Proof of Excellence - Anyone have details?

Hi all, this isn’t a PPC post per se, but I wasn’t sure where else to post it. :wink:

I’ve been very curious for a while now after reading the Peercoin whitepaper, as to what “Proof-of-Excellence” is exactly. For anyone hasn’t read it yet (you really should!), here’s the quote and all the info we get:

"During our research we have also discovered a third possibility besides proof-of-work and proof-of-stake, which we termed proof-of-excellence. Under this system typically a tournament is held periodically to mint coins based on the performance of the tournament
participants, mimicking the prizes of real-life tournaments. Although this system tends to consume energy as well when artificial intelligence excels at the game involved, we still found the concept interesting even under such situation as it provides a somewhat
intelligent form of energy consumption. " - http://peercoin.net/peercoin-paper.pdf

Ok, see that sounds VERY cool, and I love all the things left to experiment with in cryptocoins that we haven’t even begun to try out. So anyway, we have another form of proof! I was rereading the paper today and I really want to know now. My question is, does anyone have anything more on this? Another paper or post by Sunny anywhere? More to the point, how precisely would holding a tournament between some gameplaying AI’s also contribute to integrity of the blockchain?

Rather than a hash, would it involve providing some particular game, like say a class P problem (P as in not NP etc) for which we have an efficient algorithm to judge success, and then get a population of AI’s without any knowledge of the algorithm to compete in solving it? Then reward the winner? And the problem somehow has the same effect as hashing? It could actually make a pretty cool artificial life experiment if you breed the winners using genetic algorithms, try to find the best general problem solvers or something but I don’t see how it would be cryptographically secure. I mean, you need a way to judge when you have a winner, which means you need the answer to the problem being solved before the winner, meaning it’s a P class problem. Also meaning an attacker could outrun the chain and cause havoc pretty easily by using the known best algorithm for the problems, and the whole tournament thing is redundant.

Or is this more that you’d use something like PoS, while the tournament is just held for deciding on what to mint and who to reward? Meaning you’d get the AI tournament to work any any problem you chose to feed to it, so long as you knew it had an end state, and so get some useful work out while releasing more coins.

So I’m obviously missing some finer details here about how this would work. I can’t find any other info anywhere though. Anyone know or care to hazard a guess?

http://www.peercointalk.org/index.php?topic=617.0

JustaBitofTime [17|Oct 11:01 pm]: I read in your Bitcoin Magazine interview about ‘Proof of
Excellence’ has any further thought went into that approach?

Sunny King [17|Oct 11:03 pm]:

not much because i don’t yet find a game suitable enough as a basis for the idea. most games are dominated by AI so there would be a lot of complaints of unfairness

Awesome thanks!

If proof-of-excellence relates to “human” activity, couldn’t be translations between languages a base for that game? AIs/Software programs are still performing very poorly on this field.

The problem would be ambiguity of results and how to determine the winner (with votations?) …

An idea could be: if a phrase is translated the same way from a mayority of participants in a fraction of time (e.g. 1 day), the first one gets the right to mine the block.

Or perhaps something based on Ontogame: https://en.wikipedia.org/wiki/Ontogame

That would even be games “with a purpose” and so the time would not be wasted.

Yeah, see that’s the issue, how do you choose a winner? Requiring human involvement could be iffy. Maybe the Ontogame idea could work.

I did find his actual interview with Bitcoin Magazine: http://www.peercointalk.org/index.php?topic=616.0. Still no clue on how it relates to securing the blockchain.

Maybe I’ll just PM him and ask if he has any more detailed info.