i posted this in a different thread, but it fits better here.

Since the bitcoin network is considered secure by the userbase right now (and was considered secure a year ago, too), is there a security need to constantly increase the difficulty? Is the increasing difficulty superfluous from a SECURITY standpoint?

Because if it is, could the bitcoin client not be forced to work out a certain number of scientific (or otherwise useful) calculations for every block generated? I'm not a computer scientist, but here's my rough idea: Basically, the server would only accept blocks if your client sends them with a certain numbers of valid keys. Those keys would be distributed to your client by a server upon the completion of a set number of completed tasks from a "useful" project like seti or protein folding by your client.

This way, the difficulty of actually creating bitcoins could be constant, and the architecture for that would not have to be seriously modified. As time went on, and more people set up mining rigs, the number of required keys would adjust, raising the ratio of GPU time spent on scientific calculations to bitcoin calculations and effectively raising the difficulty.

I feel this would be great from a scientific standpoint- but also from a PR standpoint. Bitcoin would have few detractors if it were everything it is AS WELL AS the largest source of computer power available for scientific calculations.

The transaction fees that will exist in the future could even be avoided if laboratories were charged a small fee (much smaller than what they would otherwise pay) for use of the computing time.