Break Tank 3: Dueling at the Warcamps

[Kasimir he/him]

Guessing this means there's a floating Mark somewhere?

Are we really just hitting that badly or something? Are TDs rolling really well? Immunity? Is there a targeting pattern? As mentioned right before close, I went for Chantara over my other candidate. No dice.

Edit: JNV. Probably will change later.

Edited November 16, 2022 by Kasimir

[Chantara she/her]

11 minutes ago, Kasimir said:

As mentioned right before close, I went for Chantara over my other candidate. No dice.

No worries. I’m not offended. I  can always respect another loyal follower of the Sacred Coin.  

[+Whysper she/her]

I attacked Devotary again. Kind of a hero shot.  I actually attacked them C1 as a placeholder attack and didn't get a chance to return and see the vote and change it.

I think everyone claiming attacks is a good idea. I think that's where we'll find the lying because we'll be able to work out where the Marks should or shouldn't be and why someone claiming to attack someone didn't result in a kill or stuff like that. That's where the TDs will slip up in action claims.

[Kasimir he/him]

This makes it:

C1:

Player	Target	Success/Failure
Kas	Whysper	X
Wiz	Bookwyrm	X
Bookwyrm	Wiz	X
Whysper	Devo	X
?	Turtle	O

C2:

Player	Target	Success/Failure
Kas	Chantara	X
Wiz	Bookwyrm	X
Whysper	Devo	X
JNV	X	X

JNV didn't log in for C2, so categorically couldn't have placed any stab order. Last I checked, GMs generally don't accept doc orders.

[The Unknown Order he/him]

I forgot to kill. Was going to go Kas anyways though, so doesn't really matter. 

[NerdyAarakocra They/Them]

I'll vote for Xino again, there's no reason not to.

[Devotary of Spontaneity]

I hit TUN C1 and Bookwyrm C2.

[The Unknown Order he/him]

Who else(sh) hit the same person both cycles?

Edit: @Matrim's Dice, I can't find it anywhere else, so does the Mark take an action?

Edited November 16, 2022 by The Unknown Novel

[The Wandering Wizard he/him]

Running the numbers, if someone is stabbed six times they have about a 1.5625% chance of survival, assuming they are a training dummy.

I think, think, mind you. That you could soft clear players depending on how many times they've been stabbed. I would take this with a large grain of salt, a very large grain. Since, rebounds can't be proven, it would be based on player word and so, very, very shaky. Especially since TDs can fake claim for their teammates and get them a larger clear. I don't know, it could be very useful to try and narrow down the pool, but the TDs could be astoundingly lucky, use immunity if they haven't yet, or TDs can claim to have shoot their teammates. 

It's an idea I had, it could at least be used to calculate the chance of death a certain person has had over the entire course. Still would give us a possible very very soft clear of certain players and lead us in other directions. 

[Mat he/him]

2 hours ago, The Unknown Novel said:

Edit: @Matrim's Dice, I can't find it anywhere else, so does the Mark take an action?

Technically, but it’s the only thing TDs can do and there aren’t RBs so it doesn’t matter. But the TDs can activate their immunity and mark someone on the same cycle.

[The Wandering Wizard he/him]

3 hours ago, Kasimir said:

This makes it:

C1:

Player	Target	Success/Failure
Kas	Whysper	X
Wiz	Bookwyrm	X
Bookwyrm	Wiz	X
Whysper	Devo	X
?	Turtle	O

C2:

Player	Target	Success/Failure
Kas	Chantara	X
Wiz	Bookwyrm	X
Whysper	Devo	X
JNV	X	X

JNV didn't log in for C2, so categorically couldn't have placed any stab order. Last I checked, GMs generally don't accept doc orders.

I can crunch the numbers from what little data we have now. I'll provide a range of maximum chance of survival to minimum, since immunity could have been activated any cycle or it could have not been activated yet. 

This data is assuming the person is a TD and using the statisically appropriate values of survival.

