See primecoin.io for an example of more useful PoW.
What stops me from attacking that network by mining on my private chain with a double spend? I have no opportunity cost.
Nothing stops you mining on a private chain, but since you have a small minority of the compute power, the network doesn't care about your private chain. The same is true in Bitcoin.
> I have no opportunity cost.
You could be mining blocks that generate useful byproducts and currency.
Right, this is not an issue for small miners. But if you have a significant part of mining power, it's important that there is a significant opportunity cost mining on the wrong chain. That can only be ensured by using a PoW that is wasteful outside the currency you're mining.
Network value is different from individual value.
For them, the cost of attacking the network by mining on an orphaned chain is lower, because by definition the PoW has a useful by-product.
The more useful the byproduct, the less secure the coin.
Edit: to illustrate, imagine a PoW that computes cures for cancer. Somebody can now mine on whichever block of the chain he wants forever, since he's producing a cancer cure anyway. The incentive of a block reward (within the currency) is not primary. Thus there is diminished incentive for the network to create a linear blockchain, where everybody mines on the most recent block. Without that, you cannot order transactions.
A more apt analogy would be that each block would solve a protein folding or active site matching problem for cancer. Then each solution would be public, and useless to mine over again.
The PoW must only be useful for the coin you are mining and nothing rise.
Even then, as long as the coin value is larger you will have a consensus with the miners to to mine the latest block. Any sidechains will be neglected, which is why this "no value or nothing" attitude doesn't fit a blockchain network.
Well then they are doing useless work too, and the criticism of wasteful mining applies. You gain value from doing useful things.
You do understand how a blockchain works, right?
I can claim it's useless for miners and useful for researchers because I understand blockchains and proteonomics.
Do you understand either, VMG? Because it's fast becoming apparent otherwise.