Eric Rescorla’s article “Is Finding Security Holes a Good Idea?” argues
that because large modern software
products such as Windows contain
many security bugs, removing an individual
bug makes little difference
to the likelihood that an attacker will
find exploits later in a product’s life.
However, a significant number of
exploits are based on vulnerability
information disclosed, whether explicitly
by researchers or implicitly
when manufacturers ship patches.
Rescorla therefore argues that, unless
discovered vulnerabilities are significantly
correlated, it’s best to avoid
vulnerability disclosure and minimize
patching. This is a novel and
disturbing argument against openness;
interestingly, it centers on vulnerability
statistics—which we
might be able to measure empirically
over time.