Trash Rank Trash Clue about hanabi.github.io HOT 19 CLOSED

• In a game with rank-wild cards (pink or omni suits or multi-fives), when a rank clue is given for 1s or 2s, and all cards of that rank have already been played or are clued (making the entire rank trash), then unless it is a valid 2 Save, it is interpreted as a clue to trash.
• However, only the rightmost card which wasn't already known to be trash is promised to actually be trash. The remaining cards may not be, even if the convention being used normally promises that. This rule strikes a balance between allowing powerful trash clues to be used without rank-wild cards interfering too much by causing all potential trash clues to be interpreted as play clues, while also preventing the recipient of a trash clue from discarding non-trash cards.
  ** Example: Bob has pink 5 in slot 2 and pink 2 in slot 3. All 1s have been played, and pink has only been played up to rank 3. Alice clues 1s to Bob, hitting both of the pink cards. Because 1 is a trash rank, Bob interprets the clue as being to trash and is therefore a Trash Chop Move. He chop moves himself and marks the pink 2 as trash. He does not mark the pink 5 as trash because only the rightmost card is promised as trash.
  ** However, if Bob already knew that the pink 2 was trash, for instance because it had a 2 clue and all 2s had been played already, then this clue would promise that the pink 5 was also trash, and thus be illegal.

from hanabi.github.io.

• A Fix Clue can be a Trash Rank Trash Clue. In this case, the promise of trash on the rightmost card still applies independently of the fix. The fixed card, if not the rightmost clued card, is not promised to be trash, but the rightmost clued card is promised to be trash even if a different card is being fixed. If the rightmost card is not trash, the clue is illegal.
• Trash Rank Trash Clues do not apply to ranks which are not entirely trash. Thus, if there are any 1s yet to be played or clued, a clue to a single trash 1 is interpreted normally, such as as a Trash Bluff or Unknown Trash Ejection. Thus a Trash Bluff still promises that all touched cards are trash.
• It is possible to lie and give a Trash Rank Trash Clue when the rightmost card is not trash. Such a lie will require a fix but may be useful, such as to Trash Push a difficult card.
• A 2 Save being validly used to save a non-2 takes priority over a Trash Rank Trash Clue interpretation, however, if the recipient knows that it cannot be save clue and that the clue giver knows this (for instance because all remaining 5s are in other players' hands in a multi-fives game, or because the Pink Promise prevents a 2s clue from being used to save a higher-ranked pink card), then it cannot be a 2 Save so it is a Trash Rank Trash Clue.

from hanabi.github.io.

** Example: Bob has pink 5 in slot 2 and pink 2 in slot 3. All 1s have been played, and pink has only been played up to rank 3. Alice clues 1s to Bob, hitting both of the pink cards

the example given seems pretty weird.

firstly, you dont include discussion of pink choice, which doesn't apply here because slot 1 is not a pink card.

secondly, the current configuration is that it would just be a fix clue on the leftmost pink card. from what i can gather you are proposing that all pink fix clues are also chop moves? that doesn't seem to flow from first principles. fix clues on other types of cards are not chop moves, so i dont see why fix clues on pink cards should be chop moves.

thirdly, the way we have been already playing is that leftmost is trash, not rightmost. why specifically is it optimal to promise rightmost when prompts call cards from left to right? i think that most of the time, the whole point of fixing a pink card is to get a prompt to work again.

fourth, for practical purposes we don't typically pink fix because we can just use pink choice to get a rightmost card, it is just a more direct way of solving the problem

from hanabi.github.io.

vetoed for now i guess since alercah never responded to my concerns

from hanabi.github.io.

NoMercyYesterday at 3:12 PM
@alercah I could not follow the logic of the Fix Clue section of trash rank trash clue proposal. I think the first section has merit to discuss.

Why are you proposing the clue promise rightmost be trash? That seems counter intuitive to the clue focus being the promised trash and typically for trash clues that would be leftmost.

from hanabi.github.io.

alercahYesterday at 3:14 PM
I forget about that detail
I'd have to go look at it again
oh right
because it goes with a TCM
where the key trash card is the rightmost, since it's the one to the left of the CMed cards
e.g. if you have p2-p5-x-x it feels weird for a TCM by a 1s clue to say that the p5 isn't trash, despite trash moving the two rightmost cards
but I don't think there would be a problem with doing it the other way also
it just becomes kind of wonky because then you can't TCM with p5-p2-x-x
but it is inconsistent with UTD, so the consistency might be better
for the fix section, here's the example:

y1 needs fixing because there's a layered gentleman's discard to g2
and he's about to play it
if I clue 1s, what am I promising is trash?
My proposal is that the same convention applies
in this case, it would promise only y1 being trash
under the proposal as written though, it would also work if the fix was on g5
as it wouldn't promise that the g5 was trash, but only that y1 was
if we go with a focus-based rule
then it would work for fixing y1, since it says that only y1 is trash
it would not work for fixing g5, since it would signal that g5 is trash when it's not
note: this is multi-5s variant, where 5s cannot be clued directly, but are hit by all rank clues
the final alternative would be that there's no trash promised at all
but then micerang has two clued, trash 1s that he can't discard
because they might actually be 5s

from hanabi.github.io.

