Newbie games setup

[Talitha]

(I know that Norinel wanted to wait until his newbie game was over before discussing this, so apologies to him. I just wanted to get the idea down while it was in my head.)

There has been some discussion of concerns that the current set up of newbie games favours the mafia. (Results so far: Town 5 wins, Mafia 9 wins).

Firstly, I guess we need to decide if it's actually a problem that the mafia seem to be at a slight advantage. I agree it would be nice if both sides had an even chance of winning, but how important is it?

The current idea to address the imbalance was to give the cop a night 1 investigation but leave the rest of the setup the same. (2 mafia, 1 cop, 1 doc, 3 townies, game starts with a day).
I wonder if that will cause more confusion to newbies (who are already having to grapple with learning about Day vs Night and everything else), considering
- the fact that most games start with a night while newbie games start with a day,
- the fact that someone gets a Night 1 choice while the others have to wait until Night 2
Would you start the game with a Night or still start with a Day? Would you let the mafia communicate before Day 1 started?
To me it gets a bit complicated.

My solution would be to add another player to the game, have 4 townies instead of 3, and start the game with a night. This would make the game more similar to other games that are played on the site and a therefore a better introduction to mafia. It would also solve the awkwardness of starting with a day - with no information whatsoever - and only a made-up kill to inspire the town to action. It would give the mafia a chance to collaborate and get the feel of their role, etc.

There are a couple of downsides that I can see. Firstly there will be newbies who are killed Night 1 and that seems a tad unfair. (Although I'm sure someone would send such newbies a nice PM and encourage them to sign up again). Second (not sure if this is actually a problem), there'd have to be a different ratio of new and experienced players, 4:4? 3:5?

Are there any other problems with this idea that I haven't thought of?

Is this idea an improvement on the current set up?

[Yaw]

I decided to do my own count, because there was a statistic beyond winning in the completed games I thought important. For the 14 games completed to date:

Mafia wins, successive bad lynches: 6
Mafia wins, three-person endgame: 3
Townie wins, successive maf lynches: 2
Townie wins, three-person endgame: 3

So the games in which the town had a good chance of pulling out the win are ahead 8-6. Not only that, most of the mafia quick wins happened in the early newbie games. They don't seem to happen as much anymore.

I'd tend to think the games aren't so much broken, as they have their own strategies that took a while to work out. Now that they're established, they seem to be becoming more balanced.

All this means I don't think there's enough information yet to conclude that the newbie games are "broken", or need fixing. I think what little evidence we have points to them fixing themselves.

[MeMe]

Um...I actually prefer to just leave them as is and here's why:

Newbie games were created to cut down on newbie disappearances by getting them into their first game quickly. Upping the number of players will make it take just *that* much longer to fill...and having the game start with a night will add another couple of days to the wait for actual game play to begin. The current setup isn't

perfect

-- but the town

has

been able to win more than 1/3 of the time, so they're not terribly unbalanced...and these are more for the purpose of teaching the basics and hooking players than creating a mafia utopia.

In short, I think the current set-up has been serving its purpose excellently.

[MeMe]

Oh, hey...took a little longer to type my thoughts than I realized and Yaw hit the highlights ahead of me -- with stats, no less!

[modargo]

The game immediately becomes degenerate if you give the cop a pre-game investigation in a seven player game with a doctor. Seriously. Go through the scenarios. The analysis is that the cop and doc coming out day one is always the right play, and one of the mafia has to fake cop just to have a 33% chance of winning.

Adding an eighth player and having the mafia kill day one also doesn't do much -- it just makes the game very random. If the mafia don't kill the cop or doc, you're right back at degeneracy. Games shouldn't be almost entirely decided by a few night choices before the game even starts.

If you want a really balanced format, here's one for you. Nine players. Cop is given a random pre-game investigation of a non-mafia. Game starts in day. The roles of players who are lynched are revealed. The roles of players who are killed at night are not revealed.

This setup allows more room for error than the current setup (the town can lynch wrong twice instead of once), and it allows the delightful uncertainty of wondering if the cop is real or if it's a mafia impersonator. This is much more strategic than the current format, where the mafia can never fake cop after the real cop has died. With this suggested format, perceptive mafia can pick up on who the cop is and gain quite an advantage from it. I also think this will help give newbies experience in when to disbelieve people's roleclaims.

The pre-game sight should be random and non-mafia for two reasons. First, to prevent the same people from being seen every single game. If you ran the format above with the cop choosing who to see before the game, mathcam and MeMe would never survive as mafia. Second, the game is a bit too random (and too easy for the town) if the cop can just get lucky and find a mafia before the game.

This variation is very close to balanced. I think the town has a slight advantage if it is played correctly. However, that advantage is easily neutralized by a good mafia or by inexperienced townspeople.

