I predict that "web3" will follow the same trajectory as "ML/AI" -- keep stuffing more and more completely different models and techniques and technologies in under the same vague heading until one of them eventually starts to do something really useful and then credit that to the nebulous term itself.