For any remotely plausible definition of "worthless," bitcoins are just as worthless as any representative money, which inclues fiat money (like a USD) and commodity-backed money (like a gold certificate). The only reason money can be used in general transactions is that people choose to accept it, and they choose to accept it because they are reasonably confident they will be able to trade it again in the future for something of at least equal value.

This is the fundamental answer to "why is any representative money valuable?" If people don't have confidence that they can offer the money in the future, they won't accept the money today.

> To me the real currency would have to be something in the realm of Folding@home. I'm not sure how it would work of if its computationally feasible, but I do know people would pay for the results of useful computations such as that.

As far as I know, that can't happen, at least not in a proof of work system. The work in a proof of work system can't be severable from the specific information it is designed to secure (in Bitcoin, that would be transactions on the block chain). This isn't exactly a scholarly source, but it's from the "Chief Cryptographer for the Ripple protocol": http://bitcoin.stackexchange.com/questions/11649/intrinsical...

I'm not very familiar with other cryptocurrency systems, like proof of stake, but I know there are some attempts at cryptocurrencies that use computation with unrelated useful side effects.