## List of Principles

### Play-(or-discard)-to-save principle

In some systems like Sieve, the hand is a "holding area" for useful cards; to (temporarily) save cards, any Play Clue or a Discard Clue can be given.

- The benefit is that you can keep possibly-useful cards around as long as possible, and you can always discard the worst saved card, greatly increasing discard quality.
- Such a system may waste clues on constantly asking specific cards to discard. Maybe an old card in the hand was already a good discard, or maybe an

### Urgency principle

Urgency principle: A clue should change meaning if it is delayed (and it was previously available).

### Referential principle

To make a card play/discard, it's better to touch cards other than that card.

### Tempo

It is very good to get cards played and out of the hand as soon as possible.

#### Interaction between Tempo and Play-to-save

These two principles are directly in conflict. If

### Safe Actions

We prefer to guarantee that we can give players safe actions.

- This can conflict with efficiency; sometimes getting a safe action with some signal limits the empathy given by that signal.

#### Tempo Clue Focus

We can fill in a card to get it to play. This is sometimes the only safe action in the hand, and is always unblocked, but may be inefficient.

### Action Choice

When a player has multiple plays/discards, this is a powerful non-clue-based channel of information that should be maximally utilized.

### Empathy maximization

Among untouched cards, older cards in hand are inherently more valuable due to negative information.

### Signal synergy/antisynergy

The effect of a clue on a specific card can be to play it, discard it, or touch it. Depending on the system, a clue can also "permanently" save a card.

- Play & Discard
- These synergize. If a card is not playable, it is more likely to be discardable.
- Play & Touch
- Bad; violates Referential principle.
- Discard & Touch
- Somewhat synergize: if a card does not want to be touched, it's likely because it wants to discard.
- However, touching a good card is most likely to be useful, and good cards are less likely to want to discard.
- A compromise could be some concept of "chop moves" which are cards assumed to be good.

Ref sieve has problems with touching slot 2; slot 2 is the least likely card to be useful, and touching it has the least empathy value on average. In addition, if slot 1 is playable, one can rank clue slot 1 to discard slot 2 and later tempo clue slot 1, but this has the same efficiency as color cluing slot 2 and then fixing it.

### Sieve principle

We want to always talk about newly-drawn cards at least once before they are saved.

### Playability of cards

Newly-drawn cards are always pretty likely to be playable compared to other cards, before endgame. Maybe a strong bias towards freshly drawn slot 1 could be good?

If a saved card was unplayable but useful, each time a card is played, the chance of a saved card being playable goes up slightly. Older saved cards (that were marked unplayable at the time) are more likely to be playable. This indicates that a good system can flexibly get any saved card; alternatively, some kind of revolving focus could be good?

### Empathy maximization

Touching cards with the most negative information gives the most expected empathy.

## Development

Ref sieve discard clues touching slot 1 is often the wrong card to touch- you risk repeatedly cluing the same rank. Cluing cards that already have negative information maximizes the empathy given to all cards.

Some quick ideas:
- Chop-prioritized ref plays and saves?
-