Mini Normal 2003: donezo

[mastina]

i feel like roles are being valued based on the average of players that tend to play them versus the role itself.

I mean.

Yes?

You balance roles off of their expected usage--not their theoretical usage.

Theoretically, a miller is a negative utility role because it acts as a false guilty for a cop. That is the theoretical usage of a miller, to weaken the strength of a cop.

In practice, nobody doesn't claim the miller; it is a policy-claim role. So instead of going off of theory, you have to balance...

...Off of the average of players that tend to play the role. And site meta is to instaclaim miller.

Do you disagree with that?

[mastina]

not sure what the purpose of the ninja was exactly.

Serves as a guilty to the rolecop.

The only irk with the setup fmpov ig is the vig becomes conftown once they claim the LyLo-removing kill which gives potential for FIVE conf alignments in 2 nights

How do you figure? Vig is conftown, weak neighborizer gives two extra conftown, but rolecop isn't conftown, rolecop's results aren't conftown, and the weak neighborizer themselves isn't conftown, and the only way the vig's target is conftown is if they're a dead town body. By my math, that's three.

You can also have a lot of ugly situations, such as the neighborizer's target overlapping with the vig or being the vig, and the vig's target overlapping with the neighborizer or being the neighborizer. Also, with poor 'crumbing, the weak neighborizer dies early and if the vig shoots early as well...game gets to be in a bad spot fairly fast for town.

[mastina]

tbh I thought it was odd too. The original setup was much better for scum.

It was also scumsided.

[mastina]

For the record, my final thought RE: balance on the setup was:

Swingy as fuck, but I think the math loosely checks out to balanced, yes.

I stand by that assessment--swingy, swingy as can be, swingy as fuck. It could swing either way. But the odds of it swinging one way were loosely equal to it swinging the other way. The odds of town getting fucked loosely equaled the odds of scum getting fucked and the most likely outcome was somewhere in the middle where both sides got setbacks due to misplays.

It's not the job of a reviewer to dissuade swinginess, not under the old reviewing system this game used and not even in the new reviewing system just implemented. It's the job of the reviewer to first check normalcy and then check balance, and while this game was WAY more swingy than I'd personally prefer, the math I'd say

loosely

works out to balanced.

The guideline we as the NRG have is "a side is allowed

up to

60% chance of winning", more or less. This game might not be exactly 50-50, but I maintain we did not exceed the 60% threshold, and that the originally proposed setup did exceed said threshold in favor of scum.

[mastina]

you can kill the whole scum team by N1. only a possibility but still.

You can also kill the whole town by the end of D2. Only a possibility, but still.

Mislynch D1, weak hits scum, vig hits town, scum nightkill different town. That's half the town dead (4/8 of a needed five) by D2, and placing the game in mylo--where with a mislynch, town loses. Evens favor scum, too.

You can't deny that's strongly in favor of scum if you got that environment.

[mastina]

i disagree with 8v2 being scum sided. less scum equals less wiggle room. ESPECIALLY WHEN CONSIDERING TOWN PRs. 8v2 being "scum sided" is because town like 70 percent of the time is incompetent.

No it's because of setup math.

11-2 mountainous is considered balanced.

Not 11-3.
11-2.

You had 8-2.

That is three less townies than is balance for mountainous.
That is also on evens rather than odds.

Combine both factors (game balance being comparatively scumsided to what mountainous numbers would be, game on evens), and what you get is that scum have a lot of power in here even without a roleblocker who can fuck the weak neighborizer/vig over.

Nightkill + roleblock = two shots per night at shutting down town rather than just one. In a smaller game, that is IMMENSELY powerful. With a mislynch D1. Knowing who the scum are. You can target 2/7 town players. With 3/7 of them as PRs and two of them as important ones, that's a 28.5% chance of shutting AT LEAST one important PR down. (This is probably bad math, but it gives you the general idea.)

There's also how the town PRs have strong negative utility, capable of potentially producing two town bodies in addition to the scum nightkill.

