Gaming The Meta-Game

I have mixed feel­ings about meta-​gaming.

Some­times, it adds a lot to a game. Cer­tain games would be down­right bor­ing if you strictly played by the rules. Poker comes to mind with the idea of “play the player, not the cards.” Other times, though, it can break a game. Mafia comes to mind here — I’m sure every­one who has played that game has anec­dotes of out­side fac­tors unin­ten­tion­ally reveal­ing things and spoil­ing a round. I guess it comes down to the pac­ing and atmosphere.

But from a com­pet­i­tive or design per­spec­tive, meta-​gaming is nec­es­sary to fuel inter­est and pro­gres­sion. How do you design a game (and match­mak­ing for the game) to make it easy to pick up but impos­si­ble to set down? What makes a com­pet­i­tive game “fun”?

Sirlin’s (admit­ted “hard­core”) take on the issue is to cre­ate a hyper-​accurate match­mak­ing sys­tem where you are matched as closely as pos­si­ble to your skill-​peers and end up with a roughly 50 – 50 win-​loss record. He believes that the only sta­tis­tic that mat­ters is win-​less record, and any other stats can be gamed and thus dis­tort skill rat­ings, match­mak­ing results, and game out­come pre­dic­tions. His ideal sys­tem would be sim­i­lar to TrueSkill, which is based on Glicko, which in turn is based on Elo, the clas­sic chess rat­ing system.

An alter­na­tive sys­tem was cre­ated by Jeff Sagarin, who uses win mar­gins to bet­ter pre­dict future out­comes than the Elo sys­tem (and he applies it to col­lege and pro­fes­sional sports quite suc­cess­fully). In fact, there is cur­rently a com­pe­ti­tion to cre­ate the best chess ranking/​matchmaking/​predictive sys­tem, and Elo is not doing very well. (The dif­fer­ent teams’ method­olo­gies are not released, so I couldn’t tell you if the cur­rent win­ners are using win-​loss or points systems.)

Of course, like pointed out above, all stats can be gamed. Advan­taged play­ers can pur­posely hand­i­cap them­selves. Dis­ad­van­taged play­ers can try to force draws instead of risk­ing losses. I remem­ber when Son D would pur­posely sur­ren­der after absolutely crush­ing an oppo­nent to pur­posely keep a 0-​win record. (Even if I still do not under­stand his rea­son­ing for doing so.)

Tak­ing Val­i­da­tion The­ory into con­sid­er­a­tion, a “slip­pery slope” game will teach you the most (assum­ing accu­rate post-​mortem analy­sis) and give you the fastest skill pro­gres­sion. How­ever, slip­pery slope games are intim­i­dat­ing to new and more casual play­ers. I think the idea is to cre­ate a per­pet­ual come­back game where in-​game stats are con­sid­ered and used to give a slight boost to play­ers who are behind. How­ever, a strictly accu­rate match­mak­ing should be used in between games. This way, play­ers never feel that they are get­ting com­pletely crushed, and they can incre­men­tally grow by learn­ing from peo­ple who are close to them in skill level.

Any­way, end of geeky gam­ing ram­ble. Here’s a cool video for you:

Tags: Games, Son D