The core question explored in this post: do highly decentralized DAOs work, or should their governance start to more closely resemble trad corporations?
Vitalik's main point: there are certain use cases where decentralization is very important. There may be a relatively small number of these orgs, but they will be the very important ones.