Actually, being a D* podmin for 7 years gave me a lot of time to reflect on fair moderation. Good moderation means letting people say things that you disagree with, essentially, as long as it is legal content. However, as the owner of a webserver, I do have the prerogative to lay down the rules of what can be served there, and so a part of good moderation is setting up a fair and rational TOS that is publicly posted, and applying those rules fairly to all accounts, equally.
For example, a rational TOS would not have people banned for complaining about CP on the network. A rational TOS would not ban people for "too many comments" on a github issue ticket without defining what the time limits for posting are. Spam is unsolicited commercial advertisment posts, so those are a clearly-defined special category.
I think meta-moderation, where people are randomly assigned to review each other's mod actions for sensibility, and some consensus/reputation score could be utilized to arbitrate that. I also think random rotation of moderatorship to people with good recent reputation/kharma over a 6 or 12 month period could be utilised, to great effect, to prevent "tyrant" mods.
In a federated network such as Diaspora, I think you need some tools for federation of the moderatorship actions, such as the ability to share blocklists among pods, with standardized TOS terms that several pods could subscribe to. That's what I suggested in their github issues system, right before they banned me.