The Out-of-Order Corollary

[Zamiell]

This is not a new convention. It is the way that we have already been playing for the last 4 years.

The Out-of-Order Corollary

• First, see the section on the Out-of-Order Play Clue.
• As specified in the sections above, if Alice gives a clue to Cathy touching the next playable card, but the focus is wrong, Bob would normally give a OOO Fix Clue instead of blind-playing a card.
• Thus, if Alice gives a clue to Cathy, and Bob does blind-play a card, then by default, Cathy can mark all the cards that were touched with a note of not being the next playable card.
• For example, in a 3-player game:
  • Red 2 is played on the stacks.
  • Alice clues red to Cathy, touching two cards on slot 1 and slot 2.
  • Bob blind-plays a blue 1 and it successfully plays on the stacks.
  • Cathy marks her slot 1 card as the red 4. (Cathy knows that Alice performed a Bluff and slot 1 was the focus of the clue.)
  • Cathy also marks her slot 2 card as the red 5. (Cathy knows that if it was a red 3, then Bob would be forced to give a Fix Clue.)

[jack67889]

Which convention would make Bob blind-play finesse in the example of Jake and what would Cathy write on her cards?

[Zamiell]

Which convention would make Bob blind-play finesse in the example of Jake

a finesse would, so cathy should write r5 on chop and kt as the other card

[jack67889]

So Bob should think it's a finesse and he should not fix if he can see the clue could be given without an OOO play clue for the same efficiency?

[Zamiell]

Why is it a known bluff? Bob is thinking that he's playing the other copy of red 4, it originally looks like a normal finesse I've seen before. Only after he plays a card that's not red 4, it becomes a bluff (and then perhaps good touch applies to Cathy's slot 1 or not, I guess, depending on the result of this discussion).

we covered that already in the convention-questions channel, but to reiterate:

• in most games, there is a rainbow suit, so bob wouldn't typically be able to play r4 because then the other copy of r4 would be marked as a rainbow card, which would be a lie.
• in the end-game, this is no longer an issue, so a situation where Alice has Free Choice would result in Bob playing r4 as an End-Game Distribution Finesse.
• the end game is the most common situation where bob would play r4, but it could also happen in e.g. mid-game where there is no rainbow suit and Alice clearly has a Free Choice

however, notice that the End-Game Distribution Finesse does not conflict with The Out-of-Order Corollary. in this situation, it is of no consequence whether or not alice marks her r4 as r1/r2/r3 or r1/r2/r3/r4, because it would be trash regardless (because the r4 is on the stack)

[pianoblook]

There are plenty of undefined situations that involve a >50% correlation between a certain action and a certain underlying state regarding your hand.

To throw a random example out there, let's say Alice Chop Moves a red 2 in Cathy's hand while there are other clues to be given. Playing as Bob, from my experience I will now lightly assume that I probably don't have a r1 in my hand. At least for the sake of my argument, let's say that is probably true around 80% of the time.

Does that mean we should write up and add a Chop Move Corollary prescribing Bob to write Negative Elimination Notes for red 1 across his hand because of this action? I certainly hope not, because there's zero benefit. **Edit: And this would lead to incorrect play, importantly. It's just an incorrect assumption for Bob to make.

But uh oh, "the job of the doc is to describe the most common situations that come up, so I have typed up this section" - so I guess we need that Chop Move Corollary too, huh?

[Zamiell]

So Bob should think it's a finesse and he should not fix if he can see the clue could be given without an OOO play clue for the same efficiency?

once again, this thread is not about prescribing any behavior for bob, only cathy. don't ask me what bob should think in some contrived, undefined situation. do that in the #convention-questions channel or in some other thread.

[jack67889]

I'm sorry but I try to grasp your statements, this example resulted in this issue and you are describing in the issue what Bob should do by default. I didn't understand yet what value you see in this section. And why would it be contrived??

