Most [well-written] code doesn't corrupt itself. It requires people to get involved, who, in turn, corrupt it. Bitcoin's claim to fame is taking people's corruptibility & effectively removing it from the equation (hence "trustless").
If not, Bitcoin's no better at being money than ordinary email/texting. As long as everyone is smart and honest, why even bother with crypto?
Then you can look at XT/BU/Classic as some sort of test. If we fork once like this, there's a much higher probability that it won't stop there.
Perhaps.
Or perhaps the blocksize limit being in protocol layer (when it was supposed to be a temporary spam measure) is an anomaly.