Scum in smaller games have strong advantages. So a town needs strong advantages to counter the scum. And scum don't need an answer to every town role; if anything, it's the other way around with town needing an answer to what the scum had.

[mastina]

I love mastina ranting..

I aim to please. <3

[mastina]

but balancing roles due to expected bad usage is not good. There's only room for improvement in that scenario.

Balancing roles is done for ALL expected usage--bad AND good.

The thing is, the bad

cancels out

the good, loosely. That's what the swing is. Swing is when there's something which can be really bad or really good for a faction depending on the outcome of a role. The really bad is loosely equal to the really good.

In this case, the really good would in fact be a town in on N1 at the most extreme, but that was loosely canceled out by the really bad which would be a scum win at the end of D2.

We don't so much balance off of expected bad usage, so much as we do off of expected average usage: neither the most extremely optimal usage nor the extremely suboptimal usage.

And by that metric, this game was in the balanced zone because the town's roles with average use of their power gave the town a boost--enough of a boost to bump the game from what would be scumsided otherwise into being balanced. Was it perfect, probably not, no. Maybe it was a little townsided, maybe even like 55% townsided. But it was

close

.

Just swingy.

I'm not really fond of swingy setups for exactly this reason; players in postgame complain about how bad things could have been. But again, if the mods want the setup, so long as it is both normal and balanced and they are okay with the swing, then they can run the game.

In reviews, this manifests pretty much in the way I did it--I'll note that the setup in question is swingy as fuck, but balanced. If the mod is happy with that, then we proceed. If the mod isn't happy with that assessment, isn't happy with it being swingy as fuck, we work on reducing the swing and tweaking the setup. (Mod mods opt for the former if for no other reason impatience in that they don't like to have the review drag out. )

Mods have a fair amount of power in reviews. Especially with the new system being designed to hasten the review process.

[mastina]

So swing is legitimized as long as swinging either way is possible?

There's no such thing as swing which

doesn't

swing both ways--a setup which swings one way isn't swingy; it's unbalanced in favor of that way.

Also, to some extent, swing is inevitable. There is no such thing as zero swing. There's people who have written more advanced theory articles covering the subject than I ever could give (in that they explain the concept there better than I can), but basically, no matter what you do, your game's going to have swing, it's just a matter of how much.

And "how much swing is too much swing" is...something that's pretty subjective and thus, very ugly for reviewers to deal with. We might be uncomfortable with it, but we also are obligated to not hold up review on personal bias. So instead of focusing on the amount of swing, we focus on about how "centered" the game is. (Games can be both swingy and yet also scumsided/townsided in that they have the potential to lean the other way slightly but by default heavily lean towards one side; this, we want to avoid.)

If the game's overall position is such that, as close as can be approximated, each side stands about equal chances, loosely, of winning...then the game is balanced, even if it is swingy.

if balance is based on how players on an average basis us roles, and this average basis is "poorly", then town more times than not will have the advantage based strictly on roles regardless of how they are used.

And this is, when properly applied, not a bad thing. The vanilla game of mafia at its basis, in mountainous, is ridiculously scumsided. Even theoretical EVs tend to pragmatically come out to scum winning when by all rights they shouldn't.

Thus, the need for town powers--and town power roles are, at their core, meant to enhance, to augment, existing townplay. They NEED the advantage of those roles, because without the advantage those roles offer them, they hold no practical chance at winning the game.

It's a fine art to strike the right balance, especially since it's not an exact science. The 60% rule is because getting things on the mark 50% is almost impossible and there's two separate metrics by which to gauge balance (off of the play of players in theory and off of the play of players on average; the former leads to more "scumsided" setups and the latter more "townsided" setups), so the 60% mark is "close enough" to balanced.

To balance, you need to basically run mental math of all the roles and their interactions and have a good head for numbers, and then you need to determine if the combinations work out such that on average, the setup created has no strong inherent bias to an alignment. This is one reason I like simpler setups. Simple setups tend to be easy to just take a cursory glance at and instantly go "oh yeah that's balanced, we can do that".