[Zamiell]

I'm sorry but I try to grasp your statements, this example resulted in this issue and you are describing in the issue what Bob should do by default. I didn't understand yet what value you see in this section.

no problem. the value I see in the section is it clarifies what the common situation is for Cathy, after a clue like this happens: https://hanabi.github.io/docs/level-12/#example-1---color-disconnect-valid (but with the modification of where Cathy has two or more cards touched instead of one)

[Zamiell]

Does that mean we should write up and add a Chop Move Corollary prescribing Bob to write Negative Elimination Notes for red 1 across his hand because of this action? I certainly hope not, because there's zero benefit.

this is a good point. some sections in the doc describe "conventions" (or "moves", if you prefer). other sections in the doc are "clarifications" on how conventions interact.

if a section called Chop Move Corollary did exist, it would be a "clarification" section. but the clarification in this section would not be very useful: 1) the situation happens a medium-ish amount of the time 2) the situation is super obvious, you probably don't need a section telling you this 3) the situation "is only true" 80% of the time, and that isn't a very high number

based on these three points, the section would feel more like "bloat". so i agree with you that a section called the Chop Move Corollary isn't necessary.

on the other hand, The Out-of-Order Corollary describes: 1) a situation that happens extremely frequently (after every single Bluff that touches two or more cards) 2) a situation that is probably not obvious, at least based on how much discussion has been around this in the past few days 3) a situation "is only true" 99% of the time, and 99% is an extremely high number (remember that for the purposes of this thread we must not consider undefined moves; see my previous reply)

thus, the calculus is different, and I assert that the section does provide value, at least more than a Chop Move Corollary would

[sjdrodge]

that's fine. again, this thread isn't about bob, it is about cathy. if you want to prescribe something new for bob, please open a new issue!

You keep saying this, but other participants in the thread keep pointing out that you're prescribing behavior for Bob by saying Bob ought to give a Fix Clue instead of blind-playing in the example scenario. The reality is that Bob plays expecting to play r4.

In games, I see people get Bob's r4 in exactly this way with some frequency, but I very rarely see anyone use this well-defined behavior to bluff Bob. Perhaps that is where our disconnect is, if your comments about frequency and well-definedness are specifically about the bluff case.

In my mind, the major important point is that we should not add language to the document that would suggest to anyone that it's unusual to get Bob to play r4 in this fashion, or that would suggest that Bob should give a fix instead of blind-playing. That would be a major change to the way we have historically played, and also rather bad.

[sjdrodge]

mandating something weird like this

piano, I would not frame it that way. the way that I would frame it is like this:

piano - "Hey Zamiel, I read your new issue thread and I am very surprised by it. In the past, I have performed one or more Faulty Bluffs, and they have worked great! If your new convention is accepted, then Faulty Bluffs would not work anymore. Thus, this convention seems problematic, so I think it should be modified and/or deleted as to not interfere with Faulty Bluffs."

Zamiel - "Um, piano, the Faulty Bluff doesn't actually exist. Show me the section in the document where it says that a Faulty Bluff legal."

piano - [looks through the document] "Oh, you're right. That's strange, Faulty Bluffs are not in the document. That's weird, as it seems like a simple enough move. As I've said, I've done them before and they work great! And if I've done them, probably other people have done them too. So for the purposes of this thread, we should probably consider them as existing."

Zamiel - "If Faulty Bluffs are a common move, then lets create an issue for them and create a section for them in the doc. When the new section gets added, I'll completely delete the section called The Out-of-Order Corollary, since it will no longer apply. Furthermore, I'll add a bullet point to the OOO section that says: Remember that sometimes, clues that look like OOO are not OOO at all and are instead *Faulty Bluffs*. Don't get them confused!

Zamiel - "However, as it stands right now, that section doesn't exist. So no, for the purposes of this thread, we should not consider them as existing. For the purposes of this thread, The Out-of-Order Corollary follows from the existing conventions as written in a purely logical manner. Meaning that this isn't "my" convention - it just logically follows from what we have already written. Furthermore, for the purposes of this thread, we must not consider what "rogue" players are deciding to do on their own separate from the official moves and the official framework."

If this is really how you view this thread, what you're saying makes a lot of sense. To the rest of us it looks a little bit more like this:

Zamiel: "I'm going to add some language to the OOO section that says Bob should never blind-play from OOO-looking play clues and should instead always fix them." Everyone: "WTF? why?" Zamiel: "This is how we already play, if you want Bob to blind-play in these scenarios, you have to open an issue." Everyone: "Wait what? We all give true finesses in this way with some frequency, and such behavior is supported by the current document, let's not change this, please" Zamiel: "This change that says Bob should fix doesn't actually mean Bob should fix, open an issue if you want to talk about Bob." Everyone: "???????????????????????????"

[Zamiell]

In my mind, the major important point is that we should not add language to the document that would suggest to anyone that it's unusual to get Bob to play r4 in this fashion, or that would suggest that Bob should give a fix instead of blind-playing.

i agree that this section should not interfere with Bob assuming an End-Game Distribution Finesse.

as I've stated previously, I believe that this section is only describing what cathy should think - bob isn't involved. if you want to suggest some edits to make that more explicit, then feel free.