Bookwyrm 50-25% 

Wizard 100-50%

Devo 100-50%

Whysper 100-50%

Chantara 100-50%

And that's all the claims for right now.

[The Bookwyrm he/him]

Um....Wizard.

I never know who to vote for. This game is weird.

[+Whysper she/her]

I'll vote again for

xino

[The Unknown Order he/him]

40 minutes ago, Matrim's Dice said:

Technically, but it’s the only thing TDs can do and there aren’t RBs so it doesn’t matter. But the TDs can activate their immunity and mark someone on the same cycle.

That's what I was wondering, thank you. 

Also what in the world is going on there Wiz?

[The Wandering Wizard he/him]

4 minutes ago, The Unknown Novel said:

Also what in the world is going on there Wiz?

I'm trying to find the possibility of people surviving being attacked as if they were TDs.

So take Bookwyrm for example.

I attacked him twice. C1 and C2.

That would be a 25% chance of survival

BUT

The TDs could have activated immunity so that would remove a attack on him once, making it a 50% chance

So covering both possibilities gives Bookwrym a range from 25-50% of survival.

Does that make sense?

It makes sense in my head, doesn't like to transfer from it very well though.

[Shining Silhouette he/him]

Okay just reread the rules and realized that my prior understanding of the game was flawed in a couple different ways.

Also C1 my green Kas unvote was actually supposed to be a vote for him but I got confused because of the protect mech- I hadn't even voted him in the first place :P.

7 hours ago, Kasimir said:

This makes it:

I'm claiming nothing for C1 (I still forgot ) and JNV for C2

TBD for this cycle

[JNV]

Hi I attacked Wizard C1 and um kinda obviously didnt get around to killing anyone C2 I guess my character fell asleep in the sun or something Im just going to tentatively ignore the people who survived mroe than one attack to focus my firepower elsewhere so by the looks of it those people are Wizard Bookwyrm Devotary so yeah Im not killing you this time around congrats 

Oop ninjad lets see

1 minute ago, Shining Silhouette said:

I'm claiming nothing for C1 (I still forgot ) and JNV for C2

Wow attacking a sleeping duellist foul play I cry fowl play

[Shining Silhouette he/him]

3 minutes ago, JNV said:

Wow attacking a sleeping duellist foul play I cry fowl play

Lowkey was a sleeping duellist too so it was probably not the smartest play, but I wanted to get an action in anyway. 

[Chantara she/her]

Alrighty, who to vote for this time *mutters to self * er… Whysper because….. I like the “y” in their name 

[Kasimir he/him]

If you think C1 was the cycle the Elims invoked immunity, then you're pretty much committed to the view C1 survivals are evidentially meaningless.

If we work with that supposition, then Bookwyrm looks like the guy I have some reason to want protected, on the basis of two survivals. One survival does look good for Chantara, Devo, and JNV, but not in itself decisive. I think I am okay with staying on JNV for the moment, but Bookwyrm is an option. (I tried to crunch P(V!Bookwyrm | Dual Survival) but it's just a hot mess I'm not touching because P(Dual Survival) is kayana >>

If we ignore that supposition, then Wiz also begins to look good. Some credit to TUN and Whysper as well.

Given multiple kills are possible, it's odd there's no Turtle kill claim at all, actually, since that's free Village cred. But then again, I wouldn't be surprised if the TDs are saving that quirk of the rules for closer to win point.

Edited to add:

6 hours ago, The Wandering Wizard said:

Running the numbers, if someone is stabbed six times they have about a 1.5625% chance of survival, assuming they are a training dummy.

I think, think, mind you. That you could soft clear players depending on how many times they've been stabbed. I would take this with a large grain of salt, a very large grain. Since, rebounds can't be proven, it would be based on player word and so, very, very shaky. Especially since TDs can fake claim for their teammates and get them a larger clear. I don't know, it could be very useful to try and narrow down the pool, but the TDs could be astoundingly lucky, use immunity if they haven't yet, or TDs can claim to have shoot their teammates. 

