Oblique reference to Game Theory

[Akkori]

http://www.spectacle.org/995/scorp.html

FOr some unknown reason, the story of the scorpion and the frog popped into my head today, and I decided to look it up and read it to refresh my memory of the full parable. On the second click, I found the above page, and at the bottom of it found an unexpected reference to games. So I figured someone else might be as bored as I am and read it...

[pxib]

http://www.spectacle.org/995/scorp.html

FOr some unknown reason, the story of the scorpion and the frog popped into my head today, and I decided to look it up and read it to refresh my memory of the full parable. On the second click, I found the above page, and at the bottom of it found an unexpected reference to games. So I figured someone else might be as bored as I am and read it...

There's all sorts of assorted theories about games and then there's capitalizedGame Theory.

I could make a comment of the former type by noting that griefers sometimes share the scorpion's suicidal nature. The point of their exercise is power, and if they find themselves unable to achieve it through the intended corridors, they will unearth it elsewhere. It is easier to make your own team lose, for example, than to be the heroic captain of their victory. You are reviled, sure, and like the Scorpion's defection it holds no obvious victory... but you are reviled universally. Even the other team's victory feels hollow.

It is a rare hero who can achieve that reknown.

[eldaec]

Pxib hits the mark better than the link. In a typical mmog situation, the attraction of influencing outcomes, plus the 'I'll try button because I can' factor, together with limited or no consequences for losing, are the factors that drive scorpion behaviour. In 99% of cases, it is not 'in the scorpion's nature', instead it is a rational decision for someone seeking fame, or just wanting to see what happens if... there is a reward in knowing you influenced events, there is a reward in just doing something different for the hell of it, and in most cases there is no cost to losing, and very little opportunity cost in not winning.

[Tairnyn]

The Ultimatum Game is another Game Theory problem that has interesting results when applied to people rather than machines. The game is as follows:

Two players are given a joint monetary reward. One player chooses the split between them and the other player has two options: a) accept the split or b) deny the split and both parties get nothing.

From a machine perspective the rational choice as the offerer is to offer your 'opponent' the minimum amount in the split. The other player, being fully rational, should choose to take the unfair split because a little bit of payoff is better than none. However, when applied to people the experiment has some interesting results. For the most part human players, especially women, will offer fairer splits. In addition, depending on the value of money to the person, the accepting player will deny unfair splits to punish the offering player.

While it's a different kind of scenario it gives some weight to the inaccuracy Game Theory when applied to people. There's a thick layer of psychology and social norm that affects judgment in these situations. It would be interesting to perform this experiment with 2 groups, one in which the players know their opponent is in a room next to them and the other is told that their opponent is playing over the Internet from far away. I'd assume the Internet version would lead to more selfish choices, but even then it would probably deviate significantly from the purely rational behavior that most Game Theory is hinged upon.

[Abelian75]

It's not so much that Game Theory doesn't accurately apply to people, it's more that it's hard to really get people to put themselves in a "one shot" game scenario.  Put another way, you probably aren't modelling real world scenarios very well if you're only looking at games played a single time.

Where Game Theory starts to match up rather closely with real-world social behavior is when the game is played a random number of times.  Specifically, you say something like "after each round, there's a 90% chance that another round will be played."  Above a certain % threshold, you end up finding that strategies like tit-for-tat (if they screw you once, you screw them once), "grim" strategies (if they screw you once, you screw them forever), and other normal-sounding human strategies are in fact very good strategies.  And indeed, when you think about it, this is pretty much WHY screwing people over continuously isn't that great an idea... because it's likely you'll meet them again.

Strictly speaking, btw, Game Theory isn't being applied to machines at all normally.  Rather, the solutions it comes up with are entirely unapplied.  They're entirely true statements based on assumptions, like all of mathematics.  But just like all of mathematics, you have to make sure you're applying it to situations it actually models somewhat accurately.  In the case of a random human you present with two prisoner's dillemma choices, that person actually cares about more than the reward number they get out of the game, so the model is a poor one.

Of course, even in the cases where the game theoretic strategy is roughly equal to a normal human's strategy, the normal human STILL was probably not thinking solely in terms of their optimal reward over time, but that may suggest that perhaps our inclinations to not screw over our friends is a mental shortcut to the optimal strategy of the most commonly encountered style of game in a normal person's life.  That is, we're designed to play "games" in which the game may be repeated an uncertain number of times with other humans, not to play games only once with a person, and never again.  An oversimplification, of course, and most likely different people lean more toward different types of strategies.

