What's wrong with percentage?

[Ümläüt]

Since I plan to eventually mod a game, I've been spending a lot of time learning standard role terminology, including modifiers. One mechanic I was explicitly interested in using was investigative roles with a random element, because if I needed a "weak cop" or "weak tracker" or something, I didn't want to deprive the players of the opportunity to play every night by submitting actions. So imagine my surprise when I read the wiki entry on Percentage:

Percentage roles are extremely unpopular due to incorporating a large and uncontrollable element of luck to a game, potentially negating any skill the player with the role may have. In extreme cases percent-based roles may be seen as an excuse for the moderator to toy with the players. It is recommended in most cases that percentage roles not be used at all, as many players associate it with bastard modding.

Well, if it's unpopular it's unpopular, and maybe there's nothing I can do about that. But the justifications given don't seem very compelling to me.

• large and uncontrollable element of luck:
  Well, they add an element of luck, sure. Whether it's large depends on how it's done. I'd say it usually just replaces one kind of luck with another: it's bad luck for a mafioso to be targeted by a cop or tracker on the first night, but using a percentage cop or tracker can mitigate that, at the expense of exposing that role to the possibility of their own bad luck. As for uncontrollable, that leads into my next point.
  
• potentially negating any skill the player with the role may have:
  This is only true if you think it's impossible to reason with uncertain information -- but this is the whole point of Mafia! Any newbie can make himself useful as a cop, but a 50% or even 75% cop has to
  think
  about the investigation results and continue paying attention to the discussion to see if they make sense. This offers more opportunity for skilled players to outperform unskilled ones.
  
• may be seen as an excuse for the moderator to toy with the players:
  I assume this means the moderator might PM a player with (e.g.) "50% <Role>" but actually just report failure every time, or whenever it's funny. This isn't a problem with the percentage mechanic, it's a problem with bad moderating.

If this is just a matter of players having a learned association "percentage => bastard mod", rather than the mechanic itself being bad, that can probably be mitigated by restricting it to open games, or at least games where all roles are 100% self-aware.

What do you all think? Am I missing a major point here, or do I have some hope of running a game with percentages without pissing everyone off?

[BNL]

If you want to have a 50% Cop why not use an Odd-Night Cop instead? It's more predictable.

Moderator toying can be counteracted by having players to send a dice tag.

The element of luck is actually large. Even in mini-normals, you normally have only up to 6 days, and the cop will likely survive for much less, so even any variation from the expected outcome can be a ~50% difference which is arguably huge. Also cop targeting an innocent is actually helpful, so I won't count it against the swingyness of the cop itself.

If you are talking about cops which give an incorrect result 50% of the time, then that's really bad. But if you do tell them that their investigation failed

due to their luck factor explicitly

, then that's probably fine. Not knowing if your role actually worked or not is probably why people don't like such percentage roles.

[Ümläüt]

If you want to have a 50% Cop why not use an Odd-Night Cop instead? It's more predictable.

Well, I wasn't necessarily thinking of 50% specifically. But my main reason is that the most important thing is keeping players engaged with the game, and giving power roles something to do every night is a way to do that, as well as keeping the town in a bit more suspense. This may frustrate players when their role fails, but I'm surprised to hear that it goes beyond the expected "bah"-type frustration to an actual complaint.

I agree that percentage incorrect results are terrible, as opposed to percentage no-results. You might be right that the uncertainty about reason for failure factors into people not liking roles... but it's slightly surprising to me, then, that the norm for MS is that very little is revealed, even to power roles or their targets, about why certain things happen or don't happen. E.g. if the mafia's kill attempt fails, they don't know whether it's because the person commuted, because they were roleblocked, because the person was protected, or what.

(When I started playing ages ago on another forum, the norm was that the whole town got posts like "So-and-so was targeted by Mafia team A but rescued by the doctor" every morning. This is a pretty sharp contrast from what is made public here in most games, so it strikes me funny that acceptance of percentages would hinge on getting that sort of feedback.)

