Time to slow down?

It's worthy to have less bugs. It's not worthy to just have less reported bugs. As long as
there are factors limiting testing (because the process is developer, not tester oriented) you
only have access to the second number, and can't really assess coder quality.

A better metric would IMHO be the mean time a bug gets to be fixed. If this time gets low
enough one could argue the bottleneck is no longer developers but testers.