But the more you introduce chaos to your roles. The stronger, more influential you make them. The harder it is to lock the game down. And Esp/Nero from the onset wanted the 2x Weak Neighborizer as a role. That's about as swingy a role as you can include in a game of this size, so building a setup around it involved giving the town enough power where they don't need to rely on a single role yet also making it so that scum wouldn't be overwhelmed.

[mastina]

youre talking about bad play from a vig

No, I'm talking about numbers for a vig.

With a mislynch on D1, the vig has 8 targets to select from.
2 are scum, 6 are not.

Statistically speaking, the vig is more likely to hit town than scum.

Yes, admittedly, the inverse is true for the weak;
With a mislynch on D1, the weak neighborizer has only a 2/8 chance of hitting scum, so statistically speaking the neighborizer is more likely to get an innocent than a guilty.

You need only look at the results of the game to see that the vig DID shoot town and the neighborizer DID recruit town for this.

weak hitting town n1 when its the strongest possibilty, weak hitting town n2 when its still the strongest possibilty.
vig possibly holding and either eliminating mislynch oppurtunities for scum or actually killing half the scum
and rolecop possibly getting a guilty on the role blocker, a scum sided role.

Frankly, if scum with two methods of shutting power roles down (the nightkill + the roleblock) let themselves get in the situation where weak gets two results or vig gets a shot on scum or rolecop gets a guilty on scum.

They probably deserve to lose.

This was almost a micro in size; the scum could not afford to hit VTs and leave the power roles unchecked, yet that's exactly what they ended up doing. Sure, they roleblocked the vig N1, and later nightkilled said vig...but at the time they nightkilled said vig, said vig was a VT thanks to the previous roleblock--unconfirmed and without a shot, thus both mislynchable (well theoretically at least) AND not a threat via night actions. In fact, killing the vig confirmed the existence of the scum roleblocker, a HUGE tactical mistake.

[mastina]

the game was obviously centered around a swingy role.

I mean, I can't argue with that.

It was, but it was the mods' choice to use that role. Our job was in part to make the game a little less reliant on that one role.

[mastina]

my personal belief is the game should be won in the day phase by either alignment.

Definitely by town, but the job of scum with their dayplay is just as much to influence nightplay as townplay--by which, I mean. Scum should be able to, using

dayplay

, manipulate town PRs into using their actions in pro-scum ways rather than pro-town ways.

Make the rolecop get useless results, make the vig shoot town, LET the vig shoot town (roleblocking the vig, while it stops the vig from being conftown, is probably less of an asset than just letting the vig shoot town), preventing the weak neighborizer from targeting scum or even better cause the weak neighborizer to target scum N1 with NO BREADCRUMBS (or breadcrumbs leading to town).

I think you can understand what I'm getting at; the job of scum with their dayplay is to augment their night play, whereas inversely, the job of town with their nightplay is to augment their dayplay. Yes, vice-versa applies to some extent as well, but the primary directive of a town player is to first win by dayplay and then by nightplay; the primary directive of scum is to control the results of the game through nightplay (primarily, the nightkill) and only then manipulate dayplay.

the night phase is made to intentionally help town

Actually, the night phase is where scum are strongest--the power of the scum nightkill should NEVER be underestimated. The nightkill is singlehandedly the reason why mountainous games which are theoretically balanced pragmatically end up as scumsided; there's a good reason why one of the main ways to make mountainous games be balanced is simply removing the scum nightkill. Because the scum can kill the players they feel are most critical to leading the town to victory, and with said players absent, the town is put into a state more favorable to the scum.

That's why I said that if the scum let the town control the result of the night in spite of having a full roleblocker and their nightkill, they probably deserve to lose. Because during the night, they should absolutely NOT be letting themselves be outplayed by town, because night is where scum are naturally at their strongest.

You can be the best manipulative scum player in the world by dayplay--but because dayplay is the realm of town, you can still being that master manipulator lose by dismal nightplay, because while strong dayplay influences scum victories, it is not enough to hand them victory in of itself. (Most of the time, anyway.) The nightkill is singlehandedly the most important tool scum have; misusing it can and will cost scum the game. Utilizing it to its maximum utility, inversely, gives scum a near-guaranteed win.

