This should be easy to verify with sampling: I mean count the stars in a spot in the sky and extrapolate to the rest. Actually the recent Sky and Telescope art

This should be easy to verify with sampling: I mean count the stars in a spot in the sky and extrapolate to the rest. Actually the recent Sky and Telescope article about killer asteroids mentioned another interesting form of sampling. The article said that astronomers have a high confidence that 90% of all large asteroids which cross earth's orbit are known. But how can this be? This is the argument: we "tag" an asteroid when we find it. Then we look for these types of asteroids over some period of time. 90% of the asteroids found in this period are already tagged ones, so we know only 10% remain. I'm trying to get my head around the reliability of this method- at the very least the sampling has to be non-biased and sensitive.