Mastin's Guide to Scumhunting

Type: Views Author: Mastin2

History

Original Publication: September 9, 2011 by Mastin2.

Original Thread.

Last Revised: March 20, 2017 by Mastina.

Introduction

Nowadays when I scumhunt, I tend not to rely on the scumhunting tools I used to. While things like meta and VCA will never be obsolete, they have their time and place for being used; my methodology by and large instead focuses on three basic parts and two extras to support them. Those tools are…

Interactions

I very frequently say, “This was NOT townVtown”, “this interaction looks townVtown”, “this interaction looks like scumVtown”, or even, “this interaction reeks of scumVscum”.

Why I Use This

Interactions are honestly my strongest, most frequent tool, and I use them even during the first day. There is a camp of people who feel like doing associatives before a flip is stupid. They are not necessarily wrong. Making stupid arbitrary assumptions that a player must be town because someone else is scum or that a player must be scum because someone else is town is generally not a smart idea. However, I specifically said interactions for good reason. You can get good, accurate, reliable reads from interactions when you augment them with other methods.

Interactions work because mafia’s a social game. A key part of any social game, inherent in the definition, is that socialization requires interaction. Specifically, inherent in the scumhunting process of using interactions is the requirement to think that town players and mafia players, thanks to differing standings, will therefore interact and behave differently: mafia are the informed minority, starting out with inherently more information than the town; town players are the uninformed majority, and therefore start out inherently lacking the information available to the scum.

This is particularly easy to conceptualize in terms of “condemning interactions” (interactions which strongly tie a player to scum), and “no condemning interactions”. Buddying (especially White Knighting) is a form of a potentially condemning interaction. True OMGUS (used not as a buzzword, but from the true meaning of the term) is another. A classically held one would be Chainsaw Defense.

It should be noted that just as important as the condemning interactions is the lack thereof, because sometimes a player who otherwise would look scummy simply lacks anything which would otherwise indicate they are scum. No scumteam with them works. Alternatively, literally every scumteam works with them. (Which is typically a sign a player is not actually scum.) This type of “individual scumminess” is typically the hallmark of an individual who lacks a reliable network of support, i.e., town.

Honing Interactions

Just because you know that interactions are a good tool doesn’t mean you know how to use them. So here’s how I value things.

Early Trumps Late

When in doubt, check the early game, because the interactions there will always be relevant to the game. Nine times out of ten, they will not fail you. The reason why is simple: the scum are still getting a feel for the game. (Everyone is.) They don’t yet know what they need to do in order to successfully manipulate the town.

Later into the game, any remotely competent scum player will have learned which buttons to push: they know what they need to do in order to maximize their chances of getting a factional victory, be it strategically bussing, not bussing, distancing without bussing, whatever; the scum have gotten a feel for what is “needed”. This is most evident when a scumbag is about to be lynched—it’s universally known you should ignore the posts by the scum at that point because they’re so filled with intentional WIFOM that there’s no way to get a reliable read out of it.

It applies then, it applies across the board: the longer the scum are in a game, the more they know exactly how to get the town thinking that they’re not scum. It becomes easier to fake being wrong convincingly. In contrast, with early-game interactions, they don’t yet know what will work. They’re still trying to figure things out.

Know Your Accuracy

Specifically, the earlier you use interactions, the more false positives you are likely to receive. The later game gives you vital contextual clues as to what was a lead and what wasn’t. The more information you have available, the more accurate your interaction analysis will be. While I will use interactions on D1, they will not be nearly as accurate as those on D3.

Know What You Can Get

Using interactions is actually a great way to townhunt earlier in the game. While the use of interactions to catch scum increases exponentially the later into the game you are, they still are best for finding town when using early interactions. I get lots of false positives on scum interactions. I almost never get false positives for LACK of scum interactions. Meaning, if you see someone who looks like they lack condemning interactions, it has a significantly higher percentage of being correct than a player with condemning interactions.

Look For Trends

Patterns in interactions, patterns in data, these are both things I attribute to being pieces of interactions. You are looking to reverse-engineer the events of the game and find a way for them to plausibly fit.