(Incidentally, this is also one reason why dead townies in dead PTs rant about how obvious the scum are--of course the scum are obvious to the dead townies, that's why they're dead in the first place! Because the scum didn't want them alive to point out the obviousness of it; they wanted to leave only the townies too "stupid" to see it alive, more or less.)

Power roles

are

made to deliberately help town--they serve more or less as the counter to the scum's two main advantages. The nightkill first and foremost as scum's strongest weapon, and being informed as scum's second-strongest weapon. (The scum, knowing who is town, who is scum, and by their roles having some insight into the nature of the setup, have a better ability to manipulate the gamestate to their favor using this information alone. When you pair that ability to manipulate off of knowledge WITH said nightkill, it makes scum able to dominate a game unchallenged unless town PRs are in place which can provide a counter.)

Primarily, town's roles can be sorted into three categories, each serving as a counter to one of the scum's weapons. Investigative roles serve as a counter to the scum holding more information, by giving town a way to artificially boost the amount of information they possess. This game doubled up on these, at the cost of not only giving up protectives but also increasing the risk of losing information.

Killstopper roles (which protectives are a subset of) serve as a counter to the scum nightkill, by preventing the scum from eliminating the town player serving as the greatest obstacle to a scum win. This game had none.

And town killing roles serve as a source of an extra town-controlled death, to prevent the scum from having total control over who lives and who dies. Them being in the hand of a single player means there's no direct interference scum can give. Scum, simply by not bussing, can make lynching scum MUCH more difficult, but in contrast, scum can't so easily stop a vig bullet from shooting them. (And in Normals, even under the new guidelines, any killing role also doubles as a type of informative role because it is public knowledge that scum cannot hold extra killing roles.)

These are what the town's roles do to counter the scum's innate inherent natural advantages. The town is given boosts--significant boosts. But these boosts, if balance is at the level it should be at, don't inherently overwhelmingly give the town the advantage, so much as they do neuter the natural scum advantage.

The idea of town power roles is thus, the "necessary evil" of reducing the power scum naturally hold. Because scum do, by default, hold power.

The problem is that the more power you introduce, the greater the swing, so the harder it is to get an exact calculation. The night is then a delicate balance, so you have to take educated guesses on what the level of power each side has actually translates to. (And make no mistake, it's always an educated guess. Sometimes, we guess wrong. Other times, we guess that it's close enough, but close enough still ends up at being like 60% in that one side does overall hold a slight but not significant edge over the other. But it's always a guess. Never an exact science.)

basically i feel like the "pure luck" is more likely to happen to town in THIS scenario.

The more swingy the game, the more luck plays a part in the outcome, yes. That I can't deny, and with this game being very high in swing, it was very high in luck for the town/unluckiness for the scum.

However, I stand by the statistics that, statistically speaking, the town is actually less likely to luck out--I maintain that, by the odds, the town is most likely to have night action results which provide a minor boost but don't break the game in favor of the town. As far as luck in night actions go, the town got about as lucky as it could: the neighborizer neighborized one of the other two PRs, and the vig shooting said PR was prevented from shooting said PR, and then the neighborizer told said PR their target which landed on scum, giving town a guilty. You CAN find results better than that for the town, but not by much; the town got really, really, REALLY lucky, about as lucky as is possible without being the MOST lucky.

Like, in terms of luck. The outcome of this game is in the top-5 town-luck outcomes I'd estimate. Yes there are outcomes more lucky for the town, but not very many. This game had dozens, hundreds of possible iterations which could have played out, and of them only the smallest fraction favored town more than the outcome of this game did.

By the numbers, the vig is likely to shoot town. And did!
By the numbers, the rolecop is unlikely to produce a useful result. It can clear the weak neighborizer, it can clear the vig, it can catch the scum, but it can be roleblocked and it gets NAI results on half the players in the game (the 5 VTs) and critically above all else it doesn't

know

this going in so it has no idea what to aim for, how likely it is to get an innocent or a guilty.

