Any specific black powder strategies?

[cjw29]

Hi there,

I was wondering whether there are any specific strategies in regards to black powder as you can end up with numbered clued cards clogging up a hand because these cannot get coloured... (apart from trying to use finesses where possible of course) I tried to see if the Hyphenated/Zamiel's group had any variant similar to it but they don't...

[Stroom]

I think using a color clue that marks nothing could be seen as an indication that the numbers might be black powder. But it would not always work...

[postmans]

cjw I'm quite certain Zamiels group has the variant (it was played a bit when announced, but I think didn't become one of the more popular variants) and as such also conventions on it. But they might be simply combinations of white and reversed conventions.

Main problem seems to be to safe the 1, and after that indeed to get the plays out.
For a hand such as 1b 4R 4black 5R marked as - 4 4 5 where 4 black is playable, but 4 red is not. You might agree that a red clue "reveals" the 4 black and as such might be a play clue on 4 black, rather than a finesse on 4 red. But you need the table on board

[cjw29]

So the classic 5 stall strategy used in early game... how should this be modified as it could be interpreted as play 5 in black powder? Don't do 5 stalls or wait until clued 5 again to play? but that would be possibly wasteful....

[Travis Hall]

I’ve noticed that off-chop 5 clues in the early turns have gotten much less common in the few months I’ve been playing with masters, so maybe it’s best to just recognise that they aren’t really necessary, and don’t do them.

The most important thing to recognise is that you can’t simply assume that 1s are playable by default. Context needs to be taken into account.

(A note on notation... Since we’ve long used b to denote blue, I’d suggest using p to denote the black cards - as in Black *P*owder. So, 5p, 4p, etc. The other option would be to take a cue from Magic: the Gathering, which uses U for blue and B for black, but there’s probably not that big of an overlap between the communities.)

I’d suggest not cluing 1 to early sets of 1s if they include 1p. The exception is if 1p is on chop - at which point, you have to assume the clue is a save. Use colour to clue 1s in hands that include 1p. If you can’t clue them with colour, recognise that losing a 1 is unlikely to be too horrendous, as there are two more of them; let go of 1s that are to the right of 1p if you can’t find a better way.

There are three 5p. Clue 5 to a 5p early if you can, so that other 5 clues will become more clearly plays or saves. Especially look for opportunities to clue 5p when it is to the left of another 5. Doing this means you won’t have to save that 5 later (which we try to avoid anyway, but let’s face it, many 5s get saved.)

You can’t colour-clue black cards, but you can finesse or bluff them.

A lot of this isn’t specific to Black Powder, but these techniques will become more necessary when playing with that expansion.

[Romain672]

Convention for 'Reversed' (which is Black powder) for hyphen-ated group are present here: https://hanabi.github.io/docs/variant-specific/reversed (note there is some terms which doesn't exist in bga, so just skip them)(white reversed is the name of that variant, so apply anything you read about white)

But I will insist on one point, these conventions aim to be effecient.
Bga's one usually prefer simplicity. So 'we' should not blindly copy them (and if you do so, that has a chance to became ambigious over time, especially if it's a rare spot).

If I summarize their position:
- 1s clue in chop is a save on that 1 in chop, and a play clue on others 1s
- reclueing a 1 make the player skip the next 1 to play and make it saved as k1
- a 5 clue off chop is a play clue
- if you play with 2-saves on, 2-saves on reversed 2 is impossible, same for a '2-save' on black 4
( - implied bluff works as normal)

And hyphen-ated group use 'k' for black (and 'p' for the fifth normal suit which is purple so often used, and 'u' for null).

[Travis Hall]

I’ve been thinking about how Black Powder changes the cluing of 1s.

Now, I’ll start by noting that I prefer to note principles that extend from how the cards interact, rather than lay down arbitrary conventions that players should memorise. I like there to be reasons behind the ways we play, and then we can use our understanding of those reasons to deduce what to do when the situation is unusual and the clues not straightforward. So, that said...

