The decision-making process of the future will be dominated by the organizations that provide the backbone of block generation. If they don't like a change in the code, they'll refuse to upgrade but will still keep 50%+ of the CPU power. A major dispute between bitcoin.org and the backbone would lead to a serious fork.
Do you think the damage/inconvenience this would cause would make a strong incentive to just start a new chain with the different rules?
I was thinking this is a nice thing about bitcoin. Only a majority is required for an attack, but you need even more than a simple majority for a smooth rule change (only for certain kinds of changes I realize).
Even if you get 60% to say they'd prefer it to change, will they want it enough to work on a network 60% the size? If half of them won't give up that much then you've only got 30%. Not to mention the difficulty of gauging how much support you have before you actually try. "I'll switch and I have 20000khash!" Okay, add up all the claims...