It seems to me that nine poeple instead of seven isn't that much more. From what I've seen, the newbie games fill up quickly enough that it wouldn't be a problem. Instead of 19 over three months (the current rate, give or take), there would be fifteen. And again, I'll say that I think it's important to make the games a bit longer and more forgiving. Three days maximum, and only one possible wrong lynch is just too few.

[PolarBoy]

It sounds nice modargo, but something I don't like about your setup is that it has a lot of specialized rules. That means that when a player gets done with his newbie game, he'll jump into a mini-game or a regular game and be told to throw a bunch of rules out the window. That's raising the barrier to entry somewhat.

I'd like to here reiterate MeMe's point that the idea behind newbie games is not to have a perfectly balanced gameplay experience, but to familiarize new players with the basic principles, rules, and conventions that are used in mafia here.

[God]

I love it modargo, your 9 player newbie game has my vote

[Yoko Kurama]

7 gets the job done the games are only used as a teaching device if some one wants to make a 9 player game they can do a mini

[mathcam]

I agree with MeMe. The current set-up is working well, regardless of possible imbalance. It's function as a learning tool is great, even if not a good exercise in game design.

Cam

[Norinel]

Newbie 16 is over, so here are my thoughts.

I don't think the problem is that the mafia win too much. I think the town has just been playing poorly and not using an optimal strategy. (The chances of a town win given random lynch in the two mafia five townie scenario is 8/35, which isn't all that much worse than they've been doing so far)

As Antrax pointed out over here, and we figured out in Newbie 16, (Even though the mafia won- it was one of the worse cases going into Day 2) the openness of the newbie setup, and there being as many protectors as there are people who need protection, means that it's possible to have a breaking strategy. (Probably the worst potential breaking strategy is complete mass claim Day 1, and even that does better than the town has in the completed newbie games assuming random lynches) IMO, there's a difference between a game that's imbalanced because it has a trivial strategy that significantly racks up the chances of winning for whomever performs it and a game that's imbalanced because one side statistically wins more than the other. As a learning tool, I'd find the second more appropriate. (Think about if in every single game down the line the town were to do a total claim or a cop/not cop claim or townie/not townie claim Day 1 without arguing since everyone knows it's better for the town) In other words, if we don't want to increase to nine players, I think it's the town that needs weakening. (Or better yet, something that gives the town equal chances without being easy to break, but I don't see any way of doing this without going to nine.)

So my idea is to change one thing: have four townies and only one powered pro-town role- either a cop or a doc by random secret decision. It makes the game more like 90% of the other games on the site by adding a little mystery to the setup, it weakens roleclaims by ensuring that anyone who truthfully claims non-townie will probably be dead by morning, and it (As far as I can tell) puts any sort of Day 1 mass claim completely out of the picture. I haven't run the appropriate numbers yet, but it seems like it would favor the mafia more than the current setup.

On the subject of statistics, don't some of the super-old hands have some simulation type programs? If so, I'd be interested in seeing what the odds are like for the current setup given some sort of (rather complicated) random bandwagon to claim strategy. Too tired/lazy to write one myself.

[Talitha]

Hmmm.. thanks Norinel (and everyone else for the input too).
I'll have to chew over this one for a little while before I can decide what I think about it. (Not that my opinion is particularly important here ).

[Yoko Kurama]

Hehe.... I didn't want to say it up front but Newbie 16 was my "pointing out a flaw." I talked to dirge about it before hand and he told me not to bother with it but I couldn't resist a sure fire way to win!!!! Then I lost ...... shouldn't let people tick me off I guess ehh.....

I agree with your new setup and think that it is just the thing the game needs. My problem though is will the mafia be way too powerful?

[mathcam]

Here's the deal with the newbie game setup.

If we assume that it's optimal strategy for the cop to come out on day 1, which I suspect (like others) that it may be, then it's also optimal play for one of the mafia to also claim cop.

If the town lynches the actual cop, then the mafia has a 59.4% chance of winning (assuming I didn't do some arithmetic wrong), so the mafia is actually the favorite in this scenario. If the town lynches the mafia instead, the town has >70% chance of winning, but I don't think it's

much

bigger, though I didn't work out the exact number.

So in this scenario, the town starts the game with about a probability of winning about 57%. This doesn't seem unbalanced to me.

And

this is all assuming that the cop comes out on day 1. Maybe they can do better if they don't do this.

This is also assuming that we are all robots playing mafia, and are totally unable to determine anything from intuition or deductive reasoning. So I suspect that a strong gut belief one way or another should take precedence over the cop coming out as optimal strategy.

But I do like Talitha's proposal above.

Just my two cents.

Cam

[Dragon Phoenix]

The simplest way to cure this is simply forbid role claims on day 1.

[PolarBoy]

What constitutes a role claim?

[ages]

I think forbidding role claims in a Newbie game is a bad idea because:

It isn't a normal game mechanic, so not a good teaching aid.

How do you classify a role claim? Is hinting alowed?

