Board Game Arena

Hello everyone,

After hundreds and hundreds of games with the same group of players from Strasbourg (FR) during which we kept improving our way of playing to Hanabi, we ended up putting our game convention in black and white.
This is based on the rules of the game and the principles from which strong recommendations derive. We've summed it all up in the chart below.
Your comments are welcome (in particular on translation from French to English) !


Bonjour à tous,

Après des centaines et centaines de parties avec le même groupe de joueurs de Strasbourg (FR) pendant lesquelles nous n’avons cessé d’améliorer notre façon de jouer à Hanabi, nous avons fini par mettre noir sur blanc notre convention de jeu.
Celle-ci est basée sur les règles du jeu et des principes desquels découlent des recommandations fortes. Nous avons résumé tout cela sous forme du graphique ci-dessous.
Vos commentaires sont les bienvenus (en particulier sur la traduction de français en anglais) !

Hi, I really like this kind of post, ty for it.

First remark, discarding 'oldest first except if you got a reason to discard to discard newest', it look like to me discarding newest first would just be better overall.
(bga discard oldest like strasbourg which add an extra step instead of not thinking and discard leftmost, hgroup discard newest)

Second, 'make the saves possible only for the prevous player do the save, except if you got information which are globally known' is not done on bga or hgroup, but is a thing which came regulary, and which change lots of others things.
You lose the possibility for 'chop management' (ie, stealing clues to make the right person discard), but I guess you could gain a lot with others things.
(for bga readers, playing slot 1 is not an 'other thing' because the clue will look like a bluff, and if you don't build others things around it, then that doesn't seem valuable)

Third remark, about 'cards should be played from the most recent to the oldest'. What do you mean by it? Are you talking about playing all red cards on a clue which touch three reds? Are you talking about priority of plays, playing r1 in slot 1 vs playing y2 in slot 2?

Last remark, 'when you got one or more plays, play them without delay'. It's a nice simple rule, but imo, it should not be enforced. You could build others things around it (ie, make him less likely to not play), but never clue if you got a play look pretty bad imo. Again, strongly linked with 'chop management'.

Feel free to answer me in french if you aren't fluent in english. (I read the english picture)
I like how you put links on Rules and Principles to build the others things.

Hi Romain

Thanks for you input, it’s good to talk to other fans of Hanabi !
Here are my comments to your remarks:

Your 1st remark
I think our convention results like the BGA one if you apply the other parts of our convention.
Players have less information on their newest cards than oldest one, so, according to me, it seems to be risky to discard newest cards in priority (I would really be interested to know your arguments to discard newest cards).


Your 2nd remark
I think you talk about that part of our convention “Unless there is an obvious reason for all players on the table or for a reason of multiple save, a clue should only be done about the hand of the next player”, don’t you?
This is a powerful consequence of two principles: “Each player must do his best to send clear messages to their gaming partners” and, indirectly, “Every player must trust their gaming partners”.
Breaching it opens strong possibilities of free clues like Finesse, Bluff and many other ones with the help of another convention which is “If you do not understand an action done by another player, it is because this action must mean something on the cards of your hand” (this last sentence is very important in our way of playing because it enables to give information at any breach of our convention).
I understand what you mean regarding chop management, but I have many experiences of players spending a clue token to give information to a far player instead of discarding and, by this way of acting, preventing one of the next players to make a Finesse or a Bluff.
Every player must deeply believe that their chop has been managed by their gaming partners, except during the first round, so discarding must be the action by default if you have no relevant information to give to the next player or if you have no card to play. The only exception is if you see that the next player cannot correctly manage the chop of his neighbor by spending one clue token, then you must help him to do it (that’s what I called “multiple save”).


Your 3rd remark
Yes, that’s what I mean. If a player informs me that I have 3 red cards in my hand in slots 1, 2 and 3, I will play first the red card of the slot 1, then the red card of the slot 2 and then the red card of the slot 3, except if another player spends a clue token (which will stop me) or if I receive any information to change that order.


