another gap question

[loafbit]

I get what you're saying, but let me try at a different angle. You could've clued k5 (yes, even though I had it) but you didn't. In other words, you didn't clue k5 because you saw the path to w2 w3. Therefore, and this is the point of contention, you are the gap.

The focus of the white clue was the new white (w3). You were going to play w1 even if I didn't give white. I gave white, you did the thing you were going to do anyways, and Gemmy responded to white, seeing that you didn't try to clue k5. Why should Gemmy keep digging? Well because the white clue would be meaningless if she stopped at k5.

First, about the gap. Why Gemmy played 5k immediately after 1w played? Imagine Matrics or oktryit had 2w in finesse pos or their cards were X-Y-Z-2w, and X Y Z are all playables, would Gemmy still play for white in a situation like this? Of course not. So Gemmy played 5k showed he *failed to see* 2w playable in others people's hand, so he played his 5k as 2w, the white bluff ends here, no need to digging for 2w, and after that dmpr0 knew his white is 3w. On the contrary, how layer happened? If Gemmy saw Matrics or oktryit could played to 2w, then he knew white clue had nothing to do with him, so Gemmy's decision to not involve in the white clue created a gap here, Matrics or oktryit knew they must keep play untill 2w.

Second, why shouldn't dmpr0 clue k5? The essence of the matter is why BGA convention allow skip/distant bluff and what are the pros and cons of it? You can think about it yourself.

[dmpr0]

Honestly I am starting to double think the situation. I did not have any obligation to do anything (like a bluff/finesse requiring my play) so I could easily clue k5 - that would invalidate that w clue on me if it was a simple bluff and cause bombs. So such a clue should not be given. If we trust the clue-giver to give a good clue, this must be a layer then? We can only consider bluff when I am forced to do something and can't make a gap in any way, or have matching cards to play in a general case.

[Blacktango]

The clue tells you to play the Bk5 and informs you of the W3/M3 in your hand. I think this is good enough

You can guess that someone is bluffing when they ask you to play at a time when you would usually clue the next player.
In that case, you should trust the clue giver and play as they ask, rather than giving a clue.

I agree that this case with the W clue is not the most valuable.
However, the bluff can sometimes be very effective, allowing you to save a critical card in someone's hand without delaying the two plays.

[_mg6maciej_]

Honestly I am starting to double think the situation. I did not have any obligation to do anything (like a bluff/finesse requiring my play) so I could easily clue k5 - that would invalidate that w clue on me if it was a simple bluff and cause bombs. So such a clue should not be given. If we trust the clue-giver to give a good clue, this must be a layer then? We can only consider bluff when I am forced to do something and can't make a gap in any way, or have matching cards to play in a general case.

Self-doubt is a good first step to learning and improving.

There seem to be three possible conventions to be used by a given group that follows what's commonly referred to as bga convention system, but let me prefece what I think are pros and cons of each by saying that among the games that are lost, majority are due to human error: inattentiveness, low perception or sometimes trying to be too clever. More complex convention in a system means more human error. BDRs are probably the 2nd most significant factor, but it's not even close. Efficiency is almost never a factor unless errors are made. I'm talking strictly about the original Hanabi and ignoring kids version of the game, because it makes it significantly easier to win even if multiple blunders are made.

1. it's a bluff unless the gap was created at any stage of fully resolving the clue
This is what I have presented on the previous page with several examples and what seasoned players like Blacktango in the previous post seem to advocate for.
Pros:
+ the definition of a gap is quite simple
+ bluffs are a lot more common than layered and patch combined, and when playing with strict promise bluffs almost as potent
+ when players fully grasp the concept and treat it as first class citizen, you are allowed to give such clues more often, partially ignoring the con below
Cons:
- a player who is not involved in a clue might mess it up while assuming it's a finesse clue involing them, so care must be taken not to give such clues under many circumstances

2. you can only ever bluff the next player, it's layered or patch otherwise
Pros:
+ even simpler definition of a gap
Cons:
- very rare to happen
- drawing the line there disables a lot very commonly used bluffs like skip bluff (when the next player thinks they are responding to the clue by playing a previously marked card, sometimes even involving a prompt) or bluffing over a player that has all cards known or one that is or is about to start responding to another clue (usually long finesse or layered), or even the simplest finesse + bluff (like yellow or 3 on 3y making 1y and 1b play by the next two players) if you are strict about it

3. trying to marry the two conventions above
Pros:
+ you preserve the possibility to bluff distant players (see con of conv 2) while avoiding the possibility someone will mess it up (con of conv 1)
Cons:
- unclear or nonexistent definition of a gap, or when you are obliged / compelled / forced to respond to previous clues and when you can delay your play, creating more opportunities for human error (number 1 factor of game losses)
- you are gaining some layered at the cost of many more bluffs and while layered is fun and efficient, more cards marked for play by a single clue means other players will have nothing to do and will discard sooner, creating more BDRs (number 2 factor of game losses)

All in all, there is no single way master level gamers play, so the best thing to do is discuss what convention a group will play before the game start. I like conv 1 above the most, because of its similicity and ability to save more cards with bluffs, avoiding more BDRs. And using that conv, avoiding such clues when they could be overriden by one of the other players.

How complicated defining what being obliged is might be shown by multitude of examples, so here is one:

You are Bob.
Emily started the game and clued 1 to you. Most likely 1 and not a color because one of your leftmost cards is multi, blocking a color clue, but it's not impossible Emily is an excellent player casually setting up Alice for a potential color self bluff (neg 1 on 3 other cards) next round if you hold a card like 2g and happen to draw a playable after playing your 1.
Alice clued white to Cathy, indicating 1w is there, either finesse or marked.
You can obviously see the other two 1w on Emily, but are you obliged to play your 1 right here right now or can you finesse Cathy's 1y with say 2y on Alice or give a neat blue promise bluff to 4b on Alice?

Alice: X 2y 4b X
Bob: X X (1) X
Cathy: 1y 2(w) X X
Donald: 2b X X X
Emily: 3b 5k 1w 1w

Solution: {
It's a patch finesse for Emily to play 3 cards after you play marked 1b and Donald plays 2b. Cathy creates the gap as defined in conv 1, by discarding or performing another unrelated action. Both you and Donald should be thinking you have 1w. Patch is almost always a (nice) surprise. If you don't perform your "forced" action and finesse / bluff 1y instead, Donald bombs 2b. His pov is his hand should be 1w X X X if you finesse or 1w 2b X X if you give a promise bluff.
}