alternative proof of work

[jeremy.rutman2]

I have been trying to come up with an alternative proof of work allowing arbitrary calculations to be performed instead of hashing.

I blv I have a way to combine blockchain protection with e.g. folding (or any other difficult operation)
In case of proteins, based on a large library of proteins to be folded, in case of any other calculation a (large) library of such calculations.

The idea employs a blockchain just as btc except that the hash of
(transactions in the current block+hash of previous block)
is used to choose a problem from the library.  The lowest energy solution (for proteins, or best/only solution for other problems) is sought.
Whoever comes up with lowest local minimum within 10-minute block time gets block reward.

(solution + transactions + hash of previous block)
are hashed together for the next block a-la bitcoin.  Faking a block is now similarly difficult.
A bit more detail here http://telavivmakers.org/index.php/Protein_coin - including way to allow users to design algorithms yet prevent best coder/folder from gaining control.

[odolvlobo]

"Whoever comes up with lowest local minimum within 10-minute block time gets block reward."

How do you get around the need for synchronization, or detect someone spoofing the timestamp?

[jeremy.rutman2]

ya its a good point, consider otoh that the blockchain notion came from the 'timestamp server' which is just that which proves that a given operation was performed after time t, while what  would be needed is a way to prove that a given operation was performed before  time t

[sssubito]

To be usable as alternative proof of work your proposal must have the fundamental property to be fast and easy to check if a solution is valid and of course that finding the solution is arbitrarily hard.

I understand what folding proteins is about, but I don't see how it applies to the needed properties as proof of work.

How can you proove fast that a submitted solution is actually one that is valid compared to other submissions?

[jeremy.rutman2]

The valid folds are required to be local minima of the folding energy landscape.
These are easy to check (taking O(K) cycles in a K-dimensional landscape) but hard to find (taking something like O(A^K) cycles).

[jeremy.rutman2]

its an interesting puzzle
using a blockchain one can verify that a given block comes before a second, and if the second is timestamped
e.g. using a timestamp server
then the second definitely came before something that comes after time t...which tells you nothing about the timing

ya its a good point, consider otoh that the blockchain notion came from the 'timestamp server' which is just that which proves that a given operation was performed after time t, while what  would be needed is a way to prove that a given operation was performed before  time t