Yes, of course, you're totally bang on with your assessment. If there were a serious attack like this I suspect there might be a fork, maybe...
-----
If you want to perform such an attack regardless of cost, and with unlimited budget, I believe it has already been proven that the Byzantine general's problem is unsolvable, (lacking sufficient incentives)
It's likely that the incentive structure is the only thing that truly protects the blockchain.
A fork, sure, but a fork to what? Short of changing the hashing algorithm, I don't see what's going to prevent a destructive miner from shitting all over the new fork, too.
So: double-spends for direct economic gain don't seem likely to be a huge problem. Miners disrupting the network for political reasons (or ransom, for that matter) seems like it could be an issue, but it's hard to estimate how likely such an attack would be.
I've thought about this, actually. Not as a malicious attack, but in the (far more likely) event of getting stuck on the wrong end of a hard fork. And yeah, it's a problem.
Best case you're waiting many hours, perhaps days per block, until a difficulty change. (This is it's probably unacceptable to the mainstream, but to the participants of MPEX and bitcoin-assets --well-- I wouldn't be surprised if they were willing to accept several weeks or months of extremely slow blocks until a difficulty change. I'm not an insider, but that's the feeling that I get when reading the logs.
But that's just the best case scenario, in the worst case you'd be suffering attacks from the overwhelming majority of hash power (I think mitigation for this would be hard-coded nodes, attaching only to other nodes within the web of trust.) They could potentially operate a viable network with at minimum the 6 or 7 nodes they already run.
Both scenarios are not desirable. I think the algorithm change is probably the most likely outcome. It provides protection from asic attack, and, let's be honest , there is a strong incentive to bring back cpu or at least gpu-friendly mining.
In any case, I think a fork is risky, more risky than most people realize, and I think that
Mircea Popescu and his followers are a formidable and compelling group and should not be discounted.
Interesting times ahead.
A handful of loons. I discount then. Why wouldn't you?