It's an idea I had, it could at least be used to calculate the chance of death a certain person has had over the entire course. Still would give us a possible very very soft clear of certain players and lead us in other directions. 

If you fake claim for your teammates, you're effectively drawing a connection between yourself and them. It's fine until the Village gets lucky because when one of you flips, your connections will be looked at. I'm sure this will be IKYKed to hell and back but it's still a consideration.

Edited November 17, 2022 by Kasimir

[The Wandering Wizard he/him]

1 minute ago, Kasimir said:

If we work with that supposition, then Bookwyrm looks like the guy I have some reason to want protected, on the basis of two survivals. One survival does look good for Chantara, Devo, and JNV, but not in itself decisive. I think I am okay with staying on JNV for the moment, but Bookwyrm is an option. (I tried to crunch P(V!Bookwyrm | Dual Survival) but it's just a hot mess I'm not touching because P(Dual Survival) is kayana >>

 

Perhaps this might help with the numbers a bit. 

Survival Rates for TDs:

1 attack - 50%

2 attacks - 25%

3 attacks - 12.5%

4 attacks - 6.25%

5 attacks - 3.125%

6 attacks - 1.5625%

7 attacks - 0.78125%

8 attacks - 0.390625%

9 attacks - 0.1953125%

10 attacks - 0.09765625%

Of course this is all successful stabs, so removing any possible attacks that were nullified by the immunity this can help us find possible TDs. So claim your kills afterwards! It'll help us try to solve this mess

5 minutes ago, Kasimir said:

If you fake claim for your teammates, you're effectively drawing a connection between yourself and them. It's fine until the Village gets lucky because when one of you flips, your connections will be looked at. I'm sure this will be IKYKed to hell and back but it's still a consideration.

I know, it's a hot mess. >>

There's always the possibility that a TD could be fake claiming a rebound against a villager, making this a greater mess of a IKYK. 

Kas, you deserve my vote for now.

---

Lehaz's respect, Keredin had. Sparred with him often, he had. Enjoyable, it was. Learned much from their bouts, Lehaz had. Muttering to himself. "Always, more to learn, there is." On the other side eight Sades Slayers walked out. Lehaz grimly looks at their numbers and he speaks boldly "Trap it is. Always with this one. Courage we must muster, quickly! Upon us the foe be soon."

[Kasimir he/him]

12 minutes ago, The Wandering Wizard said:

Survival Rates for TDs:

Okay, so mild digression. Anyone who dislikes this, just scroll below this section. I find this an interesting avenue and Araris probably would judge me if he saw my thoughts in the GM PM but lucky for me, he hasn't  

Theoretically what we are interested in is P(V!Bookwyrm | Dual Survival) or P(E!Bookwyrm | Dual Survival) - I don't really think it matters which one you calculate, because you just really want to be able to quantify how much more likely or how less likely the dual survivals make Bookwyrm likely to be Evil. For simplicity's sake, we are ignoring C1 with the assumption immunity was invoked.

For this, you want to invoke Bayes' Theorem: P(E!Bookwyrm | Dual Survival) = P(Dual Survival | E!bookwyrm) * P(E!Bookwyrm) / P(Dual Survival)

The main issue is that P(Dual Survival) is overall odds of dual survival, bracketing Bookwyrm's alignment. And the thing is, in a world where V!Bookwyrm wasn't marked, P(Survival) = 1. In a world where E!Bookwyrm, then it's very simply 0.25 because each time, an Elim has a 0.5 chance of dying, and each stab is independently flipped for. How to agglomerate the two, no idea, my brain is half asleep. I suppose you could do it by counting cases and then taking the number of cases in which dual survival happens, maybe? But they'd have to be a complete set of cases. I don't really know that I want to chase this rabbithole (I mean I probably do but like...Bookwyrm is one player and functionally I am better off trying to ID where else is worth stabbing rather than getting distracted.)