Zamiel: "This change that says Bob should fix doesn't actually mean Bob should fix, open an issue if you want to talk about Bob."

please show me where it says that bob should fix.

[sjdrodge]

i agree that this section should not interfere with Bob assuming an End-Game Distribution Finesse.

The end-game has nothing to do with it. This move can be performed at any point in the game.

do I sound like a broken record yet?

Yes, please consider the possibility that it's not everybody else in this thread who is going crazy.

Zamiel: "This change that says Bob should fix doesn't actually mean Bob should fix, open an issue if you want to talk about Bob."

please show me where it says that bob should fix.

Both myself and others have already done this several times in the above thread by quoting your proposed change directly.

At this point I think it's unlikely that you'll understand the majority position, so you'll simply have to accept that you're outvoted here.

[Zamiell]

The end-game has nothing to do with it. This move can be performed at any point in the game.

only partially agree, read my response to floriman

quoting your proposed change directly

i just scrolled through the issue briefly and it looks like you haven't quoted any of the section?

[sjdrodge]

i just scrolled through the issue briefly and it looks like you haven't quoted any of the section?

https://github.com/hanabi/hanabi.github.io/issues/620#issuecomment-804662503

[Zamiell]

what you've quoted is out of date and does not apply

[sjdrodge]

Yes, I haven't quoted version two of the proposal, but others have.

[Zamiell]

please re-read the thread from start to finish. i maintain that nothing in the The Out-of-Order Corollary section says that bob should give a fix clue in the hypothetical where it looks like an End-Game Distribution Finesse is happening. if you think it does, please suggest an edit or provide a pull request

[0nen]

I agree with the majority here.

[sjdrodge]

Wait hold on. Are you trying to gaslight me or something? The offending phrase that I quoted at the start of the thread remains in the second version of the proposal almost unedited: "As specified in the sections above, if Alice gives a clue to Cathy touching the next playable card, but the focus is wrong, Bob would normally give a OOO Fix Clue instead of blind-playing a card."

[Zamiell]

"As specified in the sections above, if Alice gives a clue to Cathy touching the next playable card, but the focus is wrong, Bob would normally give a OOO Fix Clue instead of blind-playing a card."

this sentence is meant to be a descriptive statement, not a prescriptive one.

[Zamiell]

i.e. "here's a thing that has happened, now what should cathy think?"

[pianoblook]

Thus, if Alice gives a clue to Cathy, and Bob does blind-play a card, then by default, Cathy can mark all the cards that were touched with a note of not being the next playable card.

I maintain that this sentence is just patently false. In what world does this "default assumption" actually exist? In what world does this assumption have positive value to prescribe to Cathy? This just seems bad, and wrong, and not helpful under any circumstance

[pianoblook]

but at this point I don't think there's anything else we can say, other than a strong Downvote for this proposal

[Zamiell]

In what world does this "default assumption" actually exist?

• For example, in a 3-player game:
  • Red 2 is played on the stacks.
  • Alice clues red to Cathy, touching a red 4 on slot 1 and a red 3 on slot 2.
  • Bob sees that this is a weird clue, because Alice has a Free Choice between cluing red and cluing 3. Alice is intentionally focusing the wrong card.
  • Bob blind-plays his Finesse Position as an Undefined Move. It is a blue 1 and it successfully plays on the stacks.
  • Cathy marks her slot 1 card as the red 4. (Cathy knows that Alice performed a Bluff and slot 1 was the focus of the clue.)

piano do you agree that the document does not specify this move, and that it is illegal? by default, cathy should never assume that this is a move that has happened. if you disagree, then we should get the move defined and added to the document. this is an important building block for understanding what i am outlining here

[0nen]

Level 10. The Bluff.

In this situation, players can perform an alternate strategy by fibbing to the next player. By cluing an unrelated, one-away-from-playable card, the next player will think that it is a Finesse and that they have the connecting card. Then, they will blind-play their Finesse Position card. This is called a Bluff.

It doesn't say you can't clue more than one card with a bluff. The focus was one away from playable, so it's a bluff. Simple.

[Zamiell]

It doesn't say you can't clue more than one card with a bluff. The focus was one away from playable, so it's a bluff. Simple.

it's not simple at all. if bob played red 3, then red 3 would be duplicated, which would be illegal. thus, bob would not blind play.

[timotree3]

In the hypothetical with red 2 on the stacks it's true that Bob does not believe it is a finesse, but that does not prevent it from being a bluff.