(And using “p” for the Black [P]owder cards...)

1p needs to be saved, like any other unique card. So, a clue of 1 that marks chop should be assumed to be a save (of 1p, unless you’ve somehow managed to discard two identical 1s and left the third to reach chop). At the start of the game, any other 1s touched will be playable, but if this comes a little way into the game and some 1s have been played, that may no longer be true. (Player should pay attention to context, and also be aware of the likelihood of a 1 further left being clued with colour if playable.)

If a single 1 is clued, and is off-chop, that should be a play cue. (Just because it’s not a save.)

Without Black Powder, we know that cluing multiple 1s is a powerful tool to get some efficient plays early in the game, and if they are clued and nothing contradicts, we play them out. I’m not inclined to lose that tool entirely. If a player holds duplicate 1s, we can wait until they discard the rightmost, and then clue them; and we play 1s from the right, because we know the trick of waiting to discard duplicates, and thus the rightmost is playable and gives time to give other clues to stop play of other duplicate 1s.

In a similar manner, if the 1p is held in a hand with multiple 1s, and is the rightmost of them, we can wait for it to reach chop, and then clue it, and that will tend to be the efficient play. So, if multiple off-chop 1s are clued, the rightmost should not be 1p. Play from right, as before.

If a hand has multiple 1s including the 1p, and the 1p is not on the right, you can clue them, but you need a mechanism to warn before the 1p gets played. Lacking colour, it is difficult to do this for the 1p. Thus, I would not assume a clear stop clue is available. That in turn means that cluing 1s in this situation is a bad idea. Try cluing relevant colours instead, and wait until the 1p reaches chop.

And if you need to get a particular 1 played more quickly, then you aren’t concerned with getting the most efficient play, and can just clue for play with colours instead. Usually.

When playing Avalanche of Colours (and possibly more rarely without it), this can mean there is no clear clue to call for play of a 1. That’s okay; there are three copies of each 1. If things are that difficult, you can let it go and know that recovery is possible.

So, in short... play off-chop 1s when marked, and be aware that off-chopped 1s will probably get played, and clue appopropriately.

[Ankeszu]

I prefer the following way:
1 clue touching chop will be normally treated as a play clue, unless reclued - same as normally a reclue on 1s means that the next 1 to be played is not playable. That also means that if you are the second player and you have been clued 1s on chop in the first move on the game, you can't play it in the second move of the game, as there was no chance to reclue.
1bk is less probable than any other playable 1 (in multicolor, 1BK: 1/70 vs 1 non-BK: 18/70) so this attitude rather saves clue tokens, and hardly limits the plays in the early game.

Let's see the following hands:
a) 2m 1b 1bk 1g - firstly clue 1s, 1g play if no reclue, then the hand is x 2m 1b 1bk. Reclue 1 would point 1bk at the exact spot of the next 1 to be played, and also show the next 1 (1b) is playable. I also saw that some players would prefer to use blue in this case (to point 1b instead 1bk to play, and save 2m btw) but IMO it looks too much like a finesse, and not necessarly logically stoping from 1bk play if player holding 1bk want o play another 1 (for example because hoping for 1y, as the cluing player has 2y in finesse spot / few 2s behing 2y and inconvenient to clearly clue if no 1y).

b) 1bk 1y 5r 1g - same as before; two ones play and then a reclue and 5 save. It can be discussed further if an early 5 save (if no 5bk played yet) could work as both reclue 1bk & 5save.

c) C: 1g 4r 3g 1bk - 1s clued twice immediately, and we know that the second one was playable, unless stated otherwise.
If at the same time we have a player with a hand like D: 1b 1m 1g 1y, we can clue 1 to C and green to C, which imo sould be understood as 1bk save with additional green save - and a notice that the other player is to be played 1g - because all other players see its more reasonable this person play. D player would initially not now whether he has 1g 2g or just many 1s, but three D left auto-plays (after C discard 1g) probably would soon help to distinguish, and if there is still a doubt, players can always reclue 2g.

