The government's attempt to control the dissemination of encryption techniques and algorithms illustrates my point best. If knowing the implementation of an encryption algorithm makes it less secure, then it was never secure in the first place.
I'm not an authority on this, but my impression about the reasoning behind the U.S. ban on "exporting" certain strong encryption methods was that these methods were considered "too strong". The perceived problem was not that America's own encrypted traffic might become compromised, but rather that the U.S. didn't want others using these methods, because then it would be harder for the U.S. to track what others were doing. I could be wrong about that, but it seemed plausible when I first heard the idea.