Tempo Clues are defined as clues that "get" no new cards. The purpose of a Tempo Clue is to get a card played that already had a clue on it.
If a clue is given to a card that is already known to be playable, then that would not count as a Tempo Clue. It would instead be a Burn Clue (i.e. a clue that accomplishes nothing).
As explained in the beginner's guide, we do not typically perform Tempo Clues, because they do not meet Minimum Clue Value Principle.
When a Tempo Clue touches multiple cards, the focus is on the leftmost card. The non-focused cards are not promised to be playable.
We treat Tempo Clues differently depending on whether they are valuable or not valuable. See the next two sections below.
In some special situations, Tempo Clues have enough value to justify giving them:
When everyone knows it gets two or more clued cards to play (in total across all hands).
When the clued card is not a 5 and it is "out of order" (meaning that it is not possible for a Prompt to get the card played at this moment in time).
When playing the clued card would "unlock" someone's hand. (For level 9 players, see the section on Locked Hands.)
If a Tempo Clue is given in any of these situations, it is considered to be "valuable".
Valuable Tempo Clues can be given at any time. They don't have any special rules associated with them and they are treated in exactly the same way a "normal" Play Clue is.
When determining if a Tempo Clue is Valuable, it should only be evaluated on the turn immediately after the clue is given. (This helps keeps things simple and prevents desynchronization.)
If a player gives a Tempo Clue that does not meet any of the criteria for being "valuable", it is treated differently and given a different name.
Sometimes, players are in "stalling situations" where it would be illegal for them to discard (like when they have a hand that is completely clued). In some specific stalling situations, it is okay to give a Tempo Clue that doesn't meet the criteria for being valuable. This is called a Tempo Clue Stall. Since everyone sees that the player giving the clue was in a stalling situation, they can understand what is going on.
For level 9 players, see the Allowable Stall Clues section for the specific times when Tempo Clue Stalls are okay to give. If you are lower than level 9, the details are not important right now.
Obviously, players are only allowed to give a Tempo Clue Stall in a valid stalling situation. If a player gives a Tempo Clue that does not meet the criteria to be valuable in an otherwise ordinary situation, then they must be intending to send a deeper message. See the Tempo Clue Chop Moves section below.
A Tempo Clue Stall that is given in non-stalling situation can't be a stall - the cluer must be intending to send a deeper message.
We agree that Tempo Clues of this nature cause a Chop Move in addition to getting the focus of the card to play.
For example, in a 3-player game:
It is the first turn and nothing is played on the stacks.
Alice clues red to Bob, which touches two cards on slot 1 and slot 2.
Bob plays red 1 from slot 1. The red card in slot 2 could be red 2, red 3, red 4, or red 5.
Cathy clues number 2 to Bob, which does not introduce any new cards, but it does "fill in" his slot 2 card.
Bob now knows that he has exactly red 2 on slot 2 and that this was a Tempo Clue. Since it does not get any new cards and only get Tempo on one card in total, he knows that he should also Chop Move his slot 5 card.
Chop Moves interact with Tempo Clues in a potentially confusing way. When a Play Clue is given that touches a Chop Moved card for the first time, it is not considered to be Tempo Clue, even though Chop Moved cards are imagined to have an "invisible" clue on them. Instead, this would be categorized a normal Play Clue.
The reason for this is kind of technical. A Tempo Clue is defined as a 0-for-1 because all clued cards will "automatically" play on the stacks from Good Touch Principle. However, there is not an analogous principle for Chop Moved cards - they will not automatically play on the stacks. Thus, a Play Clue to a Chop Moved card would be a 1-for-1.
In the Early Game, players must exhaust all Play Clues and Save Clues that satisfy Minimum Clue Value Principle. Since clues to playable Chop Moved cards count as normal Play Clues, it follows that players are are obligated to give these kinds of clues before ending the Early Game.
Often times, either you or the player who comes after you can perform a clue. And the clue would have the same result if done by either player. So who should do it?
One way to decide is to look at the next player's chop. Is their chop high-value, low-value, or useless? If it is high-value, then you should discard and let them give the clue. If it is low-value or useless, then you should give the clue and let them discard.
"Stealing" clues in this way is called Discard Modulation. If other players are discarding to let you give all of the clues, you might have a high-value chop.
For example, in a 3-player game:
All the 1's are played on the stacks.
It is the Mid-Game.
Bob has a red 3 on chop. This is a one-away-from-playable card that is valuable to the team. However, it would be illegal to clue the card directly with a Play Clue or a Save Clue.
Cathy has a (playable) blue 2 in her hand.
Alice decides to discard instead of giving a Play Clue herself.
Bob clues blue to Cathy as a Play Clue.
Alice has smartly prevented the discard of the red 3, at least for a little while.
In general, 3's are pretty high value, and 4's are pretty low value.
But combined with this, cards that are one-away-from-playable are pretty high value, and cards that are two-away (or more) are pretty low value.
Thus, in some situations, a one-away-from-playable 4 may be more valuable than a two-away-from-playable 3.
According to our conventions, it is illegal to clue a one-away-from-playable card as the focus of the clue without it being some kind of special move like a Finesse. However, sometimes we still want to "save" these cards. Thus, they are prime targets for indirect saves like the Trash Chop Move or the 5's Chop Move.
However, in most cases, Chop Moving the valuable card just won't be possible. So in general, you should try to construct lines that protect high-value cards from being discarded for as long as possible.
In the H-Group, we like to find the "best" move for every turn in the post-game review. This is fun and helps everybody improve. But this can be taken too far.
Sometimes, players will give clues that are very complicated. Maybe the clue looks like it could be two different moves. Or, maybe the clue relies on non-obvious contextual factors.
Often, these kinds of complicated clues end up in misplays and lost games. And in the post-game review, the person who gave the clue gets defensive: "If you guys just played perfectly, then my clue would have worked!"
It's natural for people to feel this way, because normally, if Alice performs a Finesse, and Bob is not paying attention and misses it, then we would say that Bob is at fault.
On the other hand, things are a little different when Alice gives a really complicated clue that Bob should technically be able to figure out, but doesn't. In this case, Bob shares a little of the blame. But it is mostly Alice that is at fault.
Part of being good at Hanabi is recognizing when you should not try to do the most-optimal clue for the turn, because it would be confusing for your teammates. Clue clarity is really important and you should prioritize it! In the post-game review, you can always say: "On turn X, I considered this more-efficient clue, but it did not seem very clear. So I did this other, more-clear clue instead. What do you guys think?" That gives everyone a chance to discuss the theoretical move together and find out if it really was the best move, without introducing a huge chance of failure into the actual game.
Don't give confusing clues. If you do decide to give a confusing clue, and it doesn't work out, then you are at fault. This principle is more important than all other principles.