Sunday, March 11, 2018

Evolutionary Pressure

My favorite professor in college was Patrick Winston, who taught me basically "Intro to AI" and "How to Communicate".   One of the cooler things I remember him teaching us about was the subject of genetic algorithms, which in a nutshell is a way to solve a problem that works in the following way:

1.  Define a problem space and a measure of fitness for a solution.
2.  Randomly generate solutions to the problem with some parameters.
3.  Let the randomly generated solutions fight it out (solve the problem).
4.  Take the fittest ones, mutate them, and let them try again.  Etc, etc.

Basically the idea is to mimic the evolutionary process we see in nature in a computer simulation in order to come to some reasonable solution for a very complex (perhaps NP complete) problem.  The example that was coded up in the class involved little creatures made of blocks who had to figure out how to capture the piece of food in the middle of their little world.  I can't find it...but you get the idea.  You started off with a few simple block beings and eventually you'd get all sorts of crazy things fighting over the food pellet.

Poker ecosystems are no different.  People have seemingly random strategies, they test them out against each other, the ones with the best results get to play longer, the ones with the worst results usually go broke or play less.  People adapt, and mutate....but not really.

So I was thinking the other day about the poker games I've been playing in these past few years, and something occurred to me seemingly for the first time.  I've always known that "mega fish", your truly special game starters that everyone wants to play with, that even regular fish know are terrible, are extremely important to a poker ecosystem.  But something that had never occurred to me was the following reason for it related to long term (very long term) game health.

True mega-fish remove evolutionary pressure from the ecosystem.

And that's really the simplest way to say it.  If you have a game that features some truly terrible players there is far less pressure on everyone else to improve in order to survive.  Regular fish don't lose nearly as much when there is a mega fish in the game for a few reasons.  First, because of schooling....if someone is playing J7o for two bets it's far less of a mistake for the regular fish to play JTo behind him.  Also they don't get bum hunted as hard.   Mediocre players (break even players) stick around longer.  And perhaps most importantly or at least most interestingly winning players can leave large leaks in their games unaddressed for months or even years and still get along just fine. 

And it's kind of that last part that I was really thinking about when I was driving home from a game the other day and texting with Pete and The Hammer.  The games I usually play in just don't have any truly good players.  Nobody has even a modicum of game theory knowledge.  Everyone goes on tilt.  Almost everyone chases losses and quits amazing games when they've won "enough".   And the winners can just get away with this because the games are so often just balls nuts amazing that they can make large glaring mistakes (both in tactics and meta stuff) and still win enough to just keep going, month after month, year after year.

And I mean don't get me wrong, I'm definitely guilty of this stuff, too.  Spending the last 2 years playing basically no higher than 40/80 has definitely left me in a spot where I struggle to remain comfortable if I ever get into a reasonably tough game with 3 or 4 actually good players.  I honestly figured I'd be back to software by now, but I've frankly been winning so much for the last 3000 hours that I'm just under no pressure whatsoever to make a change.  And there you have it again :)

No comments: