github PRs are the best anti-spam we can get. Have you ever seen spambots trying to add links to their stuff in a pull request?
I would certainly not turn the main website more into a wiki than it already is. In many open source projects, I often find wikis to be slightly unreliable, with lots of contradicting informations, and sometimes out of date.
A recent example of this happened to me with Wireshark. A page on capturing the loopback interface, a single page, says:
1) It doesn't work and it is not possible to do it
2) It is possible, but you need some experimental software that has no binary releases yet
And it turns out, said software has actually been released some years ago and is perfectly supported.
So, a wiki is also a lot of work to keep up to date and accurate. I would even lean towards forcing a review of all changes before we accept them. Which is why the PR model sounds fine.
Just think about the information on haikuporter, supposedly in the github wiki. What do we point people to? "gentle introduction to haikuporter" (website) and "using haikuporter to build packages" (website). That is rather telling something about our use of wikis.