You could just remove the reward, any one can mine new block out of the mem pool, if two blocks or tx are in common, a determinstic algorithm could be used to select between the two.
I agree with you. The error in most crypto is the reward, which gives rise to strategies that do not necessarily induce the desired properties. I also think that the only viable kind of crypto currency is where the validation/consensus decision is taken on a voluntary basis, the "reward" being that the system in which you are invested, keeps running correctly.
However, you still need a kind of deterministic decision *that is hard to game* (because you can do "proof of work" like calculations to get the deterministic solution in your advantage). This is why a kind of PoS signature scheme is necessary in my opinion.
Quote
With the hash of previous block in the header including timestamp for me it's enough to prevent sybil attack. Checkpoint could be made every 100 blocks and hashed in the chain.
Checkpoints are no solution, because they are just another consensus problem. If you have two conflicting check points, which one is the "right" one ? You've just transposed the block consensus to the check point consensus.
Yes, you can *individually* decide that you won't allow any old modification of consensus. But there is no guarantee that the rest of the network will follow you.
But mainly, yes, PoW is a bad idea, and rewarding (with fees or coin creation) consensus deciders/maintainers is also a bad idea.
The thing is, ok it's easy to game, but then, there is nothing to win either
The whole problem of sybil attack become almost nothing.
Because even let say you have a whole new blockchain of 1 millions tx, vs the local blockchain.
Either
1 All the tx are identical, and it's just about reordering them to end with the same block header, which just involved sorting them on any kind of value that can be the same for every node, could be based on the hash or something else, even only the timestamp, and that just involve recomputing the merkle root and the block header hash, period.
2 One of the blockchain contain tx that the other doesn't have, and in the end, it's same than 1, just need to insert the transaction in the good block and recomputing merkle root and block header hash.
3 One of the blockchain contain a double spent for the other, and it's only in this case that there is any kind of need for a hard consensus, either via POW or other.
For all the other cases than 3, all the nodes can already know how to get the tx and block hash in the same manner than all the other node, if an algorithm is made in sort than given a set of N tx, it always end up with the same chain of block headers, then they just have to know they are on the same set of tx, and they have naturally the same blockchain.
Looking into what they are doing with the XThin block with bloom filters and that sort of things is intersting in this regard.
I'm 100% convinced if the goal is only to get to some sort of distributed ledger that protect against double spent, there are much much less expansive / risky manner to reach consensus than the game theory of reward & mining.
It's not that the mining is a 'bad idea', but it's clearly not made in sort to be an efficient distributed ledger system. It's made to induce game theory.
When you think about it, the only thing that make reward / mining game interesting is the speculative value on the other side, because if there is no speculative value, the block reward worth nothing.
In the end there is necessarily a direct connection between the speculative value of the coin on one side, and the cost invested in the mining risk taking. The two balance each other out.
And then it need some kind of 'clock' to be connected to real time economy, and it's how you get the fixed block time emission. Like this the whole thing is 'hard pegged' to real time economy, with the predictible inflation rate etc.
Remove all this problematic of speculation from the equation, and it become much simpler as a simple open distributed ledger. And the pow become overkill if it's just to solve double spend or sybil attack.