Correction 'finesse' imo leads to bomb

[dpjahnke]

Someone told me to go here to present this case. Seems basic to me but anyway...5w and 3Wwere clued White as a reverse to bring out 2w. Correction was needed to as 5w was leftmost and 3w next. The person clued 5 instead of 3 so I presumed 4w was leftmost in my hand (since it was not elsewhere) and bombed. I said that the playable card needed to be marked as the correction if there was no finesse intended. He said you cannot presume anything on a correction. 2500 games of Hanabi tell me otherwise but he told me to come here with this.

[Wreckage]

Was there another 3 in their hand that shouldn't be marked?

[Jellby]

Or another 5 that could be saved? Even if 3 and 5 were free of side effects, I probably wouldn't assume there's a 4w anywhere.

[Silene]

I would assume finesse if 3 would have been possible without downsides. If there are unwanted extra 3s on the hand then I would only see it as a correction.

[Ant Stewart]

As a general rule, if a correction is required in order to prevent a bomb, at any stage in the game, then you should not read a play clue from that clue.

I think Silene makes a good point - if the correction could be given more precisely and wasn't, then you should assume there was a reason for that.

So, suppose the hand has no other 3s, so the correction could be given without ambiguity, and they clue the 5 instead. In that situation you can assume there is a reason they wanted to clue the 5. But given no benefit either way, clueing the 5 provides more information if r4 ever does show up, so I think that would still count as a reason for some players.

[Blacktango]

I used to play like Silene too, in such a case.
A clue should have only one purpose. If you can avoid the ambiguities, do it.

I think it is useful to be able to trigger another finesse from the correction, sometimes.

[Travis Hall]

clueing the 5 provides more information if r4 ever does show up,

No, it doesn’t. If we follow a principle of cluing the last playable card in the sequence, so that all the cards can be played after the clue, then cluing 3 to 3r instead of 4 to 4r tells the holder that their other red card is not 4r. If it isn’t 4r, what is left? Only 5r.

When playing Avalanche, multis can modify this, but in that case cluing 5 to 5r doesn’t confirm the card as 5r anyway. Usually a colour clue will be given for some other reason that will happen to clarify soon enough, and in that case all the issues with multis go away. If multis are not clarified by another colour clue, the holder needs to wait for a confirmation clue, and since one clue should suffice for that, any other information given by earlier clues is irrelevant.

There is often an assumption that information is the markings on the cards (and the more information, the better, so mark as many cards as possible). That is not true. A lot of information can be conveyed by what we do not clue, by our choices of clues, and by the timing of our clues. When we play to touch extra cards, we limit the information we give in other ways.

In cases like this, we would give up the ability to signal an extension for the finesse up to 5r if we refuse to limit ourselves to cluing 3r when the finesse is not available. That is a much more useful channel for information than to directly mark 5r, noting the ability to deduce as above.

This sort of logic is also why so many players will splash a non-multi card with colour when they want to clue a multi that has been previously marked, thus producing ambiguity in their finesses when the clue receiver holds a multi.

I like to splash many extra cards when those cards have not been marked at all previously in order to prevent those cards being discarded. I find that splash onto cards that are already marked (and thus “safe”) is much less useful.

[CoolLuke]

So Travis, what if there is interference with a trash 3 in their hand? Or what if there is another 5 that will need to be saved soon, and tokens are running low?

[Travis Hall]

So Travis, what if there is interference with a trash 3 in their hand? Or what if there is another 5 that will need to be saved soon, and tokens are running low?

Then you clue 5, and the other players can see why. Real masters pay attention to the context. This means you can’t clue self-finesses this way, but the colour clue being corrected has usually ruled that out anyway.

[Romain672]

Or you wait for the trash 3 to be discarded (if you can affoard it which is usually no), or wait for the other 5 in chop and do a 5-save.
Or you clue 5 when w4 is drawn.
Or you use 2 clues on that hand.
Or you clue 5 when the risk is minor or null (be careful because timing give strong sign of w4 drawn).

If none of those options please you, you will have to take a risk somewhere.