Known Bluffs

- Usually, when a Bluff occurs, the blind-playing player has no idea that is a Bluff - they assume they are playing a specific card into a true Finesse.
- However, in rare situations, a clue will be given that looks like a Finesse, but the next player will know for sure that they do not have the "connecting" card.
- In these situations, if the next player cannot see a better explanation for the clue, then they should blind-play their Finesse Position card as a Known Bluff.

[0nen]

It doesn't say you can't clue more than one card with a bluff. The focus was one away from playable, so it's a bluff. Simple.

it's not simple at all. if bob played red 3, then red 3 would be duplicated, which would be illegal. thus, bob would not blind play.

Why can't it just be simply known that you are being bluffed and it would not duplicate? Or if there is duplication it would be known trash due to other cards on the board. We wouldn't bad touch if it can be avoided

[Zamiell]

In the hypothetical with red 2 on the stacks it's true that Bob does not believe it is a finesse, but that does not prevent it from being a bluff.

agreed. but a known bluff in exactly this case is extremely rare. thus, cathy should not consider this possibility, and should write a note of exactly red 5 on her slot 2 card. (this is historically how we have played)

[Zamiell]

Why can't it just be simply known that you are being bluffed and it would not duplicate?

are you talking about a known bluff?

[0nen]

Why can't it just be simply known that you are being bluffed and it would not duplicate?

are you talking about a known bluff?

No, I'm saying it would either be a bluff, or if it duplicates then the other person knows their extra card is trash. So not known bluff.

[timotree3]

If you read the third bullet point in my Known Bluffs quote, it states that Bob should blind-play finesse position if there is not a better explanation for the clue. Therefore unless there is a better explanation for the clue, it is a perfectly conventional and defined clue as a known bluff. @Zamiell, do you assert that there is a better explanation for the clue in the hypo with r2 on the stacks? If so, which explanation, and why is it better than a known bluff?

[Zamiell]

@Zamiell, do you assert that there is a better explanation for the clue in the hypo with r2 on the stacks? If so, which explanation, and why is it better than a known bluff?

Known Bluffs are more commonly performed when e.g.

• you are supposed to be playing r3, and you see one copy of r3 in the trash and the other copy of r3 in someone elses hand
• you are supposed to be playing r3 and the card you are blind-playing has negative red

but yes you are correct in that a known bluff technically could/should apply here

[Zamiell]

with that said, i maintain that:

cathy should not consider this possibility, and should write a note of exactly red 5 on her slot 2 card. (this is historically how we have played)

[timotree3]

If you agree that a known bluff should apply, do you still assert that without a new convention of Faulty Bluffs, such clues are undefined?

[sjdrodge]

I think maybe the major confusion that you're having here Zamiel is how different the OOO Finesse scenario is from the scenario(s) we're talking about in this thread. With the OOO Finesse, there's real tension between interpreting it as a Known Bluff and an OOO Finesse, and I'm pretty sure the prescription is OOO Finesse by default. For example, Cathy has r3 b2 r1 y5 and Alice clues red to Cathy. Bob should think he's supposed to fix and play r2 later, not blind-play now and have r1 prompted later.

In the scenarios in this thread, Bob's alternatives are not blind-play now or blind-play later, Bob's alternatives are blind-play and no fix required, or give a fix and never blind-play. The latter is strongly counter-indicated when Alice could've just given a straightforward clue that didn't require a fix. Myself and others are shocked to hear that you believe the reasoning from the OOO Finesse scenario overrides the consideration that Alice shouldn't give lies for no reason.

[Zamiell]

do you still assert that without a new convention of Faulty Bluffs, such clues are undefined?

no. allow me to change my position:

i'll modify the hypothetical:

• For example, in a 3-player game:
  • Red 2 is played on the stacks.
  • Alice clues red to Cathy, touching a red 4 on slot 1 and a red 3 on slot 2.
  • Bob sees that this is a weird clue, because Alice has a Free Choice between cluing red and cluing number 3. Bob sees that Alice is intentionally focusing the wrong card.
  • Bob blind-plays his Finesse Position as a Known Bluff. It is a blue 1 and it successfully plays on the stacks.
  • Cathy marks her slot 1 card as the red 4. (Cathy knows that Alice performed a Bluff and slot 1 was the focus of the clue.)
  • Cathy is unsure whether to mark her slot 2 card as [red 5] or [red 3, red 5]. She considers both possibilities:
  • If there are blocking cards in her hand, Alice might clue red to Cathy intending for Bob to give a follow up OOO Fix Clue. If this was the case, then her slot 2 card is probably not red 3, or else Bob would have given a OOO Fix Clue. Thus it would only be a [red 5].
  • If there are no blocking cards in her hand, it is possible that Alice performed a Known Bluff by using Free Choice. If this was the case, then her slot 2 card could be a red 3. But it would also be possible to be a red 5. Thus, it would be [red 3, red 5].
  • Cathy knows that Known Bluffs using Free Choice are extremely rare. As an unstated convention that is not officially in the document (but learned through experience in playing with the group), Cathy knows that she should disregard this rare possibility, and instead just simply write a note of [red 5].
  • Since Cathy wrote the wrong note on one of her cards, Alice has performed a Lie to Cathy. Alice should not have given this clue.

