The New Newbie Game Setup

[Toomai]

I plan to continue my stats-tracking with this new setup. So I'll need to get things set in stone as soon as possible so I can re-create the new spreadsheet.

Here's my current list of assumptions, please correct me if necessary.

• We are using the new Creature-split table everywhere, and the old 3x3 table nowhere. This is important because they have different rules as to which re-rolls are necessary.
  
• The SE and IC system remains unchanged; games can be 5Nb-3SE-1IC or 6Nb-2SE-1IC depending on supply and demand, and this is not expected to change.
  
• There are 27 possible 3d3 rolls:
  • There are 15 rolls that are invalid setups, and are re-rolled:
    • 1,1,1
      
    • 1,1,2
      
    • 1,1,3
      
    • 1,2,1
      
    • 1,3,1
      
    • 2,1,1
      
    • 2,1,2
      
    • 2,1,3
      
    • 2,2,1
      
    • 2,3,1
      
    • 3,1,1
      
    • 3,1,2
      
    • 3,1,3
      
    • 3,2,1
      
    • 3,3,1
  • There are 12 valid setups of equal probability:
    • 1,2,2: Tracker & Cop vs Roleblocker
      
    • 1,2,3: Tracker & Jailkeeper vs Roleblocker
      
    • 1,3,2: Roleblocker & Cop vs Roleblocker
      
    • 1,3,3: Roleblocker & Jailkeeper vs Roleblocker
      
    • 2,2,2: Doctor & Cop vs Rolecop
      
    • 2,2,3: Doctor & Jailkeeper vs Rolecop
      
    • 2,3,2: Rolecop & Cop vs Rolecop
      
    • 2,3,3: Rolecop & Jailkeeper vs Rolecop
      
    • 3,2,2: Doctor & Tracker vs Goon
      
    • 3,2,3: Doctor & Roleblocker vs Goon
      
    • 3,3,2: Rolecop & Tracker vs Goon
      
    • 3,3,3: Rolecop & Roleblocker vs Goon

[Toomai]

So a system like A23 B33 C22 or whatever might be helpful (I'd tend to visually separate out the scum die roll from the town rolls but that's admittedly mire of a "me" thing)

I really like this for readability, but it comes with a disconnect from the codified setup table.

[Toomai]

Maybe we should take the two town dice and change them from "2 or 3, reroll 1" to "1 or 2, reroll 3". It'll look less confusing at first glance (people won't be asking where the 1 is).

		Table A		Table B		Table C
First die	Mafia	Mafia Roleblocker		Mafia Rolecop		Mafia Goon
		Number 1	Number 2	Number 1	Number 2	Number 1	Number 2
Second die	Town	Town Tracker	Town Roleblocker	Town Doctor	Town Rolecop	Town Doctor	Town Rolecop
Third die	Town	Town Cop	Town Jailkeeper	Town Cop	Town Jailkeeper	Town Tracker	Town Roleblocker

This means the list of setups would be:

• A11: RB x Tr + Cp
  
• A12: RB x Tr + JK
  
• A21: RB x RB + Cp
  
• A22: RB x RB + JK
  
• B11: RC x Dc + Cp
  
• B12: RC x Dc + JK
  
• B21: RC x RC + Cp
  
• B22: RC x RC + JK
  
• C11: Gn x Dc + Tr
  
• C12: Gn x Dc + RB
  
• C21: Gn x RC + Tr
  
• C22: Gn x RC + RB

[Toomai]

I think this is almost the clearest chart, but one thing that was clear in one of the charts but is not clear in this one: what are mafia's safeclaims? You can kind of parse it out if you think about that in Table A the safeclaims are doctor-or-cop, in table B they're tracker-or-cop and table C they're roleblocker-or-jailkeeper, but that's not obvious from the chart and it should be.

I disagree that safeclaims need to be obvious from the setup chart.

Additional note: If a Town Tracker targets a scum PR who is PR'ing someone and killing someone else on the same night, the Tracker will receive a result of "Your target visited both A and B last night", which is basically as strong as a cop guilty. I don't think this has happened before in Newbie setups, so this may need to be clarified in the ruleset as a possibility.

