The interrogation strategy begins with selecting a player to interrogate. At the beginning of every move, we analyze our players' disjunctions in the representation scheme. We select the player with the most number of uncertain cards. This is the player that we know least about.
Next, our strategy tries to learn of at least one card that this player has.
To do this we ask for a list of cards that ensures that at least
one card of this player will be included. That is, if each player was dealt cards,
and there are
cards in the player's disjunction that we are not
sure about, and say we know
of that player's
cards, then
we ask for
cards. That is the minimal number of cards
that has to be asked to guarantees an answer. In addition
to the above cards, we throw in a number of random cards from the
list of cards that definitely don't belong to that player. This
measure of obfuscation is meant to make analysis of our own moves
more difficult for other players.
Note that the above is the minimal amount necessary to get an answer. We do not, though perhaps we should have, try to learn about the cards that will teach us the most. For example, if we need to learn about 2 cards from a player, perhaps one of these cards appears in another player's list of clauses. Trying to learning about one card over another one may lead to a chain of clause inferences that will yield more information.