Your 4th remark
With our convention, a player can have to play a series of cards and other players must await he plays all of them before giving him a new piece of information on his hand (if they do it before, they will stop him and give him a wrong information on his hand). But he might have another important card in his hand to play and he will be able to learn it after he will have played his complete series of cards. That’s why we recommend that a player should play without delay (furthermore, playing good cards is the goal of the game).
By the way, breaching this convention (i.e. do not play a card immediately) gives information to other players (see my comments about your second remark for more info).
Moreover, this convention includes another strong convention, that we call “Finesse oblige” (sorry for my French). It means that if if you are encouraged to play a card by finesse (so the clue token has not been spent to give information on your hand, but on another gaming partner’s), you must play the card immediately and not postponing this action (by giving a piece of information to another player, for instance). This is because you cannot know if it is a Finesse or a Bluff and cannot know your card. Not playing it immediately can generate a big misunderstanding.


I hope I am clear. If not, feel free to tell me

1)

Jean-Baptiste wrote: ↑18 March 2022, 16:33I think our convention results like the BGA one if you apply the other parts of our convention.
Players have less information on their newest cards than oldest one, so, according to me, it seems to be risky to discard newest cards in priority (I would really be interested to know your arguments to discard newest cards).

Ok I start to see the upside.
But the point is 90+% of the time, you are sure they are trash.
Let me give you an example:
b3 is on the stacks. Donald is holding b5 but don't know it.
Alice clue blue to Cathy on b1+b1.
Bob who play between Alice and Cathy play his slot 1, it's y1 and it's sucessfully play.
It's now your turn as Cathy
If you discard rightmost, that's the basic safest move to do.
If you discard leftmost, you tell you know the focus of the clue which was the most important card is not playable.

2)

I understand what you mean regarding chop management, but I have many experiences of players spending a clue token to give information to a far player instead of discarding and, by this way of acting, preventing one of the next players to make a Finesse or a Bluff.

Yeah it's not something easy to do. Bluffs and Finesses got value during early and mid game, before starting to have many clues and try to save any leftover good cards / preparing the end game. Point is is to try to see what will happen if you steal a clue in 2/3 turns, and if it's fine, that normally said that every player in between got an useless card in chop or a known trash card/known play.

Every player must deeply believe that their chop has been managed by their gaming partners, except during the first round, so discarding must be the action by default if you have no relevant information to give to the next player or if you have no card to play.

We agree on the fact your chop isn't critical. It's nearly always not playable too.
But you don't know if it's useful or not, it could be any single red 4, where the other copy could be in the bottom of the deck (and if it's the case, that discard reduce the max score). If the whole board is trash, i don't hesitate to steal absolutely any clue. Don't forget that discarding make your chop not finessable/bluffable anymore.

3)

Yes, that’s what I mean. If a player informs me that I have 3 red cards in my hand in slots 1, 2 and 3, I will play first the red card of the slot 1, then the red card of the slot 2 and then the red card of the slot 3, except if another player spends a clue token (which will stop me) or if I receive any information to change that order.

Ok ok, I never played like that, only like bga and hgroup, which is to play the focus, and save the others, so I don't have a good grasp on it.

4)

With our convention, a player can have to play a series of cards and other players must await he plays all of them before giving him a new piece of information on his hand (if they do it before, they will stop him and give him a wrong information on his hand).

Second time I read about it. Never tried it. Are finesses included? Because if so that look bad to me because it's a nice clue to give anyway.. If they aren't incldued, I'm fine with it.

By the way, breaching this convention (i.e. do not play a card immediately) gives information to other players (see my comments about your second remark for more info).

Hgroup play with 'Occupied play clue' where if you save a card while holding a play, and not in your next player, then (except for edges cases), that save became a play clue.
In others cases, or in all cases on bga imo, that's for chop management, but some years ago that caused lots of troubles inside the community.