[hitogoroshi]

I don't necessarily agree with the reasons given, but I still think percentage roles are just bad design.

Thought experiment: in a setup with two scum left and no other town power, what's the difference between a 50% cop and a tracker?

[Kagami]

The most critical issue with a percentage role is balance.

Imagine you have a 50% cop. If the investigation misses every night, it's a VT (or worse, since the claim will be regarded with suspicion when it comes). If it hits every night, it's a full Cop, which is quite strong. You've introduced a swing component to the game that is altogether unnecessary, and is in no way influenced by the players' ability, which will potentially make the game significantly scumsided or townsided.

It also involves withholding information from the player which is relevant to their play. The mod could just as easily roll the dice pregame and tell the player he's a "N1, N3, N4, N6 Cop," or the like. A 50% cop who will never have a successful investigation but is playing a survival game believing she might is essentially being cheated.

The argument

for

percentage roles presumably leans on some notion that the unpredictability is fun, when it really isn't in most cases. We don't play 10 minute long games where we can laugh off a loss due to RNG; we play games that last for months, and it's a pretty terrible feeling after months of work to say "o, we played a great game and should have won, but that guy got really lucky, so instead we lose." When that happens, it cheapens the value of good play, while what we really need right now is setups that inspire players to perform better.

Also, minor note, tracking the correct player or being investigated as scum is not a matter of luck. The scumteam makes a deliberate choice of who they send out in an attempt to outguess any town trackers or roleblockers. Investigating scum is no more luck than lynching scum is. In both cases, a non-incriminating result has real value anyway.

[Kagami]

• potentially negating any skill the player with the role may have:
  This is only true if you think it's impossible to reason with uncertain information -- but this is the whole point of Mafia! Any newbie can make himself useful as a cop, but a 50% or even 75% cop has to
  think
  about the investigation results and continue paying attention to the discussion to see if they make sense. This offers more opportunity for skilled players to outperform unskilled ones.

I don't think it's the case that Town Cops just say "o, I'm a cop. I'm not going to pay attention to the game and let my role do all the work." In general, the cop needs to read the game to make good judgments about who to investigate. I could write an entire article about optimal cop-targeting. Any work the cop puts into making their decision will only be undermined by a percentage role. Losing a carefully thought out investigation can be demoralizing.

If you're really concerned about player engagement, you should create roles that inspire engagement in more reliable ways, imo.

[Iecerint]

Maybe you could introduce a role where a player can choose between a sure result of lower power or a risk result of higher power.

That way, the player makes a meaningful gambling decision, so they don't feel like they are at the mercy of the RNG.

[Davsto]

• large and uncontrollable element of luck:
  Well, they add an element of luck, sure. Whether it's large depends on how it's done. I'd say it usually just replaces one kind of luck with another: it's bad luck for a mafioso to be targeted by a cop or tracker on the first night, but using a percentage cop or tracker can mitigate that, at the expense of exposing that role to the possibility of their own bad luck. As for uncontrollable, that leads into my next point.

Key word: "uncontrollable". You cannot doubt there's some skill in a player choosing who to target. Do they go for the scummy player who'll likely be lynched anyway, just to be sure? Do they target the seemingly obvtown player to avoid them just skimming by? And, besides, even if you do inspect town, you still get a town result which is useful in its own right. With a 50% Cop, whether or not you get a result is entirely random; you could get 5 results in a row, you could get five misses in a row.

[Ümläüt]

I'm starting to think focusing on 50% really obscures the issue. Say we have, like, an 80% cop instead. Then the chance of failing three nights in a row, is only 0.8%. That's not impossible, but it's less likely than (say) the town lynching the only Mafia power role right out of the gate by choosing someone randomly. On the other hand, the choice of succeeding three nights in a row is about 51%, but this isn't a major imbalance if it happens since the role is balanced for a high chance of success already.

