Please tell me if you agree an ossifying of the protocol - the fact it will become increasingly hard, probably impossible to make changes as adoption grows - is what we'll likely see.
Not that I was asked, but I'll offer an opinion anyways.
At worst harder, but not impossible.
Today we have a sort of self-enforced (by the core devs) consensus system, plus of course the ultimate ability to vote with your node and with your mining capacity. I wouldn't expect the latter to ever change (indeed some blocksize limit is required to maintain this goal). For the former, however, I doubt that having this little governance around important changes to Bitcoin will last forever -- 20 years hence I would expect a much more regimented procedure, somewhat more akin to a standards organization than what we have today (perhaps with a combination of academic, miner, and corporate interests represented, but that'd be an argument for a different thread).
More governance is both bad and good-- in particular on the good side, bright lines can be drawn when it comes to voting in a way that doesn't happen so much today. If the ISO can finally manage to crank out C++11, despite the contentious issues and compromises that were ultimately required (and C++14 just two months ago too!), pretty much anything is possible IMO.
If you're that worried about a ossification, perhaps you'd prefer a dead man's switch: in 20 years, the blocksize reverts to its current 1 MB.