Pestilent Double Bluff > Promise Double Bluff

[I Love You More Than]

Hello Forum,

I’d like to present my arguments for why Pestilent Double Bluff (PDB) is stronger, clearer, and more broadly useful than Promise Double Bluff (PrDB) and why PDB should be the default convention used on BGA in the majority of games.

Because these two conventions directly conflict, only one can be active at a time. Below I explain why PDB > PrDB in practice.

Why Pestilent Double Bluff Matters
PDB appears frequently in real games. Consider a very common situation:

A clues 2 on the green 2 to D
B plays 1y, believing it is 1g
C plays 1b to repair the situation
D now knows the 2 cannot be yellow or blue—otherwise only one of those cards would have played

When C plays to fix the broken situation, they are healing the clue and preventing D from bombing. This “healing” element is the core of Pestilent Double Bluff: a broken situation must be repaired to prevent the destination card from bombing.

This act of healing is what gives PDB both its name and its structure.

What a Double Bluff Actually Is

A double bluff is a clue that targets a destination card and allows two cards to play in sequence. Both cards played do not match the sequence to the destination card but are played without causing any bombs.

- A potentially connecting card
- A healing card

At first glance, the clue looks like a simple finesse:
1r is played on the stacks.

A clues 3 to D. To D, this presents itself as 3r being the destination card in their hand.
B plays 2r (the connecting card)
C plays 1w (the healing card)
D now knows the 3 is not red since C played to heal the situation.

This pattern is simple, efficient, and highly replicable.

Why PDB Should Be the Default

Greater frequency.
PDB is easier to use, more intuitive, and appears naturally in many board states. Situations where a player can give a PDB occur in roughly one out of every three games, making it far more common than PrDB.

Its reliability and frequency make it a strong candidate for the default double‑bluff convention on BGA.

Where PDB and PrDB Diverge

Promise Double Bluff requires significantly more conditions to be met.
Example:

A clues 3 on 3r to D
B plays 1b
C plays 1y
D, when receiving the clue, must scan the table for a 2 in the leftmost position or behind a playable card of B or C that could connect to their 3
Only if such a 2 exists can D interpret the clue as a Promise Double Bluff.

The odds of this alignment are extremely low. Yes, it does happen but rarely. And when it does, the payoff (card plays, card identity resolution) is not meaningfully greater than what PDB already provides.

So the question becomes:

Why prioritize a convention that is exponentially rarer, harder to spot, and offers little additional value?
PDB gives the same benefits while occurring far more often and with far clearer structure.

[Blacktango]

Hmm, It takes me a long time to understand both your first examples are PrDB, and the third one is PrDB (actually, I'm not even sure about that).

The real question is all about the last example: Would you allow the clue from your last example? What would that mean?

1.There is no 2, so the promise is broken. However, the clue is valuable because it allows two cards to be played. Therefore, ignoring the promise part is OK.
2. There is no 2 so the promise is broken and the clue is forbidden
3. There is a 2, and it may be in D's hand, so they will memorize it, it won't need to be saved or clued to be played later in the game.

If 1, then it is a PDB, if 2 or 3, then it is a PrDB.

If D doesn't memorize the promised 2 in their hand, forbid the clue certainly wouldn't be worth it.
If they do, then yeah, is it still worth or not?
Do you really need the PDB?

[Romain672]

In the first example, the only downside is C assuming that D see both y2 from their pov.
In the second example, it's the same thing C could assume that D could see both r3 (or more likely the only copy of r3 remaining, potentially in C's hand).
And the third example doesn't connect. It look like to me that you could clue 5 on r5 for no reason and get two blind plays too. Hgroup got 3-bluffs, that's why that clue would work there.

If you assume like hgroup that assuming copies of cards is not worth it, then first and second example become logical, and third isn't.
By the long past, I think that assuming more than 2 blind play from a clue is generally bad when thinking of conventions, that's kinda what you are doing there.

[_mg6maciej_]

3. There is a 2, and it may be in D's hand, so they will memorize it, it won't need to be saved or clued to be played later in the game.

(...)

If D doesn't memorize the promised 2 in their hand, forbid the clue certainly wouldn't be worth it.
If they do, then yeah, is it still worth or not?

I don't think it's possible for self-promise to exist on double bluff on a 2-away without additional context. His 3rd example to D should look like it's a finesse + bluff and D marks card as exactly 3b unless he sees any 2 (or 4k) somewhere else, in which case he marks it as possible 3 of that color in addition to 3b.
Additional context that would allow it to be a (valid?) self-promise is if both copies of 3b were visible both to the clue giver and the receiver, for example E had both (or one of them was in discard pile). But even if E has both or if both are in two different hands (or in discard pile, but if both 3b are in discard pile you are doing something wrong ), should the players holding these cards assume it's a valid self-promise or should they think they hold the other copy of promised 2? It's unclear to me what the answer should be, so I'd most likely avoid such an ambiguous self-promise clue and either clue that 2 directly for a simple double bluff or clue color for a single bluff and marking both 2 and 3.

