bytemaster , you should look at PPCoin and proof-of-stake
I have looked at those systems and the proof-of-stake creates a whole new can of worms, but is on the right track of attempting to securely reach a consensus without any single actor being able to dictate terms.
Part of the consensus process would be factoring in the weighted average coin-age spent in a block. Any blocks that do not spend trx referencing old outputs that have been broadcast to the network would automatically be suspect. This would prevent someone from moving a lot of coin between their own accounts to help drive consensus.
For someone to 'attack' this network they would have to:
1) Set the mining reward on the blocks they generate to 0... this means they would not be able to recycle / profit from fees on their own transactions.
2) Have enough hashing power to be the first one to find a hash after 10 minutes have passed.
3) Generate blocks that follow all of the rules of the network, including 90% of known valid broadcast transactions prioritized by dividends paid and weighted by coin-age.
To perform a 'double spend' the spender would have to isolate a single client from the global consensus, but this attack would not be sustainable because of the coin-age requirement as well as validation that you have the right consensus that occurs off-chain. Imagine every major business published & signed the consensus hash. An attacker would be unable to isolate someone and convince them they had received funds based upon a bogus consensus / double spend.
Lets put things another way, assume the government *did* invest in ASIC hardware and started controlling 90% of the hashing power? How would the community respond? Change the hashing algorithm? What if the government declared a war-on-crypto and intended to out mine any new chain ever created so that they could censor transactions or accumulate the entire money supply or implement a tax or registration scheme?
Moral of the story... CPU, GPU or ASIC power is centralized control.