Mafia: Mutiny on the SS Incorrigible

[Pleonast]

Sorry for the omnibus reply, but I think it’s better than several posts in a row.

@Chronos

Quoth sachertorte:The first iteration of this game, the Scum stood out like sore thumbs in the voting record. Granted, that’s just one game, and it may be that the Scum that game just weren’t very good at bussing (or decided they didn’t need to: They did after all end up winning). But I’d be loathe to accept the conclusion that multivoting helps Scum hide their intentions, without a lot more evidence the other way.

Scum stand out in this game too. The fraction of town-votes on scum vs scum-votes on scum was 10% vs 0%, 20% vs 25%, 70% vs 25%, and 42% vs 11%. I think more votes placed means a bigger record and makes it harder for scum to hide.

glowacks, Pleonast’s games tend to be amenable to strategies like that in theory, but the tricky part is that, in practice, it’s very difficult to get people to go along with even a very good plan. “Lynch everyone who doesn’t go along” isn’t very good when half the players aren’t going along, because they don’t understand the plan, or think it goes against the spirit of the game and prefer to do it the “old-fashioned” way, or think that their own personal plan is better, or suspect the person who came up with the plan and therefore think it must be a trap, while meanwhile the genuine Scum see that there are a bunch of Townies not going along, and so cooperate because it’s safe.

Unfortunately, our group of players have pretty much abandoned the principle of considering ideas on their own merits without becoming suspicious of the proponents. Apparently an effective scum tactic is to suggest a useful pro-town idea and then let townies block the implementation.

@TexCat

I didn’t really like the multi-votes – exactly because of the DrainBead situation. It seems like it tends to lynch everyone’s fourth choice. And it definitely muddles the vote record.

That sword swings both ways. It means scum that are low on everyone’s suspicion list is still vulnerable to the lynch. The effectiveness of the lynch is highly dependent on the quality of the votes. Perhaps this voting method magnifies that effect.

If players were not happy with who was being lynched, they could change their votes. If your fourth vote is headed to be lynched and you decide you don’t really want them to be, then change the vote.

~ ~ ~ ~ ~

Stan, you make good points, but I’d disagree with saying players are “forced” to vote for town. It may be difficult to vote only for scum, but it is possible. In my opinion, players spend too much time focusing on a single player and arguing for that lynch. Accountability for votes is solely up to the players–if players want to accept votes made “because the rules force me”, then those votes become acceptable.

@Stanislaus

Ultimately my biggest problem with the forced multi-vote is that it takes control of the game away from me. I want to decide how many cases I’m making. I want to decide how many votes I feel like putting out today. I want to decide what my threshold is for making a vote. But I didn’t get to make those decisions, and to be honest, that sucked a little.

Heh, now you know how I feel in games that restrict players to only one vote. Usually my games have optional multi-votes, which takes the best of both mechanisms.

@Stanislaus

Well yeah. That’s what happens if you force Town to vote for Town.

I take it to mean that the lynches were completely driven by town. The only way scum can stop a lynch that has half the town in favor of it is by coordinated voting. Which scum usually try to avoid for the obvious reason.

[bbonden]

I’ve followed the last several mafia games, but never played, so take this in that light.

I think that the multi-vote, forced vote was powerfully unbalancing in favor of scum.

In the platonic ideal, all information is town-favoring. Town gets information, analyzes it, and produces lynches. The more information, the better the lynches.

But there are three confoudning issues.

The first is the difference in vote systems. In a one-vote system, the voting record is not raw data - it is at least partially refined data. Every action (vote) has a consequence, and everyone can be held accountable based on those actions. With the multi-votes, not every vote has an expected consequence (it’s easy to know whether a bandwagon is building), meaning that the otherwise-refined data has become murky. The votes on the lynch leader and perhaps the nnumber two may be telling, but the rest is just noise.

The second confounding issue is the quantity of data. For scum, this is less of a problem, as scum doesn’t need to analyze the information perfectly; scum already has all of the information that it needs. Plus, to the extent that scum needs to analyze data, it has true and perfect collective wisdom on its side. For town, on the other hand, the hypothetical benefit of the additional benefit is outweighed by the burden of sifting through a bunch of meaningless data. Also, the need for everyone to make four separate cases means that there will be more weak cases made by town, which, I submit, hurts town.

Third, time is precious. People do not have an unlimited amount of time to analyze data. I have followed the last several games easily, including voting records. This game I could not follow at all. I am not sure how a participant could even begin to make heads or tails of all of the voting records in real time. Again, not a problem for scum - they don’t care (much) about voting records. Big problem for town, as the noise may discourage participation or, worse, may discourage detailed analysis because no analysis ever would be complete.

[Pleonast]

You’re probably right about the amount of information being overwhelming. Making more cases is highly beneficial to town, but only if they put the necessary effort into it.

I’m not sure how off votes are more noisy with multi-votes than with single votes. Typically there are only two (or sometimes three) players close to being lynched. Those are the only votes that need to be analyzed closely. That’s true, independent of how many votes each player has made.