[timotree3]

I'm relieved by your new position. :smiley: It makes much more sense to me than the previous ones. I think it's actually an interesting question which note Cathy should write. I used to write that r5 note in Cathy's position but I'm not sure if I still would.

[sjdrodge]

This hypothetical and the hypothetical offered in my comment at the top of the thread as essentially the same, to my mind. I would expect Cathy to maintain the full superposition unless context collapses it, just as I would expect Cathy to assume Good Touch Principle instead of marking the r4 as trash in the first example. That said, I would also be wary that a mistake/misunderstanding is possible in these scenarios since they are edge cases that don't come up a lot. I believe that is the way that most people currently play, and I'm pretty sure the responses in this thread support that belief.

Let's not change the document in such a way that enshrines what many of us to believe to be the worse option, which is Cathy eliminating a possible card from the superposition without enough evidence to demonstrate that it ought to be eliminated, and let's definitely not change the document in any way that would suggest Bob shouldn't blind-play in the first scenario.

[Zamiell]

closing this in favor of a new issue

[Zamiell]

/deny

[conventions-bot[bot]]

• Some time has passed since this issue was opened and the group appears to have reached a consensus.
• ❌ This change will not be integrated into the official reference document.
• This issue will now be closed. If you feel this was an error, feel free to continue the discussion and a moderator will re-open the issue.

(For more information on how consensus is determined, please read the Convention Changes document.)

[florian-5f3759df]

Edit: Never mind, the issue was closed.

What I think might be a productive example to get to the disagreement:

No variant, r3, g1 played Scenario A: Bob: b1 - x - x - x Cathy: r4 - g1 - g1 - r5 Alice clues red to Cathy. Since it seems much cleaner for Alice to clue 4 or 5 and let Alice clue red, Bob attempts to play the other copy of r4.

Scenario B: Bob: b1 - x - x - x Cathy: r1 - g1 - g1 - r5 Alice clues red to Cathy. Bob plays b1, attempting to play r4.

How I thought I understood the implications of the conventions, but apparently was wrong: Cathy can play the extra card by good touch. In Scenario A, this will work out well. In Scenario B, Cathy will bomb r1 as r4, so this was a clue that's illegal as a bluff, and the clue was only allowed if it was the truth.

What the consequences of the proposal sound like to me: Cathy is supposed to discard the extra red card, and therefore Scenario B is a legal bluff - Cathy will simply toss r1. The clue in Scenario A is illegal, Alice can only give this clue if it's the truth.

[0nen]

do you still assert that without a new convention of Faulty Bluffs, such clues are undefined?

no. allow me to change my position:

i'll modify the hypothetical:

• For example, in a 3-player game:

  • Red 2 is played on the stacks.
  • Alice clues red to Cathy, touching a red 4 on slot 1 and a red 3 on slot 2.
  • Bob sees that this is a weird clue, because Alice has a Free Choice between cluing red and cluing number 3. Bob sees that Alice is intentionally focusing the wrong card.
  • Bob blind-plays his Finesse Position as a Known Bluff. It is a blue 1 and it successfully plays on the stacks.

I know this is closed, but I just want to point out:

If Bob plays finesse it could still be r3. This would provide extra info: r5 is definitely in someone's hand, and Cathy has a known trash red card. Any player that doesn't see r5 would put elimination notes on their hand for it.

tl;dr It isn't 100% a Known Bluff, it's just very likely to be one.

[dobi815]

Any assumption to fix someone else's clue is dangerous. e.g. of bomb from just today https://hanab.live/replay/478480#4

The driver ofc. is the truth is visible in front of kopen's eyes and they are deciding between fixing MangoPie cleanly vs. blind playing but lanvin cannot tell the difference.

Fixing must be the exception and cannot be put ahead of blind playing. Its not how I play