EDIT: The above section is mostly right, but straight double bluffs with self-promise can exist in very specific situations. For example:
Context: 2r is played
A clues 5 to D
B plays 1g
C plays 1b
If D doesn't see anything connecting, D should think he has 4r leftmost (self-promise) and 5r marked.

Self-promise can more easily exist on triple+ bluffs or finesse + bluffs. His 2nd example is a triple bluff making 1r, 2r and 1w play. So it's known not to be 3r or 3w. If D cannot see any other 2 (or 4k) in finesse on anyone, he understands his finesse position is either 4k or 2 of one of the remaining colors.

An example with finesse + bluff and self-promise marking a 3-away:

Context: 5k is played
A clues 4 to D
B plays 1r
C plays 1w

D sees nobody has 3r or any 2 leftmost, so he should assume he is holding 3r in finesse position. If someone has any 2 leftmost (or behind just played 1r or 1w) or any unplayable 2 is marked, then to D the clue looks either as possible finesse + bluff and his 3r is self-promised or just a double bluff with 2 cards promised: the 2 in another hand and 3 of the same color as self-promise, so in both scenarios D will keep the same promised card, but only play when he knows which situation it is.

So in short, self-promise should be assumed to be a valid option if no other explanation exists, because it's a lot rarer.

[Kaznaka]

I like the idea of healing much more than promises. You can extend the logic to trash moves like trash double ignition. When player A clues a known trash on player D's chop, and the card next to the trash is also trash, player B plays because otherwise D will bomb card slot 3, and player C has to play to repair the situation. Imagine if B blind played 4w, if player C doesn't do anything, player D will bomb card slot 3 as 5w. Promises have its uses, but it shouldn't be used to limit double bluff potentials.

[springiest]

i agree with your overall premise that PDB is more useful and more common than PrDB, but i think ur 3rd example needs a bit of correction based on what cards are played before A's clue.

• if r1 (and no other card) is already played, then B blind-plays b1. amazingly, C needs to do nothing more! since b1 does not connect to a 3, its just a simple bluff on one-from-playable r3.

• if some non-red 1 (say g1) is already played, then as usual B blind-plays b1. in this case, can C do nothing? well, b1 does not connect to D's 3 as before, but r3 is not a valid bluff target. if C does nothing, D will assume they have g3 (because no other 3 is one-from-playable). after both B & C blind-play, D knows that their 3 is not g3 and not y3, but it can be any other 3 in the game.

• if BOTH r1 and some other 1 (say g1 again) are already played, then B still blind-plays b1. now can C do nothing? yes! now D notes that their 3 is either r3 or g3 only!

[Kaznaka]

i agree with your overall premise that PDB is more useful and more common than PrDB, but i think ur 3rd example needs a bit of correction based on what cards are played before A's clue.

• if r1 (and no other card) is already played, then B blind-plays b1. amazingly, C needs to do nothing more! since b1 does not connect to a 3, its just a simple bluff on one-from-playable r3.

• if some non-red 1 (say g1) is already played, then as usual B blind-plays b1. in this case, can C do nothing? well, b1 does not connect to D's 3 as before, but r3 is not a valid bluff target. if C does nothing, D will assume they have g3 (because no other 3 is one-from-playable). after both B & C blind-play, D knows that their 3 is not g3 and not y3, but it can be any other 3 in the game.

• if BOTH r1 and some other 1 (say g1 again) are already played, then B still blind-plays b1. now can C do nothing? yes! now D notes that their 3 is either r3 or g3 only!

You're forgetting D's pov. If C does nothing, D will bomb looking for b2 because he thinks his 3 is blue. The key here is that a number is used to clue, so it is intentionally ambiguous in order to make C also respond or there will be some desync in information.

[springiest]

If C does nothing, D will bomb looking for b2 because he thinks his 3 is blue.

D shouldnt think they have b3. since 1 doesnt connect to 3, D (and everybody else) should assume a bluff over a finesse if possible.

• if r1 is already played, it IS possible because r3 is one-from-playable.

• if r1 is not already played, it ISNT possible because two-from-playable r3 is not a valid bluff target.

of course, this changes drastically if bob blind-plays a 2 instead of a 1.
lets say both r1 & g1 are already played, and A gives the same 3 clue on D's r3. if the card B blind-plays is g2, then even though r3 is a valid bluff target C should still respond else D will assume they have g3. C believes they are blind-playing g3 as a Pestilent Finesse, but of course it can be an unrelated card (like y1) as a Pestilent Double Bluff.

[Kaznaka]

in BGA I can tell you 1 absolutely connects to 3. If I gave a 3 clue and a 1 blind plays, they are going to look for the connecting 2. This is why the 3 bluff, a concept from H-Group, is impossible in BGA.