Conversation

2) There are pros and cons to each model. There's a spectrum. Generally, the closer to full token voting you get, the more decentralized and "pure" it is. The closer you get to dictatorship, the easier it is.

3) Also, though, left is only one kind of risk--another is e.g. YAM-tastrophies. and those are also more likely the closer you get to full token voting, because it's harder to patch bugs.

4) I find

style full governance really cool and powerful. I also find the original

@iearnfinance

multisig style fairly useful.

5) In the end, I think I mostly agree with

@VitalikButerin

(and I think (?)

@haydenzadams

though not sure) on this -- I lean towards: a) flexible at the very beginning to build b) transition to "anyone can build what they want but already built things are immutable"

6) the nice thing about (b) is that it means no one can make changes to steal funds, but also if there's an upgrade you can push it (to a new copy) and let people migrate (though if you do this too much it's a pain). And then, eventually, "specialized governance":

7) token-based governance over things like fees etc. which can't retroactively steal funds, and immutability over the fundamental behavior of the protocol (modulo starting a new version with new properties).