What NOT To Do

When commenting on the nature of interactions, never leave it at just that. If you see a town-town fight, try to break it up and direct attention onto players who are actually scum. You want those players to be working together. If you see a scum-scum interaction, you absolutely need to explain why you think it’s “scum theater”. If you think one participant is scum, you better be explaining why you think that person is scum and the other person is scum. But you should be making these comments, and not ignoring them just because neither player has flipped yet.

Biasing Yourself

There is basically no situation where you are doing yourself a favor by arbitrarily assuming something must be true, because of whatever factor. This player must be town because that player is totally scum. This player must be scum because they are the only unflipped player on that wagon. Things like this, you want to avoid. You cannot make progress if you are discarding evidence to fit a preexisting presumption.

You can make notes: “this fight doesn’t seem town-town” would conflict with “I am townreading both participants”, sure enough. But how do you know which is the right conclusion? You cannot out-of-hand arbitrarily assume one is correct and immediately throw out the other. So instead of blindly forcing these conclusions, reason it out and find which is more likely to be true. Resolve conflicting data by looking closer at both sides, not by looking at only one side.

Further Reading

Interactive Tells, an article I wrote on the subject after the original publishing of this article.

Motive

You can think of this as being summarized by one question: “What was this meant to accomplish?”

What was their reasoning? Does it make more sense as coming from town, or scum? What reason does the player have to do this action as town, versus what reason do they have for doing this as scum? What does this player have to gain as town from this action, versus what this player gains as scum from this action. Equally as important, what does this player have to lose as town, versus what they have to lose as scum? What drove them to do what they did?

Getting inside the head of a player is the key to success, since superficial scumhunting often only catches weak players rather than actually scum players.

Intention

Often, what a player is doing is different from their intention, because humans are flawed beings who make imperfect moves. So instead of looking at what a player is doing, look at what a player intended to do. Think of things from their perspective. Do they genuinely think they have delivered what they say they they have? Even if what you see from them is in your mind a piece of junk, by looking at their viewpoint, you might understand why they feel what they have done is legitimate.

Mindset

Looking at how a person is literally thinking can be extremely revealing as to their alignment. If you see someone pulling actions which look consistently scummy (like, say, constant bandwagonning), your first thought is probably going to be “this guy’s obv-scum”. But if you take a look into the mindset of a player, think about their actions. Do they look like scum lazily taking advantage of every wagon, or town doing what they see as legitimate scumhunting? Far more often than not, it will end up being the latter.

Certain mindsets are almost impossible to come from scum, and if you are looking at mindsets, they are also incredibly difficult to fake or cover up. What a player feels is not easily hidden. Look at their jumps in logic. This can admittedly take practice, because while motive measures what they actually post, mindset measures what they were THINKING at the TIME they posted. You are looking for a naturally-flowing trajectory in thought, and to figure out how they got to point A to point B.

This transition is not necessarily linear or smooth. (In fact, a perfectly linear progression is often a sign of a scum player.) However, while town players' chaotic minds will make sudden changes, these should be traceable to a definitive trigger that sparked the thought; with scum, this trigger is often absent.

Advanced Thinking

One refined version of this way of thinking is NKA. Yes, Nightkill Analysis is actually a form of motive, applied to the whole scumteam: you are trying to figure out why the scumteam chose that player for death. This can be any number of reasons: accuracy of reads, Power Role hunting, the charisma of the player, fear of the player's competency off of past experience, the status of the player as obviously town, in an attempt to avoid failed/suicidal kills, or usually, the player who is the greatest combination of these factors plus whatever other oddities the scumteam generates for reasoning.

This reasoning allows you to paint a picture of who the scumteam is, as certain players are more likely to make certain kills than others. Additionally, if you apply nightkills over time, typically, a trend in the data emerges. This trend crosses back over into Interactions as a scumhunting tool, as seeing the trends in the data is applying a motive on a game-wide scale.

Know Your Players

This begins to get into the use for meta, as it should be utilized. having a generally good feel for a player and their style makes it far easier to use motive as a scumhunting tool, as your familiarity with their style will allow you to know which moves are more likely to come from town, which moves are more likely to come from scum, and which moves you might otherwise misattribute to one of those two are simply part of that player.

It is generally best to place emphasis on the last part in particular: some players are far more likely to make certain moves than others. Knowing “how they play in general” is something you can and should value far more than “how they play as scum” or “how they play as town”, both of which you are far more likely to be wrong about especially if viewed superficially.