[Norinel]

I agree that banning role claims is not the best way to fix things.

mathcam, a 57% chance of winning for the town is pretty ridiculous, when you factor in the town's ability to find scum. (Although the mafia's ability to find the doc shouldn't be underestimated either) According to one person's statistical analysis over at http://eblong.com/zarf/werewolf-stats.html (Although he's playing IRL, with a cop he doesn't factor into the calculations), he prefers setups where the town has a 23-29% chance of winning.

[mathcam]

My personal philosophy for games is that it's meaningless to say something like "25% winning chances for mafia makes the game most fair." What does that mean?

If you truly want to design a fair game, you make the design so that it maximizes fun, and

then

assign strengths to the side of winning groups so that if you played long term, it would balance out.

For example, if the mafia were only expected to win 25% of the time, I would award them 3 points for winning, and the town 1 point for winning. This is a fair game, because the expected winnings for both teams are equal.

Of course, we don't play for "points" here on mafiascum, we play for the thrill of victory, but this is even better, as it is self-balancing. You should get more of a thrill of victory if you had a 1-in-10 shot of winning then a 1-in-2.

But if you want to eliminate this altogether, you just make it 50-50, and 57%-43% is pretty close to this. The town's ability to find scum is, in my opinion, completely outweighed by the mafia's ability to misdirect if they choose, or the town's ability to lynch incorrectly on their own.

Cam

[Norinel]

"25% winning chances for mafia makes the game most fair." What does that mean?

I'd assume something like that whatever chance that is given random lynches works out in practice to where both sides have a reasonable chance of winning.

But if you want to eliminate this altogether, you just make it 50-50, and 57%-43% is pretty close to this. The town's ability to find scum is, in my opinion, completely outweighed by the mafia's ability to misdirect if they choose, or the town's ability to lynch incorrectly on their own.

I disagree, but we'd have to try things out empirically- have 10-20 newbie games where the town goes for a cop claim Day 1 and see how many of those the town wins. (Although the cop claim hasn't happened since Newbie 16, IIRC; I'm not sure why.)

And even not looking at the math, it's just a matter where eventually, enough people will realize that there's no reason to do anything else. That decreases the fun factor, especially because any sort of mass claim reduces the possible cases. (And that most people want to discourage that sort of thing in larger games)

And from an older post:

But I do like Talitha's proposal above.

Which one? Modargo alredy pointed out that an extra cop investigation broke the game even more for a Day 1 mass claim, and having eight players and a Night 1 just bases most of the game on random choices/metagaming.

In case I haven't said it yet, I would be fine with setup experimentation, as long whatever's being done is made very clear before the game.

[Dasquian]

Is there any mileage in just making the newbie game setup 12 player? 2 mafia + 1 mafia GF, 7 townies, 1 cop, 1 doc, or something?

[Flying Dutchman]

I think the set-up just needs to be balanced a bit, but adding a vig or something seems to do the trick.

(I don't know if you should use the vig instead of a townie, or extra, but I think he will make a huge difference.)

[Dasquian]

I think a vig in a 7-player game is a really bad idea - it means a possible loss for the town at the end of night 1.

[Phoebus]

Is there any mileage in just making the newbie game setup 12 player? 2 mafia + 1 mafia GF, 7 townies, 1 cop, 1 doc, or something?

Consider newbie registration frequency. We can fill in ~2 current newbie games if we were to assume 7 of those 12 were to be new.

[mathcam]

"25% winning chances for mafia makes the game most fair." What does that mean?

I'd assume something like that whatever chance that is given random lynches works out in practice to where both sides have a reasonable chance of winning.

But

most

fair? There are plenty of distributions of winning chances (i.e. 25-75, 30-70, 50-50, etc.) that would give sides a reasonable chance of winning. If we want to truly make it a test of pure skill, i.e. which sides actually is playing the game better, we give both sides a roughly 50% shot at it.

I disagree, but we'd have to try things out empirically- have 10-20 newbie games where the town goes for a cop claim Day 1 and see how many of those the town wins. (Although the cop claim hasn't happened since Newbie 16, IIRC; I'm not sure why.)

I suppose, but I strongly feel it will be between 60-40 and 50-50 in favor of the town. The cop did claim in 21, the mafia counter-claimed cop, and the actual cop was lynched. From then on, it's pretty hard for the town to win.

And even not looking at the math, it's just a matter where eventually, enough people will realize that there's no reason to do anything else. That decreases the fun factor, especially because any sort of mass claim reduces the possible cases. (And that most people want to discourage that sort of thing in larger games)

While I agree that maximization of fun is a top priority, your above paragraph only holds if it is indeed an optimal strategy for the cop to come out, of which I'm not convinced.

But I do like Talitha's proposal above.

I think was in reference to the idea of having a cop

or

a doc, not both. Sorry if this wasn't Talitha's proposal.

Cam

[shadyforce]

I don't agree with increasing the size. Part of the appeal of those games is they are small and simple and thus get newbies into the game easier than a mini-size or larger game would.