For the rest of the game, the strategy is rather obvious, with a point stated by postmans - how to mark plays for black cards behind a cards of the same number: cluing color on the other card. It's tricky with multicolor because it needs the color that doesn't touch any unmarked card. With proper attitude around the table, it's possible. (Yeah, we did bomb a bit on this )
Here is another remark - if 5bk was missed to be clued early clearly and there is another 5 in front of it, depending on the hand situation it may be worth to let this 5bk go... At the same time I prefer not to lose playable 1s early just because they were on the hand with 1bk, because that can have a higher impact on the clues that can be given later (majority of colors played 1 to 5; majority of 1s being playable at the start) and slow down the tempo more noticeably than losing first 5bk.

[Travis Hall]

I wrote a long response looking at my approach vs Ankeszu’s, and then something glitched when I tried to post it, and I lost it. So, let’s not do that again. Life’s too short.

So what I’m going to do instead is write a few shorter posts examining some of the specific situations, and see how our approaches compare, and what else we might try that could work better. Be warned, my comparisons won’t be limited to exactly what I outlined before; when we have other techniques that fit the situation better, I’d defer to them.

But I will say that I really prefer that the signals I use be deducible. I suspect that Ankeszu’s approach may have some benefits that may only materialise when you can be sure that your partners are using exactly the same techniques; as opposed to things like finesses that become obvious reasonably quickly as players experiment with the game. In BGA’s environment, I know I’m going to be playing with a lot of different players, and even things like trash bluff aren’t all that widely known. So if a 1-clue technique is a little less efficient but less likely to cause an error when playing with a expert-rated player, I’d probably go that route in general.

Anyway, specific situations in new posts.

[Travis Hall]

Let's see the following hands:
a) 2m 1b 1bk 1g - firstly clue 1s, 1g play if no reclue, then the hand is x 2m 1b 1bk. Reclue 1 would point 1bk at the exact spot of the next 1 to be played, and also show the next 1 (1b) is playable. I also saw that some players would prefer to use blue in this case (to point 1b instead 1bk to play, and save 2m btw) but IMO it looks too much like a finesse, and not necessarly logically stoping from 1bk play if player holding 1bk want o play another 1 (for example because hoping for 1y, as the cluing player has 2y in finesse spot / few 2s behing 2y and inconvenient to clearly clue if no 1y).

I think there’s a much better technique for this situation. When playing Avalache of Colors, it’s fairly common to need to clue past a multi in some way. Frankly, if you haven’t gotten a handle on that, it’s very hard to do well consistently. (Not, I suppose, that scoring actually-consistently is much of a thing on BFA at the moment; even though I rarely have partners at less than expert rating, I find myself repeating “If you can see the card (in finesse position) the finesse is not for you” way too often. But I digress.) A common technique - one that I think we should all know well - is to use colour clue that touches a multi and a non-multi, and expects the receiver to recognise that the non-multi should be played rather than the multi (even if the multi is on the left).

And an extension of that technique is to clue a colour past the next player to touch a multi and a non-multi, and require the next player to correct with a colour clue that reveals the multi to be multi. Very useful when a multi blocks play clues to two other cards.

(Obviously, I’m just talking about games with more than 2 players. I rarely play 2-player, and I think that strategically it’s quite a distinct form of the game.)

Unless the player holding this hand was the second to play, I would apply that technique here. One player should clue either blue or green (depending on the other hands) and the next should immediately clue the other of those colours.

How do you know the colour to 2m is not a finesse? Because to be clear to the player being finessed, we should use a colour that only touches the 2m. Or just use 2. (This means that there will be less opportunities for side-effect saves. I believe the clarity is worth it.)

This marks two plays and saves the 2m with two clues. After two plays, clue 1 to save the 1p. Three clues for two plays and two saves.

By comparison, Ankeszu’s approach uses two clues for two plays and one save. I’m happy with the relative efficiency, although admittedly saving 3m or 4m reduces the benefits of the extra save.

Also, everything changes if you can get the 2m played early.