I think in this game, the town in general did put less effort than usual into any of their votes. The Conspiracy games I’ve run, which are an order of magnitude more complex than this (more players, four factions, secret powers) had optional multi-votes. Town has managed to win all of them.

My inclination for this game is that it wasn’t forced multi-votes that was working against town, except for the workload effect. I think the poor use of the officer roles was the major factor in the town’s loss, which was partially due to the lack of participation.

One thing I haven’t seen discussed is the public reveal of the number of scum. That is important information in many games. Was it not useful here, or was it not used for another reason?

[sachertorte]

@Pleonast

The Conspiracy games I’ve run, which are an order of magnitude more complex than this (more players, four factions, secret powers) had optional multi-votes. Town has managed to win all of them.

I don’t recall multi-votes in Conspiracy. They may have had them, but I don’t think anyone used them to any great extent.

Also, luck and power role optimization played a big part in Town winning Conspiracy all those times. It certainly wasn’t due to smart lynching. Maybe lucky lynching, but not well-reasoned build-a-case or even vote record based.

[Stanislaus]

Making more good cases is highly beneficial to town. Making more bad cases is highly detrimental. There is an upper limit on how many good cases a given player is going to make in a Day. That’s partly workload - which is not an insignificant consideration - but it’s partly the nature of the game. If we were, as a group, collectively good enough that three out of four cases we make pick out Scum, we’d have long since given up Mafia as trivially boring. It’s because it’s difficult to find Scum that the game is worth playing. But that means being forced to make cases and place votes on our third-and fourth-choice suspects is bound to be self-defeating.

Town didn’t do itself any favours in this game. We could have done a lot better with more participation and better use of the officer roles. And better analysis. But being forced to make cases we didn’t actually believe in was intrinsically anti-Town, and I’d be reluctant to play in a game that had a similar mechanic again.

[Normal_Phase]

About town cooperation, I hate to say it, but we were all but stymied by our Captain. His absolute failure to participate in the game rendered all else moot until he was gone.

Second, it’s really hard to even think up good overall strategies when you’re scumhunting at the same time. Perhaps an initial period devoted to strategy alone (no votes) would have been an aid to the town. As it is, we went down the road early on some false assumptions and that left no time or energy for anyone who might have been so inclined to come up with something good. (I wouldn’t have – for me it’s not about mistrust so much as that’s not how I prefer to play the game.)

[Red_Skeezix]

@Pleonast

I think in this game, the town in general did put less effort than usual into any of their votes. The Conspiracy games I’ve run, which are an order of magnitude more complex than this (more players, four factions, secret powers) had optional multi-votes. Town has managed to win all of them.

That’s a tough comparison. The last Conspiracy game was won by a combination of one scum faction releasing the entire list of roles and numbers to the town and the other scum faction deciding at the end of the game that they’d rather let town win, than let me win. grumble grumble. I’m still frustrated with how that went down.

[glowacks]

I feel that because of the Town’s need to have players interested in compiling and analyzing the data they had available, participation levels needed to be emphasized at sign-up and that substitutes should have been considered. While the rules (aside from voting) overwhelmingly favored an interested Town, they were overwhelmingly stacked against a Town that had 3+ players not playing, one of which was in a position with a lot of inherent swing. It’s unfortunate that many players jump in and get sidelined by unexpected occurrences (especially with whatever happened to peeker), but we definitely need to be able to replace them. I’m absolutely opposed to a blanket ban on substitutions after seeing the state this game progressed to.

[Chronos]

Quoth Scuba:

Chronos: Scum won in Insipid partly because I didn’t – and still don’t – see the “standing out like a sore thumb” scummy votes. Both times I’ve been in the final three, I chose poorly and the scum voted with me. Please PM me some tips to spot those voting patterns.

I’m not sure I can help-- I’ve found, since then, that just as often the Scum don’t stand out like that, so my methods aren’t all that much better than chance. It may be that, in multi-voting systems in general, my methods work much better than in standard games, but I lack the data to say that.

Quoth Normal Phase:

Second, it’s really hard to even think up good overall strategies when you’re scumhunting at the same time. Perhaps an initial period devoted to strategy alone (no votes) would have been an aid to the town.

Isn’t Day 1 usually almost entirely strategy? The Scum-hunting per se can’t start until there’s some posts to hunt in.

[glowacks]

@Pleonast

Unfortunately, our group of players have pretty much abandoned the principle of considering ideas on their own merits without becoming suspicious of the proponents. Apparently an effective scum tactic is to suggest a useful pro-town idea and then let townies block the implementation.

This really saddens me. In face-to-face games a lot of what people are going by is unconscious emotional cues that people have no ability to control, while on forums all you have are walls of texts - but you have a perfect record of them. It’s so much easier to use all the information you have to your advantage when someone can’t claim they never actually said something (although sometimes you can try to say that it wasn’t what you meant), while at the same time you get to word whatever you say as exquisitely as you want and it becomes far less possible to catch people emotionally.