[Toomai]

I'm worried that this setup will lead to a lot of mod errors. My experience is that very few players and mods really understand NAR, and it's very relevant in some setups.

Here's an example (which I expect many people to get right but many to get wrong): scum RB kills town RB and blocks town JK, town RB blocks scum RB, town JK jailkeeps the scum RB. What happens?

Spoiler: What actually happens

Even though the scum RB takes precedence when there's an action loop, and there

is

an action loop here (the scum RB and town JK cross-blocking), there's also an action that isn't interfered with: nobody is trying to prevent the town RB's action (the kill attempt wouldn't count as stopping an action). So by NAR's Golden Rule, that triggers first, blocking all the scum RB's actions. The jailkeep on the scum RB also therefore succeeds, although it doesn't matter because the scum RB couldn't act anyway (it would matter if someone tried to vig the RB, as it'd cause that to fail).

EDIT: Based on discussion in this thread, a second example: scum RB kills and blocks town RB, town RB blocks town JK, town JK jailkeeps scum RB. What happens?

Spoiler: What actually happens

We now actually do have an action loop involving all the actions but the kill (and the kill is potentially blocked as part of the loop). So we pick the scum roleblock to automatically succeed (as it's the first action in tiebreak order). That works, blocking the town RB. As a result, nothing is blocking the town JK, so the scum RB is jailkept; their roleblock has already happened but their kill hasn't, and so the kill doesn't go through.

Well the good news is that these complex situations can only arise in 1/12 subsetups, and the rules are pretty clear about what happens in the other 2 setups involving multiple blockers should they cross-target.

That said, when it comes to these few convolulted cases, maybe the newbie queue should have in one of its first posts a section for "if you come upon a really weird interaction that you're not 100% on, here's exactly how it should be resolved".

(though your second example shows a disconnect between whether a

player

is blocked or their

actions

are blocked; that needs to be sorted out)

[Toomai]

I was about to say that the first post already solves interaction problems, but I checked and it kind of contradicts itself.

Setup Specific Rules:

1. Mafia Power Roles may both use their night action and preform the kill.
2. Should a town roleblocker and a mafia roleblocker target each other, all actions preformed by both, including night killing actions, will be stopped.
3. This game always uses Natural Action Resolution to resolve night actions.
4. Roleblocker actions have priority over Jailkeeper actions.
5. The Mafia Private Topic is open at all times, including Day phases.

#3 and #4 here are in conflict. Are you using NAR, or are you saying Roleblocker always goes before Jailkeeper?

[Toomai]

Isn't it still NAR if you declare one role to have precedence?

I'd say "no" in the very technical sense, since they both fire in the "block" slot, and the NAR wiki page doesn't go any deeper than action type levels. But I'm nitpicking. Cross-blocking can only happen in 3/12 subsetups and will statistically probably only happen in a fifth of those.

On the various ideas for re-visualizing the setup matrix: While they look cool and are easy to understand the setup as a whole, I think there still needs to be a way for players to take it and quickly get a short identifier for each subsetup. Players currently say things like "okay we're either in setup 1 or A" and it's not hard to grasp/look up; I think it's important to have that available somehow. All the fancy tables with "or" in the cells and such don't offer that.

[Toomai]

How does it change the balance of the setup to make one or more of the blockers unblockable? That would help simplify things a lot.

It's pretty important for the town RB to be capable of blocking the scum RB if the scum guesses wrong for who the town PR is. (and vice versa)

[Toomai]

I'm of the opinion that as long as the cross-blocking/circular-blocking interactions are given canonical listmod-designated setup-mandated resolution instructions that players and mods alike can refer to at any time (probably in the newbie queue's setup post, and ideally also in the setup post of every game), it doesn't matter too much what exactly the resolution is. If someone gets it wrong even after that, what can you do.

