https://blog.ethereum.org/2014/05/24/on-long-term-cryptocurrency-distribution-models/Useful proof of work in ethereum is a very exciting idea to computer scientists. Using functions in EVM code as the proof of work is so recursive,.
1) The POW helps fund software and hardware research to run EVM code faster, which makes all other parts of ethereum work better too. Whereas asics for bitcoin only made the hash function fast, and everything else stayed slow.
2) This is a convenient and fair way to distribute initial coins.
This is a very complex process for ethereum, but truthcoin can have this advanced behavior with no modification to Paul's design.
By simply wording the predictions for an assurance contract correctly, we can crowd-fund the creation of any sort of cryptographic artifacts. We could pay people to repeatedly execute the python code that runs truthcoin. Someone would re-write all the python in C, and optimize it to an extreme. They would do this to mine more coins and get a reward.
I am starting to think Paul might be right about this: everything valuable ethereum could do, truthcoin could do even better.