A Case for Proof-of-Excellence

At the time of this writing, proof-of-work (POW) as a consensus algorithm is losing favor for the reasons well documented other places. One of the primary issues is massive energy consumption by mining. Not pointing fingers, I’m as guilty as anyone as I sit here listening to the fans on GPUs busy mining Monero and Aeon. However, as bad as wasting power is, I feel the most tragic waste is the computing power and worse yet the intellect involved.

Imagine if all the smart people with all the high-end compute equipment were working on solving problems which mattered to society. I’m suggesting that instead of a block reward based on solving a meaningless algorithm, the reward was based on solving a scientific problem, a Proof-of-Excellence, an idea conceived by King and Nadal. By creating tournaments based on a meaningful problem, all of the resources consumed would still generate the wealth of a POW but with a collateral of impact of some benefit to society.

There are significant challenges to implementation:

1. Type of problem. The problem, or problem steps, have to be known to be solvable in a reasonable amount of time to a human -seconds to weeks- vs months or years.
2. Validation. A successful result would need authoritative validation in a reasonable amount of time either by a human or machine
3. Validation Criteria. Must be measurable based on the problem
4. Brute Force Attacks. If for instance a particular genetic algorithm were the subject of a tournament, perhaps the most efficient algorithm in the public domain with the most hashing power would always win
5. Energy Consumption. Employing AI in this competitive manner could lead to even higher and more variable energy consumption than exists today with POW
6. Cost estimation. Because of solution variance, it is difficult to predict cost efficiency of participation and participation may exceed the benefit

I propose a solution to most of these problems is to create a framework which is an elaboration on Augur. Problems could be suggested to “make a market” where anyone can participate in 1 or n tournaments. The market would describe the problem, expected outcome and verification process. Rewards would be calculated by based on predicted complexity, elapsed time, energy consumed and quality of answer. Presumably the best, fastest, cheapest answer wins. Validation techniques would be suggested ahead of time and true validation incented by the market. Of course energy consumption is minimized but not solved but at least there is a positive, possibly significant societal outcome. The value of the IP generated by the activity could exceed the block reward by magnitudes which should be of great interest to the market.

Example problems may be tumor sequencing or ontology creation. Both are well known computing problem sets with well known validation techniques. Most importantly both have tangible significant benefit to society and are achievable in a human timescale.

This Medium article is a call to action to collaborate on this idea by King and Nadal.