Yes, I instinctively agree with you. But what about sink investments? Also, what about borderline protocol implementation improvements? Death of a thousand papercuts...that kind of thing?
Does keeping the algorithm source in the open, which allows for forks when opportunity presents itself...is that pretty much the optimal strategy against any transaction protocol being abused?
It is, obviously, because if there's no agreement, then nobody (or not enough people to matter) adopt said protocol.