It seems pointlessly unreasonable to reject flat out any amount of reasoning because you don’t completely trust the source of it, and not particularly reasonable to fully accept the opinions of an someone who happens to be confirmed Town without further consideration. You obviously are going to trust the motivations behind the reasoning if someone is confirmed Town, but cold hard logic knows no motivation. With all the possible roles out there, you can work out exactly what someone might be getting at and recognize that anything specific they might say have ulterior motives. But when I put out my idea of using the brig as a back-up lynch (which it was nearly designed as) I was basically attacked because I was proposing to take a risk. A risk that I showed was extremely likely to lead to a higher Town win percentage, and I explicitly didn’t mention anyone (other than the person already in the brig) in order to show I wasn’t advocating for a specific brig, just a general principle. It was instead completely ignored and taken as a scum tell without even being analyzed. It was as if any detailed plan that involves taking a risk MUST come from scum, even if the risk can be analyzed and found to be utterly minimal and not possible to be manipulated by scum.

I understand the idea that if someone’s asking you to take a specific risk you need to be cognizant of the ways in which scum knowledge about alignments and available powers might warp things in ways that you are unsure of, but that’s not as much of a problem when something specific isn’t being asked and you’re in an open game. Notice how in Glasnost the Town was able to put their heads together and come up with a winning plan that forced a scum concession with 2 scum still not completely identified because they knew exactly how everything in the game worked. Then recall with Lost (which I read because it was referenced numerous times) that the Town lost explicitly because they thought the actual set-up was implausibly biased in their favor. Town needs to use the openness of the game to their advantage, or suffer tremendously. They may not have known exactly what our role distribution was, but they knew the kind of things we could do. Instead of planning for them, they just winged it.

[Red_Skeezix]

Glasnost is a bad example of town winning due to them following a winning strategy. Town caught huge breaks in the beginning of the game (removing the two greatest threats in two days). But recall it took 2 (or was it 3?) in game days for the same people to keep hammering home a strategy before people were willing to commit to it. I always say it, but in the end Logic plays an equal role in these games to psychology. Not only do you have to use logic to find scum and devise strategy, but you also have to be able to convince other players that you’re right and get them to follow you.

[glowacks]

@Chronos

I’m not sure I can help-- I’ve found, since then, that just as often the Scum don’t stand out like that, so my methods aren’t all that much better than chance. It may be that, in multi-voting systems in general, my methods work much better than in standard games, but I lack the data to say that.

I may have mentioned this before, but my statistical analysis of the voting data through Day 3 as would have been done by a typical Town member showed that those that had been confirmed as Town had a higher percentage as compared to the remaining unconfirmed of voting for other confirmed Town. Raw data was: 33/60 for Town, 29/64 for unconfirmed. Modeling these with a binomial distribution, I get the unconfirmed as being 1.5 standard deviations below Town which while only significant at around an 85% level, definitely is good cause to reject the hypothesis that scum are more likely to have voted for Town than Town to have voted for Town.

The only way the scum would have stood out is if you use the above as evidence that the scum actually would vote for Town less often than other Town, in which case you would have found me and Red at 4/12 and 3/12 as the “most likely” scum among the 6 unconfirmed. Stanislaus was 5/12, Tex 6/12 and fluid 8/12, with NAF at 3/4. I would have trotted out that statistic to nail fluid with if I had not already warned Stanislaus about using unsupported assumptions in analyzing vote data.

It is true that Red and Tex agreed on the most votes (8) among the unconfirmed, but Red also agreed with 7 of Idle’s votes, and Inner and Idle agreed 7 times as well. I also agreed with Stanislaus 7 times. Agreements on 6 votes were fluid-fubble, peeker-crackrat, Stan-fluid, Inner-Red. Tex-Idle, Tex-Inner, Tex-Crackrat. Overall the scum were more agreeable on average, but fluid was just as much as Crackrat and only a little below Tex and Red. Additionally while most Town had many fewer agreements overall, on average they were about the same as the unconfirmed as most confirmed Town at that point were dead.

[glowacks]

@Red_Skeezix

Glasnost is a bad example of town winning due to them following a winning strategy. Town caught huge breaks in the beginning of the game (removing the two greatest threats in two days). But recall it took 2 (or was it 3?) in game days for the same people to keep hammering home a strategy before people were willing to commit to it. I always say it, but in the end Logic plays an equal role in these games to psychology. Not only do you have to use logic to find scum and devise strategy, but you also have to be able to convince other players that you’re right and get them to follow you.

They didn’t have a forced win until then, and they were following a basic strategy that would lead them down the path, discussing what would in general be needed in order to help them into a forced win situation. They didn’t have their hands up and say “just do whatever you want as long as you justify it”, they put their heads together and worked on it. I agree that they won due to a lot of luck as well, but they still could have wandered around aimlessly and not really thought about what they were doing and what good strategy might be. They recognized that scum might be trying to influence them so they tried to make it fool-proof and a few days later managed to do so. They likely would have won without that strategy, but having the scum concede when there’s 2 of them left and the Town only has a vague pool of unconfirmed and no actual leads seems like quite a feat.