What To Avoid

Motive is meant as a tool to read others. It is quite often a mistake to try and apply it on yourself.

Tone

Tone is a controversial tool, because people insist it translates poorly over the internet. This is less true than most initially assume. You actually can read into what a person is saying, and the way they are saying it will make a difference. A huge tip for beginners in Tone is to look for CHANGES in tone.

Scum shifting gears when the state of the game suddenly shifts from being in their favor to massively against them will be quite different from town who are ticked off they aren't getting what they want. A player who successfully pushed a mislynch, then displayed doubt, is another; yes scum can display this, but it is hard for them to fake convincingly. Perhaps these are not the best of examples, but they give a fair idea of what I mean.

Scum, as the informed minority, know when they are right or wrong; town, as the uninformed majority, do not. This displays in their posting more often than it does not, even if only in subtle ways.

Natural vs. Artificial

Some have argued that town should try and make themselves look town—but many others pointed out the flaw in this, that their play would become artificial, and that if they just played naturally, they’d look better.

It’s not that hard to understand why. If someone is artificially constructing their posts, chances are, they have something to hide. Scum want to maintain a façade that is delicately built. They have to employ all the correct wording, use exactly the right phrases, in order to get the town to believe they are town. Town on the other hand want to appear as natural as possible, like they’re flowing without interruption, like there’s nothing to hide, nothing not to be seen.

What Tone Is For

Tone is, at its best, essentially a combination of both interactions and motive: it is finding patterns in speech which reveal an underlying cause, be it pro-town or anti-town.

VCA

Vote Count Analysis is an incredibly valuable tool, but it is exactly that: nothing more than a tool. It cannot catch the entire scumteam every single game in its entirety with any sort of reliability, at least, not by itself. The value behind VCA is that it is essentially a shortcut to all three of my primary tools, acting as a summary of the game.

VCA catching interactions is obvious enough, as VCA is meant to find trends in data. You can find motive behind every single vote, as a vote represents the ultimate expression of a player's actions. Yet every vote also has a tone to it. A tone harder to read than normal posts’ tone, but still present. A post-with-a-vote-only, a post with a one-liner and a vote after it, a post with a vote then one-liner, a post with a paragraph and then vote, a post with a vote then paragraph, etc. They all give off a different tone. A subtle but noticeable difference in the overall message of the vote.

Usage

There is no universal method for utilizing VCA, and even I have differing levels of it, which go into different depths and emphasize different aspects of my scumhunting tools. However, one thing is universal about VCA, and that is...there is not anything universal in good VCA. The best VCA is done on a case-by-case basis.

Deployment

My recommendation is to use VCA as an augmentation to your other scumhunting methods, as one tool among many. It is acceptable for you to use VCA to gain an initial foundation for further reads. In fact, that’s actually a reasonably good usage of it. However, it should be used as a good starting reference, not as your only reference.

Further Reading

Mastina's VCA Guide (currently available only on the forums) goes into further depth as to best utilize VCA.

Miscellaneous

Sometimes, the player who looks likes obvscum…actually is obvscum. Basically, just because an established scumhunting method may rely on superficial assumptions doesn’t mean those assumptions will always be wrong. They are also something easy to see, easy to understand, and easy to reference, so traditional scumhunting tools, of which there are too many to list, have their place.

I point out lurking. I am quite familiar with active lurking. I look out for bandwagonning. I have a keen eye for parroting. I point out people who are overdefensive. These are all things which we consider scummy for good reason. It’s what people have been using since the birth of Mafia as we know it, and there’s a dang-good reason for it—because they work, they’re generally reliable, and do catch scum. However, I tend to use them mostly as an augmentation for my primary three.

They can be used for lack of any other indication, if it is literally impossible to sort interactions, if no mindset sticks out as especially more likely to be an alignment for a given player. But these should be appropriately marked as so in your reads for being as superficial as they are. They only gain their strength when you have the backing of interactions, motive, and tone to go along with them.

In fact, they are quite good for helping to explain motive and even interactions to others—point out the mindset you have observed, and then uses all these miscellaneous methods as your evidence, your proof, of this motive being true.

