Ethics: Type-2 Metagaming

[Fiasco]

(I hope Mith doesn't mind me starting this. It came up in the other thread.)

Normally, players are assumed to maximize their probability of winning in the current game. This goal conflicts with other possible goals, such as punishing or rewarding players for certain types of behavior, or establishing a pattern of one's own behavior to refer to in future games. Pursuing these other goals at the expense of winning in the current game is called "metagaming".

An example of metagaming is the "Lynch All Liars" rule.

Assume someone was caught in a lie. Assume you're an innocent townie, and for whatever reason, you think the liar was probably also an innocent townie. Lynching him may cost the town the game. However, lynching him will also dissuade future townie lies.

What's the ethical thing to do here?

I'm going to take an unpopular opinion and say

Lynch All Liars is bad

, and

metagaming is bad

in general. You have a duty toward your fellow players to maximize your team's chances of winning in the game you are currently in. If you're interested in modifying people's behavior in future games, in-game mechanisms like lynching are not a legitimate way to do so. The only legitimate way to change people's behavior in future games is by convincing them your way is better. (Maybe we should make all new players read a list of the top ten dumb newbie mistakes, for example.) (Obviously, you'd still lynch most liars, because most liars are scum.)

I'm not sure, though. I just changed my mind on this. Thoughts?

[title edited to reflect that this is only one of two types of metagaming mentioned here]

[Fiasco]

Some good points. From the "maximize my team's chance of winning" point of view, the following are all legitimate reasons to lynch liars:

* Liars are more likely to be scum.
* Liars are more likely to be bad players (if they're protown).
* Liars are less likely to be trusted during the remainder of the game (if they're protown).
* Lynching liars will deter future protown lies in the same game.

I don't think those reasons are always strong enough, though, so I'd modify it to "Lynch Almost All Liars".

[Fiasco]

If in a non-LAL universe, it's good play for a townie to fakeclaim cop/doc, why is that something you'd want to discourage? If in a non-LAL universe, it's bad play for a townie to fakeclaim cop/doc, why not convince people of that by arguing? Is it really fair to put your teammates at a disadvantage because of your personal preferences about behavior in future games? I'm not convinced at all.

[Fiasco]

Most people here have misunderstood my position. I think lying is bad for the town in the vast majority of cases. I think most liars should be lynched. But to me, "Lynch All Liars" implies more than that; it implies you should lynch even liars that aren't scummy, otherwise it'd be "lynch most liars". As a townie, you should lynch exactly those liars whose lynch you expect to help the town more than the alternatives; no more, no less,

even if

lynching more liars than that prevents future bad play.

I'm not trying to give some general pro-liar argument. I'm just arguing metagaming isn't a legitimate reason to lynch liars, or to do anything else. This thread was intended to be about whether it's OK to metagame, not whether it's OK to lie.

WHY do you think he was an innocent townie?

By assumption, I have a good reason of some sort to believe he was an innocent townie. Maybe I have a 99%-trustworthy innocent cop result on him. I'm not saying the average liar is likely to be innocent; I'm saying that it's possible for liars to be probably innocent based on the information available, and I'm saying that in such cases, it's bad to lynch them.

LAL is a mindset to help you deal with this situation. In games, I assume all players will play to the best of their ability and that their ability is comparable to my own.

But that's not always true. You shouldn't assume things that aren't true. If you do, you're not playing optimally.

I disagree, obvoiusly. I think you CANNOT avoid meta gaming. Yes, it sucks when you get killed N1 every time, but it's cyclic.

It's simple to avoid metagaming in the sense that I just defined. Just do whatever is most likely to lead to the win, regardless of future consequences on bad play. Getting killed N1 sounds like (what I think the wiki defines as) pseudometagaming, i.e. using information from past games. I think that's perfectly acceptable. You're ethically allowed to use people's past behavior as a guide; you're just not ethically allowed to anticipate people using your past behavior as a guide and choose play that will be convenient for you in future games over good play.

You have a duty toward your fellow players to maximize your team's chances of winning in the game you are currently in.

Within reason. Clearly, you shouldn't cheat to do so. So what are your boundry conditions? Is using outside information legit?

You should play for the win within ethics. There is an ethical rule that says not to cheat. There may or may not be an ethical rule that says not to use outside information. There is IMHO no ethical rule that says you have to go out of your way to punish bad play using in-game methods.

Basically, mafia playing styles work in a Free Market system-bad styles don't emerge and good styles are recognized as good styles. Lying has never emerged as a good playing style for townies.

Lying is usually bad play, but that doesn't follow from your argument. If people lynch anyone who says "fishsticks", that doesn't mean the free market recognized saying "fishsticks" as intrinsically bad play.

