Well, if your starting point is that "open source doesn't lead to bugs being identified and disclosed" then those very posters you are complaining against are partially right, in part. Consider:
Open source: anyone can read the code, but (based on our premise) this doesn't lead to identification and disclosure of problems. It can allow a prospective attacker to identify problems and not disclose.
Closed source: only internal staff can read the code, but (based on our premise) having many eyes looking doesn't lead to identification and disclosure of problems. Prospective attackers can only do binary analysis, not source analysis, to find problems.
If binary analysis is more difficult than source analysis for finding potential bugs (i.e. potential targets for attack) then closed source is more secure in this context (assuming one or more attackers looking for potential vulnerabilities in the library/source/whatever).
Note: I'm not agreeing with the 'ubiquity' argument because it ignores read distributions of OSs. Also I'm not agreeing with the 'financial interest' arguments, because in a closed source there is the possibility that a company will gamble on an internally-detected vulnerability not being exploitable (or exploited) rather than fix it.
There are valid arguments for using open-source software, but I don't think the "many eyes" argument is necessarily a good one.