kar: lets start talking about P3 ! how many of u have played Clue !!

kar: its not the same as the game we're playing.
	(explains the Clue game)

kar: in Gen. Clue, there's only 1 category..
	n items, p players, k hidden items

jw: will n be much larger than k ?

kar: yes generally..
	what do u think is the nature of this problem ?

edan: u have to select ur list of items when interrogating others.
	everyone else also gets some information. so u have to be careful.

mb: u need to do some kind of formal logic to analyze this information.

yaniv: u might infer if A asks B a list, then A has that list.

kar: but I may be trying to confuse everyone else.

lu: its no harm asking for cards u know the opponent doesnt have.
	and u also have to choose a probability at which u want to
	guess. if u wait until u know surely, u may be too late.

eric: seems like a game tree strategy might help. u keep track of
	what everyone else.

kar: yes keep track of what everyone else knows !

ma: we can have a matrix which contains the probability of every player
	having a certain card. and we update the probabilities as we go on.

kar: okay, can u see a problem with this matrix solution ?

vlad: there's a complex dependence structure among elements of the matrix.

steve: the matrix thing would be good as a precursor of your final guess.

edan: keeping track of the whole history is important. u might infer something
	later from it.

kar: I think thats equivalent to storing the disjunctions.

yuan: what if u ask everyone about all N cards ?

kar: thats a good idea at the start, but after that everyone can just show
	you the same card over and over again.

rooz: maybe you are overestimating what others can infer from your
	interrogating style. its going to be very difficult.

dv: the moment u say something is a bad idea, I think that may be a
	good idea, just to confuse others.

mb: seems like u can combine the formal logic approach with the probabilistic
	matrix approach.

rooz: it may be good idea to ask for medium number of cards !

lu: its better to ask what they have. it giver less info. to other players

andy: I agree, in the early game its better to find out what other ppl have.

an: I think that depends on the number of cards u ask for.

kar: can anyone think of a situation where everyone gets the same situation
	but still u benefit more.

mm: if u ask for 10 cards of which 9 u have.

dv: i disagree, i think you're giving more information than u're gaining.

valerie: i think Miq's idea wouldbe good when u're pretty sure that the
	opponent has the 10th card.

mb: i think the asker now knows where those 10 cards are.

adam: I think u're difinitely getting more precise information.

dv: I think everyone else gets more incremental information.

mm: u're also giving false info. in the sense that others might think
	u dont have those 10 cards.

rooz: 

jw: I think he means that if u have information about player 2, that nobody
	else has and u work towards getting a complete picture of player 2's
	hand, u'll be much better off.

vlad: if u have information about player 2, then all players have that information.

jw: u might have asked player 2 more often than others.

ma: items that are being guessed a lot may more likely be among the hidden
cards.

kar: if u want to make a guess of k cards, u only have one chance.

rooz: if u put lot of garbage in ur interrogating cards, that'll be good.

jw: u have to have a more refined taste in garbage. better than random 'random'

edan: if u ask the same person for the same cards, they'll likely give u the
same answer. thats like losing a turn.

kar: there's a field called reasoning about logic.
Book by Joe Halperin !

kar: lets shift gears !

jw: its simpler task to simulate ur own inference procedure for others.

kar: firstly, is everyone convined that given enough rounds, u can guess
	the hidden cards eventually.

edan: n - k - c rounds to know all the hidden cards.

kar: can anyone do better than that ?

jw: maybe a binary search kind of idea.

lu: but wont be u be better off, because the opponent may say 0
	and then u dont have any more information.

han: if there are only two players, u can do better than n - k -c

kar: can u do this in better than n - k - c. If no, can you prove it ?

kar: okay how many rounds do you think it will take ?
	in this particular game ?

steve: players who move early may have an advantage.

kar: I'm not sure about that, does anyone think the opposite.

steve: if A asks B a certain thing, all the other players will get a chance to
utilize that information.

adam: I think in every turn based game, the player order becomes unimportant
	after one round.

kar: lets talk about deliverables.

lg: a player that guesses eventually.

kar: the simulator would eventually force you to ask.

dv: every player should be able to keep some information.