That said, Kagami, your point about how investigative roles actually play is well-received. If all percentage will do is make players angry or sad, it doesn't matter that much if it's actually 'fair' in some broader sense. I'm not going to argue the point about simply modifying roles with a percentage anymore. It doesn't sound like anyone will enjoy playing it.

I'm intrigued by your idea, Iecerint. Something like: choose between 100% chance of learning alignment, or

x

% chance of learning full role? (Where the value of

x

is something I'd want to choose so that the expected value of the two options is about the same.) Is that what you had in mind?

[Davsto]

But the whole problem with this is that

Imagine if you targeted someone you were a bit unsure of, with a 50% Cop. You were never sure if they were Mafia, so you didn't want to push to get them lynched, and no-one else really thought they were scum. You get no result due to RNG, and then are lynched the next day, or die the next night.

They were scum.

They survived to lylo and won the game for scum.

Purely due to the random number gods.

That is why it's a bad idea.

[quadz08]

For what it's worth, I highly recommend you pay attention to these people explaining why percentage roles are bad to you

they are all correct

[Cheery Dog]

I wouldn't even trust a 99% cop. Just give the scum a role blocker, and there's your percentage for you in a way that works.

[Antihero]

there are a lot better ways of gating roles than randomness. that's just sloppy.

[Firebringer]

there are a lot better ways of gating roles than randomness. that's just sloppy.

Naah randomness is the best method of balance in some situations.

[Firebringer]

I will go against popular opinion here and say why percentages is not a terrible idea.

1) Its a alternative method for modifications and neutering a over powered role.

2) Dice roll is an essential part of the game. Think about who gets roles in the beginning.
Aren't some players better than others?
That was done by dice roll. Some players are naturally better than others. If you think percentages makes the game too lopsided you should consider all player actions to be random. What happens when you give a cop to someone who uses Random.org to pick targets. Is that much any different than giving him a percentage of success or failure? I mean in one case you still get a result not dependent on coin flip, but it could easily get same results.

I would say that if you are going to give a role a dice roll. You should let the player know if the dice was a success or not.

[Firebringer]

Now I await for someone to take my bait.

[Antihero]

it's not AWFUL like the godfather role, it's just... inelegant. clumsy.

[Ythan]

50% roles are p ugly, any other percent above 0 and below 100 is even uglier.

[Antihero]

One mechanic I was explicitly interested in using was investigative roles with a random element, because if I needed a "weak cop" or "weak tracker" or something, I didn't want to deprive the players of the opportunity to play every night by submitting actions.

haha

one thing you'll quickly learn about this site's culture is that you don't necessarily need an action for everyone every night to keep people engaged.

[BBmolla]

Let's say you're playing faith plus one

http://wiki.mafiascum.net/index.php?tit ... h_Plus_One

Let's say the game is at 3 VTs, 1 Faith Doctor (50% chance of success), and 2 Mafia.

You guys lynch a VT.

The Faith Doctor than correctly predicts the nightkill.

The game is now literally over or continuing due to luck.

That's garbage.

[BBmolla]

Literally all your arguments are invalid unless you can prove to me how the hell the above is fair.

[BROseidon]

Maybe you could introduce a role where a player can choose between a sure result of lower power or a risk result of higher power.

That way, the player makes a meaningful gambling decision, so they don't feel like they are at the mercy of the RNG.

This feels like it would likely end up being a false decision :/

[BROseidon]

Also there really isn't a reason to add anything else to this thread because randomness is so obviously terrible that we shouldn't even need this thread.

[Antihero]

clarification:

i've never seen a "percentage" role myself, so i don't know how they work.

if it's where the action just flat out fails and gives NO RESULT (100-X)% of the time, that's really blah and there's better ways to do it.
if it's where the action gives an INACCURATE OR FALSE RESULT (100-X)% of the time, that's terrible and should never be used. the only way to correctly play that role is to act like a vt and ignore the action altogether.

[BBmolla]

I thought it was 50% succeed, 50% fail meaning "no result."

If you guys are arguing they should get a wrong result 50% of the time

Lmao