Micro 641: Confirmed Guilties (Game Over)

[zoraster]

I can check CFJ's setups for you, if you want.

Please do!

[BTD6_maker]

This is truly the biggest mistake you could have made. You sent scum a "very basic setup that shouldn't be broken" that was actually broken. If 1 is Town the scumteam is {2,B,C,D} but 3 is voting A, which is impossible. Therefore 1 is confscum.

If you ever run this setup again, let me know. I would like to help with giving scum the recommendations.

[Comparing Realities]

Suppose Infinity were Town.

In this case, Wingback, CR, and enomis are all scum for voting Infinity. In addition, Infinity is voting Creature so Creature is scum.

Hence Infinity being Town implies {CR, Creature, Enomis, Wingback} scum.

However, Farside22 (Town) is voting DLA (Town). This is impossible.

Hence,

Infinity is confscum

.

Your lies are no match for a Townie armed with logical analysis.

VOTE: Infinity

I have solved this already.

I haven't even made a post yet, and already you solved it.

Just to be sure, I'm in the process of checking if you actually did.

[Comparing Realities]

The game ended before I even got here

[Creature]

4 townies with a guilty
1 innocent child
4 scum

is a 50% EV

[zoraster]

This is truly the biggest mistake you could have made. You sent scum a "very basic setup that shouldn't be broken" that was actually broken. If 1 is Town the scumteam is {2,B,C,D} but 3 is voting A, which is impossible. Therefore 1 is confscum.

If you ever run this setup again, let me know. I would like to help with giving scum the recommendations.

Hmmm for some reason this one got changed from the prior one:

Scum 1: A, B, C
Scum 2: D, E, 1
Scum 3:
Scum 4:
Town A: 2, 3, 4
Town B:
Town C:
Town D:
Town E:

I must have come across a different version in my notes when I posted it.

Anyway, i won't run this setup for a while until I've established a few aren't broken. I think it's better for me to provide a few options for scum to pick from rather than give them discretion anyway

[BTD6_maker]

That is not broken. Other teams include {2,A,B,C} and {1,A,D,E}.

[McMenno]

ah, the ol' page 2 quiclynch

[BTD6_maker]

1234ABCDE
=========
AAAA11122: Not broken
AAA141114: Broken (If 1 is Town, {A,B,C,D} so 4 is Town and E is scum, which is impossible)
AA4344433: Not broken
AAA144411: Not broken
AAAB41114: Broken (If 1 is Town, {A,B,C,D} so 4 is Town and E is scum, which is impossible)
AA4341344: Not broken
AAB433443: This is not a legal setup. What I presume you mean is:
AAB344334: Broken (If 3 is Town, {4,B,C,D} so A is Town and 1 is scum, which is impossible)

AAAB41441: Broken (If 4 is Town, {A,B,C,D} so 1 is Town and E is scum, which is impossible)
AAA142411: Broken (If 1 is Town, {4,A,D,E} so 2 is Town and B is scum, which is impossible)
AABC34334: Broken (If 3 is Town, {A,B,C,D} so 4 is Town and E is scum, which is impossible)
AAB334134: Broken (If 3 is Town, {4,A,B,D} so 1 is Town and C is scum, which is impossible)
AAAB41124: Broken (If 1 is Town, {A,B,C} and either D or 2 so 4 is Town and E is scum, which is impossible)
AABC13434: Not broken

Again, Zoroaster, if you run this or something similar again, please let me know.

[BTD6_maker]

You will need more setups in the list. If scum pick a "BTD-approved" setup currently, AAAA11122 and AA4344433 are bad (Town can lynch whoever has 3 votes). AA4341344 is also bad (lynch whoever has 4 votes).

This is not a complete broken as scum may have picked a setup that was not on my list, plus there is a lot of WIFOM potential here. This is why going by raw probabilities is often not helpful.
However, in both remaining BTD-approved setups, scum have a 1 in 3 chance of winning. I am not sure if it is possible to get better or if it is even meaningful (due to WIFOM).

[BTD6_maker]

My idea for a new version is that if Town lynch scum then scum must compulsively kill a Townie. Then everyone voting someone now dead will receive a new Guilty. It would ideally be White Flag because the 3P LyLo is a guaranteed scum loss.

[farside22]

Oh well the idea was fun, I don't think I could do it again.

[callforjudgement]

1234ABCDE
=========
AAA141114: Broken (If 1 is Town, {A,B,C,D} so 4 is Town and E is scum, which is impossible)

I produced this list for the previous run of the setup ages ago.
Looking at the first example, we have that if 1 is town, then 4, B, C, D are the scum. But 2 and 3 are claiming a guilty on A for some reason. So something's clearly gone wrong at some point in the process.

Sorry for accidentally ruining your game with a PM I sent in 2012. I can confirm that there's no typo in the example from what I sent, so something must have gone wrong at my end, either with the searching program or with the way I translated the results for the thread.

[callforjudgement]

1234ABCDE
=========
AAB433443: This is not a legal setup. What I presume you mean is:

Ugh. The list was generated by computer, it doesn't make typos.

My current guess as to what happened is that the algorithm is right but it made mistakes in the output code, giving players the wrong names. (The presence of any illegal setups should have been a huge red flag that something was wrong.) I can confirm that the illegal setup was in the original PM.

[callforjudgement]

Actually, I just found my original notes in the program.

It produced its output in a different format, and it's possible I made mistakes translating it into the one seen here. It looks like it also only identified candidates for non-brokenness that I had to filter by hand. Now I'm intrigued to see if I can make a more automated version to remove the possibility of human error (especially as the format it was using was fairly unintuitive).

[callforjudgement]

I just ran the program again and used a more automated way to produce the output.

@BTD6_maker

: care to check the following setups for me?

Code: Select all

1234ABCDE
=========

AAA324433
ABBB11441
ABB114441

AA4324433
ABBB14441

ABCB14331
ABCC12421

AAA344433

AA4344433
AAA334433
AAAA44433

ABCC12421
ABCB12331

ABCC12221
ABBB14331
ABB113441

AA4A34433

If there are any problems here, there's something wrong with the program.
Otherwise, I guess 2012-me was either overconfident or having a bad day.

[BTD6_maker]

I just ran the program again and used a more automated way to produce the output.

@BTD6_maker

: care to check the following setups for me?

Code: Select all

1234ABCDE
=========

AAA324433
ABBB11441
ABB114441

AA4324433
ABBB14441

ABCB14331
ABCC12421

AAA344433

AA4344433
AAA334433
AAAA44433

ABCC12421
ABCB12331

ABCC12221
ABBB14331
ABB113441

AA4A34433

If there are any problems here, there's something wrong with the program.
Otherwise, I guess 2012-me was either overconfident or having a bad day.

None of these are broken. Be careful, though, as I saw ABCC12421 twice.

[callforjudgement]

OK. So it seems that back in 2012 I tried to change the numbers round manually to use 1s and 2s in preference to 3s and 4s and screwed up.
I guess if we agree that this list is unbroken, it's the one we can use in the future (with duplicates removed). It wouldn't surprise me if many of the list entries were conceptually identical and the program didn't notice.

[BTD6_maker]

If you or Zoraster ever run another one of these setups or something similar (and I have the right permissions) I can be your co-mod.

[zoraster]

You're welcome to run it yourself. I'm not sure a 9p game really warrants a co-mod.

[LicketyQuickety]

I have a game for you to look at @BTD6_Maker. Someone else was supposed to look at it but I never heard from them, so do you think you can do it?

[Radja]

[Radja]

Bump