Let's say I run a spider on these forums and figure out what words people use when they're scum and not when they're town.

I'd put that in the same category as computer-assisted chess. If it's agreed to be OK in advance, great. If it's agreed not to be OK, it's cheating.

Just a nitpick: "ethical", this word does not really exists.

Does too:

conforming to accepted standards of social or professional behavior

[Fiasco]

This thread was intended to be about whether it's OK to metagame, not whether it's OK to lie.

...so let me change my example.

You're the cop in a C9 game. D1 you lynch an innocent, so it'll be lynch or lose. N1 you investigate Bob, who turns up innocent. D2, before you claim, Bob claims cop with a guilty result on someone.

Stupid move, right? But do you lynch Bob, or not? If not, you're not a believer in LAL.

(edited to say: I agree with Seol's take on what the question is.)

[Fiasco]

There's a tradeoff between good play (that is, achieving a win for your current team (ignoring cults for the moment)) on the one hand, and achieving your personal goals on influencing people's behavior in future games on the other hand. (Possibly those goals are shared by your teammates, possibly not.) You're weighing the interest of one group of people against the interest of another group of people (including yourself). To me, that makes it an ethical issue.

My current opinion is still that metagaming is not OK; even extreme rudeness should not be a cause for lynching except to the extent that scum are more likely to be rude than townies. There are other ways to deal with rudeness that don't involve players using the mechanics of the game itself. (It's the mod's responsibility, mostly.)

Games like Mafia have rules (and specifically, win conditions) designed to set up certain conflicting goals among players. It's the mod's (or the game designer's) responsibility to choose rules and win conditions such that, if all players play as well as possible (in the sense of doing what gives them the best chance of winning within ethical constraints), the resulting game dynamics are fun. When you compromise good play in favor of other goals, you are unilaterally changing the game dynamics to something different from the game dynamics the players coordinated on by signing up for the game. If the mod did a good job, that means the game will be less fun, and if the mod did a bad job, that means the mod is relying on the players to make the game fun by making it a different game than it was designed to be.

That's my understanding of why it's ethical to play for the win when possible. I'm not completely sure it makes sense, though. Blah.

[Fiasco]

The only case in wich this decision would come up is if you are 100% sure (via copinvestigation or otherwise) that the lying person is in fact a townie.

Maybe the cop investigation is almost certainly accurate. If you're 90% sure the lying person is a townie (after taking the lying into account), you still get a worse expected score by lynching him than by not lynching him. There's still a tradeoff.

In most cases, a liar is more likely than average to be scum. There are some exceptions, though, and in those exceptions there's a tradeoff between lynching scum and lynching liars.

[Fiasco]

If you can be 100% sure, why not 90%?

Lying is partly a stupid move in itself, and partly a stupid move because people metagame against it. I accept that metagaming against lying may improve play quality on the site. But then at least admit that you're choosing to play worse to increase play quality in future games. I don't think it's ethical to use in-game means to coerce people to play a certain way, at least not until you've exhausted other means, such as writing a document explaining why lying is almost always bad. We're supposed to use votes/lynches/abilities to

actually play the game

, not to reward or punish each other for behaviors we (perhaps rightly) like or dislike.

[Fiasco]

Well, it might have been 99% or 99.9% before the lie, so the lie did let doubt grow a lot.

[Fiasco]

If you apply LAL, you're lynching even those rare liars that aren't more likely than average to be scum. If LAL just means lynching liars who look scummy (which

is

the vast majority), then it doesn't need a principle of its own, and it's not a metagame ploy.

[Fiasco]

What about a liar found innocent by a 99% reliable cop? (i.e. 99% of scum turn up guilty, 99% of townies turn up innocent)

[Fiasco]

So that means you agree LAL is bad? Because some people

don't

think you should maximize your chances in that particular game. It's not LAL if you don't LAL.

[Fiasco]

Well... I guess we could argue about this all day, but it turns out we agree about the underlying issue, which is that metagame ploys are bad. Right? I took LAL as a well-known example of a metagame ploy because that's what the wiki page says it's supposed to be, but it looks like a few people (at least you and Jeep) are now arguing that it's not.

[Fiasco]

That's not what I meant by "metagame ploy", but it turns out that also counts as metagaming according to the wiki, so I guess I should have been clearer. By "metagame ploy" I meant playing to affect future games instead of playing for the win. That's also what people mean when they say LAL is a metagame ploy.

[Fiasco]

Information can come with many different degrees of reliability. Would you really say lynching a liar confirmed by a 99%-reliable cop is good play, just because there's no absolute proof that he isn't scum?

You may be able to argue that situations like that are very rare in practice, but you can't argue it's theoretically impossible.