NoMercyYesterday at 4:39 PM
@alercah. Understood on the fix clue example. This seems consistent with the way a color fix clue would work in a novariant game.

The arguement of p5 p2 X X seems to me mirrored with p2 p5 X X, so unless there is a strategic reason rightmost is clearly superior wihch im not seeing then clue focus left most is preferred.

Seems to me the kernel of the arguement is less about left/right and more that indirectly number clued cards that could be their touched number and thus trash, or could not be and are useful pink/omni cards. I think the later is how we currently play....?

from hanabi.github.io.

alercahYesterday at 5:25 PM
Yeah.
Basically the logic for having it at all is that you can use another rank to give a play clue
usually
because if all the ones are down, TCM is a powerful move to have, but you can't have it if you just assume that it's a play clue on pink
I proposed it working on 1s and 2s only, to ensure that there's always "room" to make play clues
1s only might work too
actually
trash on focus is the better convention then now that I think about it because you can't play clue
if you have p4 p2 X X, you can just play clue to save the cards
if you have p2 p4 X X, you can't, and you can't clue the 4 to play at all atm
so having a 1s clue be a TCM promising that p2 is trash is ideal I think
the reason I think fix clues should promise 1 trash is to still be able to make progress on fix clues by indicating trash
in retrospect, I think the y1 is bad there because it bad touches r1 even with the proposed convention
but it might be fine if it's useful to clue again to indicate that g5 is a 5, which would also fix the r1
I think that promising trash on the fixed card might be the easier convention to remember
since the fixed card is effectively the focus
and then it lines up with UTD too

from hanabi.github.io.

alercahYesterday at 8:10 PM
@NoMercy I think my main remaining concern with the fix case is whether there's still good/bad touch on the trash cards
my feeling is that because fix clues don't promise good touch, it's allowed to make bad touches which will get disambiguated by a different rank clue later
but the question I guess is how strong that should be
like is it the usual "don't make a bad touch if you can avoid it" or should it be more lenient?
given that the bad touch is fixed by any rank clue which could easily serve another purpose

from hanabi.github.io.

StephenYesterday at 8:56 PM
fix clues do promise good touch afaik

alercahYesterday at 9:06 PM
I thought they didn't because bad touch might be the only way to fix

StephenYesterday at 9:06 PM
yo, I heard you like fix clues so...

NoMercyYesterday at 9:23 PM
They do - to a point. Keep in mind omni and pink alread have relatively relaxed reads on good touch principle considering increased difficultly in adhereing to it without violating save principle.

from hanabi.github.io.

Those first chat quotes are from discussion of a different proposal.

After the discussion with NoMercy, I think the convention should be that only the focus is promised trash, and the rest of the cards may or may not be.

from hanabi.github.io.

which chat quotes

from hanabi.github.io.

#68 (comment) and the one after.

from hanabi.github.io.

ok i deleted those comments

from hanabi.github.io.

The way we currently play is as follows:

In a 3-player game:

• Pink 2 is played on the stacks.
• Bob has a pink 1 on slot 4 and a pink 4 on slot 5.
• Alice clues number 1 to Bob, touching slot 4 and slot 5.
• Bob knows that normally, when pink cards are re-clued with rank, it signifies Pink Choice, and that would mean that he has to play his slot 1.
• However, Bob also knows that Pink Choice can not be used to perform a Self-Finesse. Thus, this must be a Fix Clue, and his left-most pink card must be a pink 1 in order to match the 1 clue. (It is also possible for the card to be pink 2 - the main point is that it is a known trash card.)
• Bob discards his slot 4 card.

Does your convention change this example at all?

from hanabi.github.io.

No, this example is in line with the convention: the focused card is promised trash, but no other card is.

from hanabi.github.io.

Example:

• All 1s have been played, and pink has only been played up to rank 3.
• Bob has pink 5 in slot 2 and pink 2 in slot 3.
• Alice clues 1s to Bob, hitting both of the pink cards
• Because 1 is a trash rank, Bob interprets the clue as being to trash and is therefore a Trash Chop Move. He chop moves himself and marks the pink 2 as trash.

Can you explain why Bob interprets number 1 as a Fix Clue and a Chop Move in this example but Bob interprets number 1 as just a Fix Clue in my example?

from hanabi.github.io.

closing this since alercah isn't responding
/deny

from hanabi.github.io.

• 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; feel free to re-open it if you wish to continue the discussion.

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

from hanabi.github.io.