[Toomai]

Also, here's my new idea for displaying the setup matrix and explaining the generation rules:

	A	B	C
	Mafia Roleblocker	Mafia Rolecop	Mafia Goon
1	Town Doctor	Town Tracker	Town Cop
2	Town Rolecop	Town Roleblocker	Town Jailkeeper

1. Pick one of the three columns (A, B, or C) at random. The role from the top row is paired with a Mafia Goon to form the scum team.
   
2. From each of the two other columns, randomly choose the role from either row 1 or row 2. These two roles are added to the town.
   
3. Add 5 Vanilla Townies to complete the 7v2 setup.

I now think simple is the way to go with this. It's basically the original 3x3 table, but with headers added for a more concrete explanation than "one from each column".

The subsetup shortform names would be:

• A11: RB x Tracker + Cop
  
• A12: RB x Tracker + JK
  
• A21: RB x RB + Cop
  
• A22: RB x RB + JK
  
• 1B1: RC x Doctor + Cop
  
• 1B2: RC x Doctor + JK
  
• 2B1: RC x RC + Cop
  
• 2B2: RC x RC + JK
  
• 11C: Goon x Doctor + Tracker
  
• 12C: Goon x Doctor + RB
  
• 21C: Goon x RC + Tracker
  
• 22C: Goon x RC + RB

(Maybe B21 is better than 2B1? Unsure. I think 2B1 is easier to read across the columns.)

[Toomai]

What is the scum winrate over the past ~20 completed games?

Scum winrate over the last 25 completed games is 64%, but I think this is because replacement rate has skyrocketed to 4.84 players per game in that time, the highest it has ever been in the Matrix6 newbie queue (less than 50 games after it hit the

lowest

point in history at 2.36).
I should be able to update my stats thread within 24 hours, but you can see these numbers in the raw spreadsheet yourself.

[Toomai]

Various notes from my unofficial position as stats guy:

• The general point of "statistically-significant" is 30. It will take several months to build up that many games. But also note that winrates may be skewed to start with simply because SEs/ICs don't know the setup as well yet.
  
• Because there are 12 subsetups, it will take quite a while before we'll have any significant per-subsetup stats (perhaps years). Even per-role/per-subsetup-group stats will take a long time to gather.
  
• Daytalk was strong in Matrix6 because the setup was originally designed without it. Since 3d3 was designed with daytalk in the first place, it should at least be reasonably more balanced.
  
• There is evidence that games with higher numbers of replacements are more likely to result in scum wins. I think the roles in this setup, plus adding daytalk, will result in fewer "lost interest" replacements.
  
• My stats spreadsheet, once it exists, will be using the subsetup representation in this post of mine because it's easy to implement mechanically and is simplistic visually. But if the official newbie queue post uses a different chart than the 3x3 matrix currently in the first post of this topic, then I guess I'll have to figure out how to match it.

[Toomai]

1. I'm unqualified to comment on the setup balance.
   
2. During the Matrix6 daytalk experiment, replace rate for scum newbies dropped from 50.0% to 34.1%. But replace rates for all other playertypes went up (less starkly). Since this would be a whole new setup, I would expect SEs/ICs to be more inclined to stick around to try it out for a bit. Furthermore, the newbie queue is currently experiencing the highest replace rates in the Matrix6 era (though partly just because August and September are the worst months for replacements, with people losing time since summer's ending and all that). So yes, I do believe changing to this setup would reduce replacements, at least for a bit.
   
3. The newbie queue does not have a "test environment", so if people want a change there's nowhere else to do it. And if it does go awful, Matrix6 is certainly not broken enough that we can never go back.

[Toomai]

Where are the updates? What happened? Newbie games all have BP claims now.

That's hyperbolic. I skim every game to keep the stats, and I feel like the odds someone brings it up are closer to 60% or 65%. (This is not a statistical number, it's a guess.)

[Toomai]

1) Newbies need a

simpler

set up. More focus on learning the ropes than the complexities of semi-open setups.

Counterpoint: SEs and ICs need a setup that's sufficiently interesting or they'll grow tired of it and stop joining the newbie queue.