Moreover, this convention includes another strong convention, that we call “Finesse oblige” (sorry for my French). It means that if if you are encouraged to play a card by finesse (so the clue token has not been spent to give information on your hand, but on another gaming partner’s), you must play the card immediately and not postponing this action (by giving a piece of information to another player, for instance). This is because you cannot know if it is a Finesse or a Bluff and cannot know your card. Not playing it immediately can generate a big misunderstanding.

Yeah, it's simplified by 'blind play' in hgroup, you are supposed to put full priority on your blind plays. If you know you are playing a finesse (ie, if Alice clue Donald, Bob discards, and you should answer), you have the possibility to delay it if you assume what you will do instead will be clear to others players.
On bga, that don't have a name, and I think they was sometimes more aggressive with delaying, but that's close.

b3 is on the stacks. Donald is holding b5 but don't know it.
Alice clue blue to Cathy on b1+b1.
Bob who play between Alice and Cathy play his slot 1, it's y1 and it's sucessfully play.
It's now your turn as Cathy
If you discard rightmost, that's the basic safest move to do.
If you discard leftmost, you tell you know the focus of the clue which was the most important card is not playable.

I don't understand. What is the interest to clue b1+b1? Do you mean b4+b1?

But you don't know if it's useful or not, it could be any single red 4, where the other copy could be in the bottom of the deck (and if it's the case, that discard reduce the max score)

Of course, but noboby can know it. That's a risk which is present in many discarding, it is very uncommon to be able to discard only cards that can be seen on other hands or already played all along a game. And discarding is essential to be able to give pieces of information.

Ok ok, I never played like that, only like bga and hgroup, which is to play the focus, and save the others, so I don't have a good grasp on it.

We played like you do some years ago (i.e. "playing the focus only") but I can ensure you that "playing the complete series if you are not stopped" is much more profitable. It opens more possibilities (I can give you some examples if you wish of different interesting cases).

Second time I read about it. Never tried it. Are finesses included? Because if so that look bad to me because it's a nice clue to give anyway.. If they aren't incldued, I'm fine with it.

The finesse and bluff are not included because you can only be stopped playing a series if someone spend a clue token to give a piece of information on your hand. So, as the clue token will be necessarly spent by the previous player (see my comments about your 2nd remark on my previous post) on the hand of a player after you, you will play your first card (due to "Finesse oblige") but you would have been stopped in your series and you will have to continue playing it the next round if nobody stop you meanwhile.

Hgroup play with 'Occupied play clue' where if you save a card while holding a play, and not in your next player, then (except for edges cases), that save became a play clue.
In others cases, or in all cases on bga imo, that's for chop management, but some years ago that caused lots of troubles inside the community.

According to me, it is very difficult to generalize the type of information which is given by breaching the priority of "playing a card which can be played". It is really depending on the gaming context. The only info is that there is a higher priority in the game and if you cannot see it in the hands of the other players, it would mean that this priority comes from your current hand.If it is possible, your partners should clarify this priority by spending one or more clue tokens on your hand. If it is not possible, it would certainly mean that you cannot discard the card of your chop.

On bga, that don't have a name, and I think they was sometimes more aggressive with delaying, but that's close.

You are right, it can be more agressive with delaying but it is very uncommon and very risky, because, as I said, you cannot know if it is a Bluff or a Finesse and you do not know the card you should have played if you have not delayed.

Jean-Baptiste wrote: ↑19 March 2022, 18:28

b3 is on the stacks. Donald is holding b5 but don't know it.
Alice clue blue to Cathy on b1+b1.
Bob who play between Alice and Cathy play his slot 1, it's y1 and it's sucessfully play.
It's now your turn as Cathy
If you discard rightmost, that's the basic safest move to do.
If you discard leftmost, you tell you know the focus of the clue which was the most important card is not playable.

I don't understand. What is the interest to clue b1+b1? Do you mean b4+b1?

Trash bluff?
If b4 is played, you clue blue on b1, and y1 sucessfully play out of it, that said that blue card is trash, no?

