[Amincd]

That's a good point, but if he studied the protocol, he should still know when bitcoins stop being generated and how divisible they are.

---You could tie an upgrade of the protocol to a consensus by a majority (or perhaps supermajority, if you're clever) of mining strength.---

The problem then would be that a >50% attack could overtake the entire network by inserting malicious code in the upgrade.

As it is now, the damage that a >50% attack could do is somewhat limited as long as it doesn't persist for too long.

---Or you could try to give up the total decentralization of Bitcoin, given that it's an illusion anyway and the source of most of the expense, security problems, and unwieldiness that will come to haunt the protocol.---

Decentralization is the only way a program that would be as targeted by law enforcement as bitcoin can survive.

[antiscam]

Bitcoin's not at all resilient in the face of an attack by a government.

You're right about limiting damage from a 50% attack, however. Of course, you could agree in advance on an amount of time the majority of hashing power would have to assert that an upgrade was needed, but it's possible that would raise other problems. It may be that there's no good way to handle upgrades without an out-of-band mechanism, but if so, that counts as a strike against Bitcoin, not one in its favor.