This is probably exactly what will happen, because the chain will not be able to support transactions directly in quantity, and so we end up with a plethora of middle men who are responsible for opening and maintaining payment channels. That scales pretty well, but a hybrid solution is probably best. Moores law and all that.
I think many recognize the value of federated 2nd layer solutions. Most are willing to accommodate some reasonable changes to the protocol to permit this. I think many would prefer a hard-fork 'cleaner' SegWit, but the softfork may have to do.
The real issue, I think, is that the main chain must remain a practical option for those that value it. Discovering the 'fair' cost for this while maintaining a secure network is very much an open question, and ultimately the source of all the controversy.
Personally I'd like to see unlimited block sizes, where txn fees and block orphan risk reach an equilibrium, but it's very difficult to say whether this would create unworkable centralization pressure.