Of course, but noboby can know it. That's a risk which is present in many discarding, it is very uncommon to be able to discard only cards that can be seen on other hands or already played all along a game. And discarding is essential to be able to give pieces of information.

It's not that rare, but that require luck and/or complicated conventions.
But that can be the goal of adding complexity on existing conventions.

We played like you do some years ago (i.e. "playing the focus only") but I can ensure you that "playing the complete series if you are not stopped" is much more profitable. It opens more possibilities (I can give you some examples if you wish of different interesting cases).

The finesse and bluff are not included because you can only be stopped playing a series if someone spend a clue token to give a piece of information on your hand. So, as the clue token will be necessarly spent by the previous player (see my comments about your 2nd remark on my previous post) on the hand of a player after you, you will play your first card (due to "Finesse oblige") but you would have been stopped in your series and you will have to continue playing it the next round if nobody stop you meanwhile.

Ok I finally get it, interresting, I would be interrested to try.

You are right, it can be more agressive with delaying but it is very uncommon and very risky, because, as I said, you cannot know if it is a Bluff or a Finesse and you do not know the card you should have played if you have not delayed.

Point is, if many clues were wasted (ie per example someone was locked, got only saved cards half the game), you should not do it. If many finesses and bluffs has been done, you should instead put nearly no value on bluffs and finesses since that only gain one clue, which you don't care.

Jean-Baptiste wrote: ↑19 March 2022, 18:28We played like you do some years ago (i.e. "playing the focus only") but I can ensure you that "playing the complete series if you are not stopped" is much more profitable. It opens more possibilities (I can give you some examples if you wish of different interesting cases).

Well, you can go ahead and provide such examples if you like, but I doubt you will convince the best players on BGA. Play-one vs play-all is one of the most basic questions that Hanabi theorists have to grapple with, and you’re certainly not the first to come here and try to argue in favour of play-all.

Certainly play-all provides some more opportunities for efficiency. However, when Hanabi is played (using a play-one strategy) by highly skilled players, imperfect scores are far more often the result of lost cards than inefficiency. That means I would rather retain the opportunity to mark useful cards that are not currently playable (and not then have to find a valid clue to prevent further plays).

This is especially so once Avalanche of Colors is added to the game.

You already argue against that consideration…

Of course, but noboby can know it. That's a risk which is present in many discarding, it is very uncommon to be able to discard only cards that can be seen on other hands or already played all along a game.

…and yes, it is uncommon to go through an entire game without a discard of a card that could be needed to avoid a loss of points.

But it’s not an all-or-nothing matter. We don’t have to eliminate all risky discards in order to significantly increase the probability of a perfect score. If we reduce the number of risky discards from 10 to 5, we can expect the number of imperfect scores to reduce by close to a factor of 2.

Frankly, many BGA masters are far less aggressive with splash saves than I would prefer, especially with Avalache of Colors. I see so many missed opportunities, especially with 1s clued 1 instead of a colour.

We played like you do some years ago (i.e. "playing the focus only") but I can ensure you that "playing the complete series if you are not stopped" is much more profitable. It opens more possibilities (I can give you some examples if you wish of different interesting cases).

When I started playing hanabi, I also played like that. But what are the chances that the reds are in the correct order already? Meaning, in most cases 2 clues will be required (play clue followed by stop clue), so play 1 and save the rest (then playing it later without additional clue due to good touch) is much more profitable (on average)

This is especially so once Avalanche of Colors is added to the game.

Honestly, my experience in Avalanche of colors is poor. Our convention has mainly been tested with the "5 multicolors" version.

When I started playing hanabi, I also played like that. But what are the chances that the reds are in the correct order already? Meaning, in most cases 2 clues will be required (play clue followed by stop clue), so play 1 and save the rest (then playing it later without additional clue due to good touch) is much more profitable (on average)

The stop clue must be profitable as well (which is 99% the case) and an additional free information is given as the card which has been stopped cannot be played.