I don't play hanabi anymore, so others will tell you what you should do in bga.
I will try to do a complete post, so it will be long, but there should be many interesting things, included in which conventions you wanna use. And I will start the most basic possible (I will still not explain bluffs
).
While I really enjoy reverse bluff, they are not good.
Order of players A>B>C>D. A is playing next.
First part: reverse bluff (part /A)
If I take a normal situation (situation #1):
B: x-r2-x-x
C: g1-x-x-x
If A clue red on B, and B need to take into consideration that this could be a reverse bluff, then if B got in reality r1, then not only r1 will be played the following turn, you make B has worst options on his turn because he doesn't know if he got r1/r2, if g1 is gotten or not, and he can't use the timing of g1 in finesse position to bluff someone. Which is obviously atrocious.
So in all traditional conventions sets I know, they are not allowed.
If I add something great to do for B (situation #2):
B: x-r2-(b1)-x
C: g1-x-x-x
So b1 is globally known. Here from A's pov, B can play, which make the move much better. Still, there is lots of annoying parts. Let's assume that A don't see any good target in D's hand for a bluff. Then A clue red on B. So B here without reverse bluff could give some informations about which card to play first between r1 and b1. But if you add reverse bluff you lose that information.
In bga, there is no priority on which card to play first.
In hgroup, (and something funny the priority goes from level 8 i guess to the last level 25 so that show that assuming things with them lead to lots of mistakes/need more focus/or need better players) you are forced to play your cards in a certain way, if you respect it nothing happen. If you don't, that mean that you see the next card somewhere on the board, and so that player can write elimination notes on that card (and should save it if the possibilities are reduced to one single card, since it's a globally known information). If that card isn't visible from your next player's pov, that player should play his finesse position as a priority finesse (prompt first as usual). If the card played is good but is not what you expected it was in fact a priority bluff, and nothing else happen: you get one card for free (and others players got negative of the promessed card if they want to optimise).
But while on bga the cost isn't that high, if you plan to play with a smaller group, just removing one possibility of prio remove a lot of options you got when deciding of conventions.
Going back to the example, if A got an excellent target for a bluff, then B can't use it if you play with reverse bluff [note that here, if B didn't knew the color of their 1 that was less a problem since getting g1 has a cost of duplicating it, but you want conventions to be the most general possible]. And if A has g2 then getting it has great tempo and should be done anyway in most cases.
Second part: known reverse bluff (part /B)
There is too the known reverse bluff: mostly either when the next player know it was a bluff (they per example got negative on the only color playable), when they will for sure not clue their next player (they per example got three 1s to play), or when the next player has no clue left.
The problem is that this kind of move is extremely confusing. The move will work, the problem is knowing if the card promessed is layered or not.
Layered finesse happen when (situation #3) A clue red on r2 to D, B discard, and C has g1-r1-x-x. C is supposed to play g1, and still look for r1 after because B was supposed to play its bluff. I don't know now if bga play with layered or not by default.
And so we have:
- not playing layered finesse + not playing known reverse bluff: which is just the default, and you got less options
- not playing layered finesse + playing known reverse bluff which work extremely well (no risk of bomb)
- playing layered finesse + not playing known reverse bluff (hgroup): which make everyone play layered when they can, and if everyone is on the same line, you don't add that extra step
- playing layered finesse + playing known reverse bluff: which add that extra step of making the person who answer the clue to decide if obne of their previous player could or not bomb the clued card. If yes: that's layered. If not, that's not layered. And that require others players to agree with the decision that player make, else there is a desynchronisation going on.
{edit: removed third part which don't look enough useful}
Forth part finally : your situation (part /D)
So finally what do I think about that move. Since no other finesse position is playable, that 5 can only be k5, so should be playable.
So you discard it, and k5 is played.
Now that convention if it exist look pretty strong.
In hgroup (without black powder), you got complicated moves about 5, which as a big summary is on situations like that is 'you focus the card on the right of the 5 and do various stuff depending if that card is playable/can be discarded/one-away/two-away/more than two away' and if there was normal clues to give or not.
On bga, I think there is nothing, especially because of black powder.
That move look pretty strong, so to remove the reverse part which is usually bad (situation #5, which doesn't exist anywhere):
B: x-x-x-x
C: k5-x-x-x
D: x-x-b5-x
Here A clue 5 to D. B discard, C play k5. D save b5.
I really like that move because that save you a free 5, and that call for b1/k5 from other's pov. I never hear about that move, so I suppose that's illegal on bga, but if some bga players want to add that move to their small group, I strongly advise to try it. That have some logic, especially bga's logic which like playing their finesse position, and isn't that complicated.
Now the main problem of that move is the number of cards leading to b5. You need to put an artificial limit.
Situation #6:
b1 is on the stacks.
B: b2-b4-x-g5
C: k5-x-x-x
D: b3-x-b5-x
Here if A clue 5 to D wanting b2+b3+b4+b5, B assume k5 is gotten and discard, that will be game losing.
That situation is pretty rare, and you usually have alternative. But still if you keep the ambiguity, that make everyone makes worst clue on those situations, which is really bad.
But that situation is commun, so you should get a k5 in finesse position really fastly. And once one k5 has been played, that's not a problem anymore.
I would suggest to put the limit at 1.5 cards (2.5 work too). So if B need to blind play one card, they should trust the long finesse. And if B need to blind play two or more cards, they should trust that move with 5s.
Then for your game, that's one step further, which add the reverse part, which I don't think should be a thing.
k5 clue on yourself, save your two chops 2 more turn. And the chance for Kyuby to get a bluff taget is pretty low (2 cards if they were playing). Kyuby got 4 plays though, but it's 3 players, so you could be getting away with it easily.
And so I agree with you.
Reverse bluffs are a lot stronger at 3 players than at 4 players (way less chance to misscomunicate), but starting to do differents things at 3 and 4 players can only lead to bombs and playing not trusting each other for that move anymore.
, make sure everyone understand the proposition before going deeper.]