You easily try to compromise the security of the first system as you know how it works, or you can report security problems very quickly because you know how it works; but as soon as you do that, millions of developers can correct the issue and make it more secure. Therefore, it quickly becomes harder and harder over time to find a vulnerability in the system as every one knows how it works and most of the problems are discovered and fixed.
By experimenting, you can also compromise the security of the second system. However, you (and million of the other users) cannot directly contribute to the development of the system or making it more secure, because it is a black box for you. The only thing that you can is to report the issue to the original developers and wait for a release that will (maybe) in the future fix some of those security issues.
Linux is an example of the first system. Windows is an example of the second one.
To summarize, the progress of "security through obscurity" is much slower that "security through transparency".