[Akkori]

For internet games, it seems to mostly boil down to anonymity. It's not so compelling to be an asshat if the other player is sitting behind you, in a chair across the room, or even someone you see around town. But if they some guy living 2k miles away... well, it seems safe. But it is heartening to know that there are plenty of people who choose NOT to be asshat's even with the anonymity granted by the internet.

[Fordel]

Is it anonymity or lack of consequence and accountability? They are related, but not really the same thing. Maybe I am just nit picking semantics or something.

[Viin]

Is it anonymity or lack of consequence and accountability? They are related, but not really the same thing. Maybe I am just nit picking semantics or something.

There is lack of consequence and accountability because of anonymity. So yah. I think a lot of 'immature' MMO players/Internet Community Users (that come in all ages) go through a phase where they do whatever they can because they are anonymous just to see what happens and how it plays out. (There's not many environments where you can do that).

After awhile they grow out of it and actually form lasting bonds and usually some kind of 'honor'. For example, I would never lie about my sex (and rarely play a female because I don't want to mislead someone) and I would never intentionally screw someone over just because I can. I've never been a griefer anyways, but I grew up (started in MOOs/MUSHs/MUDs around 12 yo) in online communities so I quickly became attached to friends I met/knew online only and thus never really consider online interaction 'just a game'.

[Tairnyn]

The lack of consequence is also influenced to the size of the community. I often find that higher level players tend to be more considerate since they are likely to interact again with the smaller player base that is around their level, where new players/alts are exposed to a much larger player base in which they are unlikely to meet the same player repeatedly in the wild. As a corollary, niche games with a smaller community (e.g. A Tale in the Desert) tend to be 'nicer' than those like WoW where you're often one of thousands of players that can move to another server at any time to escape a bad reputation.

[ajax34i]

Smaller communities are "nicer" because they are easier.  It takes effort to get rid of your anonymity and be recognized by at least a few people (whether in a positive or negative way), and in most games you can only see and affect a small number of players at a time, much smaller than the size of the entire community.  Plus, the more time passes, the more people you've affected as word of your deeds spreads, so old-timers have an advantage over the new guy; the new guy has to do a much more shocking or pleasing thing in order to become as widely recognized in one day as the old timers have in the past year.

But, as far as anonymity in these games, a lot of people want to get rid of it somehow and be recognized.  It's just that the effort required to do so may be beyond the scope of the game, or too much.

[Lantyssa]

The admins can also be in much more direct control of the environment and easily look out for problematic behavior.

[Venkman]

But, as far as anonymity in these games, a lot of people want to get rid of it somehow and be recognized.  It's just that the effort required to do so may be beyond the scope of the game, or too much.

I'd think it's actually pretty easy to do. You just need to integrate with government agencies, or maybe credit bureaus. That's where the complexity could come in, because that's overhead people in the West just don't think is worth incurring.

I'd be in favor of a non-anonymous world, but I think the creativity of conversation would suffer for it. Imagine some of the hyperbole used around here coming back to bite ya on a job interview, like what's happened to some MySpace users :)

[Lantyssa]

I imagine most of us here aren't all that anonymous if anyone cared to do a bit of research.  Righ's Google-fu alone could provide blackmail material for years.

[Morat20]

I imagine most of us here aren't all that anonymous if anyone cared to do a bit of research.  Righ's Google-fu alone could provide blackmail material for years.

I've got shit -- under my real name, no less -- on Google Group's archives that would derail any Senate campaign. The fact that it was a friend of mine in high school who forged my headers as a practical joke doesn't change what you get when you google my full name and include newsgroups.

I've been trying to convince google to strip the two offending posts for a year now, with no luck.

[schild]

I absolutely love google for keeping everyone's sordid past available for all to see.

If they went around "cleansing" the internet gene pool as it were, we'd all be fucked.

Who's a liar? Who's a thief? Who's a serial killer? Who isn't?

Stop trying to convince Google to delete those posts. There are people much worse off than you.

Edit: The above post was dramatic without reason. Just sayin, erasing your past should come at a large cost.
Edit2: I wonder if you could sue google for caching it without your permission. What a crazy case of name defemation.