Because wikipedia is open source, I can modify anything I want. But what happens if someone deletes, adds bad content, modifies useful content ? What does wikipedia do to prevent that ?
The last question is available for any other open source software: linux, php etc. How do they know which contribution is useful and which is harmful ? Thank you.
Review Process
In the case of Wikipedia, this is done in an ad-hoc method by peers with some exceptions. Highly contested pages are moderated and changes must be approved.
Similarly, when a patch is submitted to an open-source project, a review process is done to ensure the patch doesn't add new bugs or isn't malicious. Of course, a review process can never be perfect and catch all possible errors a patch might cause, but due diligence is typically observed in respectable open-source projects.
Open-source development isn't quite the same as Wikipedia. For one, developers can still control what code they accept. For another, there's nothing to stop you from creating your own version with custom changes. This is done either to test them prior to submitting them upstream, or if there is some disagreement about what should or shouldn't go in.
Open-source developers typically look at, run, and test all code they receive, and rely on reputation to determine how much scrutiny someone's code needs.
The Linux kernel is a great example. I run a kernel with Con Kolivas' patch sets. These patches will probably never be accepted into the main kernel, but he keeps his patches on his website in case someone wants to use his modifications anyway. It's not like he's created a whole new kernel, he just needs to keep updating the patches occasionally so they still work.
Linus Torvalds covers some of these ideas in his Google Tech Talk about his Git VCS:
http://www.youtube.com/watch?v=4XpnKHJAok8
Particularly interesting is his concept of a "web of trust" and his strategy of appointing people in charge of certain parts of the kernel (for example, Alan Cox used to be in charge of the TTY code and personally approved all patches for that part of the code.)
Eric S. Raymond has also written a lot of great material about open source development in general. His relevant papers are listed here:
http://catb.org/esr/writings/homesteading/
There was a classic example of how open source projects keep out malicious code when an attacker attempted to submit a patch to the Linux kernel containing a backdoor that allowed root access. There is an article here about the attempted attack.
It was found when alert reviewers noticed the suspicious changes. See here. The malicious change was caught before it made its way into an official kernel release.
See the code review section of the excellent "Producing Open Source Software" book.
It comes down to having a version control system (to track and when needed revert changes) and having a good code review system in place.