I'll add to my first comment. One could imagine a "whitelist attack" where someone would create a fake BOINC project, reward themselves a bunch of credits, and finally get enough mag to prevent themselves from getting voted out of the whitelist. They would be "locked in". Giving GRC balance a big enough voting weight will help eliminate this risk. Also having a large number of projects will dilute the per-project magnitude such that even a few fake projects would not be sufficient to dominate the voting system. In fact, just the last point makes a whitelist attack highly unlikely in the current state of the network.