By the numbers, the weak neighborizer is unlikely to produce a definitive guilty, can produce a FALSE guilty (imagine a vig shot going through, the weak neighborizer targeting town, and the scum nightkill going through, with one of the vig/scum targeting the neighborizer--OOPS LOOKS LIKE TOWN MISLYNCHES THE WEAK NEIGHBORIZER TARGET), can be prevented from giving a result beit roleblock, vig killing target, scum killing target, vig killing neighborizer, or scum killing neighborizer, and on top of that even if the neighborizer gets two innocents, by the time this happens the game is likely in lylo...a time where investigation results are likely to be doubted.

(Keep in mind that even if the vig holsters, with 10, D3 gives you six alive, mylo; with a vig shot, that's five alive, lylo. The neighborizer being a NEIGHBORIZER would be proven, but them being WEAK would not be proven, so there's room to lynch the neighborizer and/or a neighborized player if scum work up the paranoia to push that angle.)

When you look at it from that perspective. The vig likely to shoot town, the rolecop only having ~50% chance of producing a useful result and not knowing they even

can

produce a useful result (one of the things which weakens rolecops is that rolecops work with imperfect setup knowledge and thus don't know what they should be doing; you can't assume a rolecop operates with perfect setup knowledge when designing/reviewing a game because they never do), the neighborizer having multiple ways to fail including the ability to

falsely condemn an innocent town player

(THIS IS A MASSIVE ONE) and the ability for the game to end up in mylo on D2 when the town has basically no information at all.

The fact that most of this ended up not playing out is, statistically speaking. Quite improbable. The outcome of this game was not one of the more likely ones. It's not the luckiest town outcome, but it is still an incredibly lucky town outcome. One which I maintain wasn't as likely as you think it was.

[mastina]

My last question is at what point was the backup ninja introduced and why did you feel the need to mess with scums knowledge of the setup when all scum had was a role blocker?

Honestly you'd have to ask the reviewers/mods because I gave initial feedback on the original setup, and then the next time I checked into the review, the finalized setup was already posted (the setup took less than a page to be finalized so that was a fairly quick process); I didn't really pay attention to the in-between steps because the in-between steps were null-and-void as they weren't current.

When I review a setup, I tend to give feedback on only the most recent idea and refer back to the original idea--basically, because I'm looking to preserve the moderator's original vision. (Looking at the steps in-between usually is a waste of time then.) The only time I look at intermediary setups is when said moderator's vision changes into being one of those intermediary setups, at which point THAT becomes my reference to use in place of the original setup, if that makes sense.

I felt the setup's most critical piece, the essential core of the mod vision, was the 2x Weak Neighborizer, so everything revolved around it, referring to what the mods wanted it to do with the setup proposed, and I thought it was

more or less

loosely in line with what they were going for, just with better balance provided.

That's a personal reviewing method tho. You can kinda sorta think of it as similar to the 60% rule: "eh, close enough to striking the balance between 'what they originally wanted' and 'what is a balanced setup'".

In other words. I don't actually know this because the change was made during a time I wasn't giving input. The results OF the change looked to be in the right area, so I signed off on it without caring how the changes got to be at the point they did.

[mastina]

Hopefully she doesn't feel like she's banging her head against a wall.

Quite the opposite, I worry I come off as being preachy and that my spews are too personal-opinionated (and come across as trying to win an argument) when neither is my goal. I try to be informative, objective, and accurate in my explanations to let players know, "this is how the process usually works and why it works that way", but I never know if I get it right.

Explaining processes to let people know how they work, why they work that way, so that they can walk away hopefully feeling both a little more at ease with the process and also enriching their perspectives so that they can explore things from an angle they hadn't thought about before, skills which with luck transfer into future games. If I do any of that, then I can call it a success.

[mastina]

I think this kind of setup where small % NAs can blow the game right open is just controversial by nature.

No argument there.

It's something I love in themes because really themes are fun, but in Normals I kinda loathe it because I know that postgame there's gonna be hell to pay as I'm put in the awkward position of trying to explain why the hell I passed it knowing the outcome they're angry about could have happened.