The other issue is how to calculate P(E!Bookwyrm.) You could argue that since we know this is 6/6/4, functionally P(E!Bookwyrm) is just 1/4. Suffice to say I'm not really sure this is the right way to calculate P(E!Bookwyrm), given that's sort of dependent on how the distro is done. But okay, maybe we can say 1/4 is a sufficient proxy value.

12 minutes ago, The Wandering Wizard said:

So claim your kills afterwards! It'll help us try to solve this mess

Agreed

Well, that and kill pattern analysis IMO. Like, knowing who hit where just helps us answer one question: is there a dispersion pattern? If there is, then the obvious place to start TD hunting if we take at face value that all survivals were Village is to just go for where people haven't been stabbing. (Of course, might result in the TDs marking there too...)

...Huh. Now this reminds me of the Bayesian models used to build search grid patterns telling SAR where to look given the squares you have already searched. Not sure how they handle error, but there's sort of that there in the background - if we assume truthfulness (lol), then where do we go given the results we currently have?

Edited to add:

12 minutes ago, The Wandering Wizard said:

I know, it's a hot mess. >>

Turtle trauma intensifies >>

On further reflection, I am okay with voting Xino as well. Want to see how things shake out before placing my final vote.

Edited November 17, 2022 by Kasimir

[The Wandering Wizard he/him]

8 minutes ago, Kasimir said:

Okay, so mild digression. Anyone who dislikes this, just scroll below this section. I find this an interesting avenue and Araris probably would judge me if he saw my thoughts in the GM PM but lucky for me, he hasn't  

Theoretically what we are interested in is P(V!Bookwyrm | Dual Survival) or P(E!Bookwyrm | Dual Survival) - I don't really think it matters which one you calculate, because you just really want to be able to quantify how much more likely or how less likely the dual survivals make Bookwyrm likely to be Evil. For simplicity's sake, we are ignoring C1 with the assumption immunity was invoked.

For this, you want to invoke Bayes' Theorem: P(E!Bookwyrm | Dual Survival) = P(Dual Survival | E!bookwyrm) * P(E!Bookwyrm) / P(Dual Survival)

The main issue is that P(Dual Survival) is overall odds of dual survival, bracketing Bookwyrm's alignment. And the thing is, in a world where V!Bookwyrm wasn't marked, P(Survival) = 1. In a world where E!Bookwyrm, then it's very simply 0.25 because each time, an Elim has a 0.5 chance of dying, and each stab is independently flipped for. How to agglomerate the two, no idea, my brain is half asleep. I suppose you could do it by counting cases and then taking the number of cases in which dual survival happens, maybe? But they'd have to be a complete set of cases. I don't really know that I want to chase this rabbithole (I mean I probably do but like...Bookwyrm is one player and functionally I am better off trying to ID where else is worth stabbing rather than getting distracted.)

The other issue is how to calculate P(E!Bookwyrm.) You could argue that since we know this is 6/6/4, functionally P(E!Bookwyrm) is just 1/4. Suffice to say I'm not really sure this is the right way to calculate P(E!Bookwyrm), given that's sort of dependent on how the distro is done. But okay, maybe we can say 1/4 is a sufficient proxy value.

I couldn't count the number of cases, too much thinking for my brain right now. 

It becomes a even larger mess when you add immunity. Tried that for Devo and I stopped trying pretty fast. Far too many possibilities for right now, and possibly any time. It's why I tried to narrow down the field to the largest possible chance of survival and the largest chance of death. Less indepth, but less time consuming and brain hurting. >>

12 minutes ago, Kasimir said:

Well, that and kill pattern analysis IMO. Like, knowing who hit where just helps us answer one question: is there a dispersion pattern? If there is, then the obvious place to start TD hunting if we take at face value that all survivals were Village is to just go for where people haven't been stabbing. (Of course, might result in the TDs marking there too...)

 

Brings us back to the problem of the IKYK, plus there's one mark hanging out somewhere...

