I'm curious as to why ETH should fetch a large value. Bitcoin has value because it is explicitly a store of value, as well as a payment system. Transfer fees are pretty much insignificant (cents). Ethereum's value only comes from it's ability to run decentralized Turing-complete code. Since you're paying per execution, most people will actually want the value of ETH to be low as possible, so they can execute their programs as cheaply as possible. This will lead to people rapidly dumping when they feel the price is too high, and sitting on the sidelines to wait until prices are reasonable and then buy in and quickly run their scripts at a cheaper price. Or, more likely, most people won't hold ETH at all, and only buy in a small amount at market price to cover the cost of whatever script they need to run. Large initial inflation combined with no software running on the network in the beginning will mean ETH effectively won't have any intrinsic value until the first programs are released. Any rises in value after the IPO and before the release of programs will be purely speculative.

The coefficients will be revised as more hard data on the relative computational cost of each operation becomes available. The hardest part will be setting the value of x. There are currently two main solutions that we are considering:

Make x inversely proportional to the square root of the difficulty, so x = floor(10^21 / floor(difficulty ^ 0.5)). This automatically adjusts fees down as the value of ether goes up, and adjusts fees down as computers get more powerful due to Moore's Law.
Use proof of stake voting to determine the fees. In theory, stakeholders do not benefit directly from fees going up or down, so their incentives would be to make the decision that would maximize the value of the network.
A hybrid solution is also possible, using proof of stake voting, but with the inverse square root mechanism as an initial policy.