are always right and google doesn't allow violations. The 10% seems to refer to checks that catch some significant bugs - things bad enough that they are worth looking at, but the check isn't perfect.