Why haven't we tried to clear people who were targeted C2 from being marked yet?

I know that it's likely that at least someone is lying, but it could help narrow down the field of marked individuals.

12 minutes ago, Kasimir said:

...Huh. Now this reminds me of the Bayesian models used to build search grid patterns telling SAR where to look given the squares you have already searched. Not sure how they handle error, but there's sort of that there in the background - if we assume truthfulness (lol), then where do we go given the results we currently have?

What is a Bayesian model? I've never heard of it before.

[Devotary of Spontaneity]

39 minutes ago, Kasimir said:

Given multiple kills are possible, it's odd there's no Turtle kill claim at all, actually, since that's free Village cred. But then again, I wouldn't be surprised if the TDs are saving that quirk of the rules for closer to win point.

If you fake claim for your teammates, you're effectively drawing a connection between yourself and them. It's fine until the Village gets lucky because when one of you flips, your connections will be looked at.

I'm guessing it was someone who hasn't posted for a while/ever, like Dannex or Misting.

And it's still early enough that attacking a duelist is unlikely to kill them.

 

[Kasimir he/him]

Just now, Devotary of Spontaneity said:

I'm guessing it was someone who hasn't posted for a while/ever, like Dannex or Misting.

I do have a guess, yeah, but I've decided it's better not to mention it. I will say I feel it's moderately unlikely to be Dannex though.

1 minute ago, Devotary of Spontaneity said:

And it's still early enough that attacking a duelist is unlikely to kill them.

True, but claiming early always gives you more credibility than pulling it out at lylo. Wouldn't really be lylo in this case but close enough.

2 minutes ago, The Wandering Wizard said:

I couldn't count the number of cases, too much thinking for my brain right now. 

It becomes a even larger mess when you add immunity. Tried that for Devo and I stopped trying pretty fast. Far too many possibilities for right now, and possibly any time. It's why I tried to narrow down the field to the largest possible chance of survival and the largest chance of death. Less indepth, but less time consuming and brain hurting. >>

Yeah. My current attitude is just ignoring the probabilities for now and taking all of this as a sign to look elsewhere. There's probably a bunch of TD lying and all but the small slice of stabs we can see is functionally close to a circular shooting gallery. I don't want to overthink this: 1/4 of this game is TDs, and there's likely at least one TD outside of the search grid, and as much as I like chilling with y'all, I would very much like to stab and enter Stabvana soon

Maybe this matters if I'm one of the remaining duelists trying to find the last two TDs, but right now, the odds are still good.

5 minutes ago, The Wandering Wizard said:

Why haven't we tried to clear people who were targeted C2 from being marked yet?

Well, I guess it could be done, but the question is really what you expect to do with it. Not stabbing a marked player does matter to us, but the real, direct/immediate wincon is stabbing a TD so I guess that's where the interest is. Knowing that none of the stabbed people were Marked C2 doesn't really help us in terms of trying to work out who the TDs might be. It's true that if you are Marked, you are not a TD, but again, we don't gain points from confirming Villagers - it's true it's good to know not to stab there, but it just seems a lot of work for little pay-off.

8 minutes ago, The Wandering Wizard said:

What is a Bayesian model? I've never heard of it before.

I'm waiting to get judged by Devo now... >>

A Bayesian model is just a model that employs Bayes' Theorem. Bayes' Theorem is considered quite useful because it tells us how to update our priors (initial probability assignments) given new evidence. There's a whole can of worms surrounding this, including big-R Internet Rationalist communities (who really misuse the actual academic term 'rationalism' but NEVERMIND I'm digressing again) - and there's actual academic debate about whether Bayes' Theorem is really that good a model of human rationality or belief revision. But in short, yeah, it's been used to build some pretty effective models for Search and Rescue, and so on.

Like, it's a rabbithole you don't have to bother with, and I think this elides a lot of the disputes, but then again, that doesn't really affect the business of finding TDs which is what we wanna do, so lesgo