General Advice

More on Meta

Meta can help you with all five tools, if used accurately. Getting a feel for how a player plays as scum, how a player plays as town, and how they play in general can give you a much better baseline for which interactions to place emphasis/value on, and which to NOT place emphasis/value on. Similarly, it can highlight which intentions are town/scum, and what the player’s mindset is. It also helps establish what tones they typically take.

The reasons I don’t recommend using meta by itself are that it requires too large a sample size, requires too much time, and is often subject to change: certain players know and can manipulate their metas better than others. Experience from a single game is also from a single game; differing circumstances can and will alter how a player plays regardless of their alignment, something meta rarely tends to account for.

It should be noted that I have developed a distinction between two things which are both traditionally associated with meta. Meta, how a person plays; Personality, who a person is. One can be manipulated. The other cannot. A person’s personality is just who they are, at their core. If you use meta, you NEED to be able to differentiate between the two. Meta is part of the game; Personality is part of the person outside the game.

If you lock onto a person’s Personality, you can understand how they player far better. Get into the psychology of the game; understand the way their mind works. Personalities might change over time, but these changes are very slow and gradual as a person ages with new life experience. As a result, they tend to overall remain the same, staying reasonably consistent. If you can lock onto a person’s Personality, and get a feel for their general meta, you have basically learned how to use the tools to consistently nail their alignment every game.

Practice Makes Perfect

Nobody gets to be as good as they are by natural talent, by natural skill alone. Nobody can one day wake up and suddenly magically become a Scumhunting God. To get there requires hard work and effort. Namely, time, to refine methodology. You will note master a technique the first time you try to use it.

These techniques are no different; when you start to use Interactions, Motive, and Tone, you’ll likely make a lot of mistakes at first, because you haven’t learned the tricks of the trade, yet. I can give you a guiding hand to get the process started, but the subtle nuances and intricacies cannot be taught; they must be learned. So, only after you’ve used them often enough will you make them work reliably and frequently.

Read My Guide

While you’re free to agree/disagree with as much of it as you like, the principles I laid out in there have helped me a great deal over the years.

Be Confident

If you consistently doubt yourself, nobody will listen to you and nobody will follow you. You cannot let every possibility interfere in your presentation. There are times where it is okay to exaggerate, as long as you are not outright lying, and in terms of getting townreads or scumreads, this is incredibly useful. Show confidence, show assurance, show that you have a strong will to push your reads. People want to listen to those they think hold strong, yet reasonable, reads.

Keep Notes

Let me tell you, the quality of my play skyrocketed when I began to take notes and degenerated massively when I stopped taking them. Note taking allows for better organization and memorization of key details in the game. Especially in games which feature an absolutely massive amount of pages, taking good notes will allow for you to quickly reference key pieces of information.

Always Seek Improvement

I like to think of myself as an “Eternal Newbie”. That is, I never stop growing. I never stop learning. I never cease to try and better myself as a player, and am never content with myself. I never settle down into a particular style. I constantly evolve myself, with every game, using experience as both alignments to improve my game.

This is absolutely vital to consistently performing well. The game of mafia is constantly evolving. Other players all around you will change with the game, so if you aren’t doing the same, your performance will stagnate, or even worsen. Never be afraid to try new things. Some will work, most will not. Incorporate the good, evaluate the bad and determine if they have use in a different time. Put genuine time and effort into trying to learn from your past mistakes, and you will get better.

Keep An Open Mind

There may be a fine line between an open mind and doubting yourself, but just as important as knowing when to firmly plant your foot down is knowing when you need to be malleable, knowing when you need to change and adapt. Be willing to see alternative viewpoints, but don’t instantly accept them as being superior to your own. If you can strike the balance, you’ll be able to see things in a new light, while still looking confident about yourself.

Look At Others

You can always learn something from looking at others. Look at the players who are champion town players, and what makes them so good at being town. Look at the champion scum players, and the tactics these scum players employ. Look also to one of the greatest sources of new ideas: newbies. Try to encourage them to do their own thing, because that idea the newbie is pursuing just might be the next big thing, and be revolutionary enough that when refined, you yourself can learn from them.

Final Word

I realize nobody will be able to use all of these tips. (Or if they can, won’t be able to use them all effectively.) But everyone should be able to use some of these tips. Now go out there and go wreck scum.