[draft] ideas for defending against wild malicious political ais

Wiki admin: - an open system run by a wiki and logged to the same wiki that uses machine learning to "moderate" a discussion forum (where "moderate" is whatever behaviors the community appreciates) - provides for a way to freeze / fork the algorithm so people can keep things how they want when they want - likely will need to engage in behaviors of information security too. A defense AI would end up tracking the goals of the malicious AIs and "dancing" with these goals in some way. Once the system is together enough to identify goals to some vague degree, we can begin mediating with them to reduce the influence they have on us.