Attendance Present: JW MH DA RM BH BGB AC JRF IW NN RB SM HB JDG HBB SK BPS SJY SP BG RG MD DK; Absent: AM
KAR entered class about 10 minutes early. GSKC had set up the game simulator on the screen. KAR gave GSKC a deck of cards, and asked him to separate them into 10 hands to match the display. KAR requested that students mention their names when raising comments or asking questions, to help KAR and GSKC (and the rest of the class) to become acquainted with the everybody's names.
KAR started the discussion. He asked the class for ideas on things to think about when playing our particular variant of poker. SP suggested that the ranking of cards was important. KAR amended that: the ranking of the hand is what is used to judge players, e.g., a three-of-a-kind hand would be higher ranked than a pair hand. SK asked for KAR to confirm that the players were allowed to pass, and choose not to swap cards at all if desired. KAR confirmed it. MH then proposed that players should try to maximise their score instead of concentrating on coming first among the players. After all, the order in which the players are seated does affect the ranking somewhat.
JRF thought that players should never propose swaps that allow the player you're swapping with to leapfrog you in the rankings. KAR then demonstrated a situation where a 4th ranked player proposes a swap with the 7th ranked player. The outcome of this swap is that the 4th ranked player now becomes 2nd ranked, and the 7th ranked player becomes 1st ranked. Should the 4th ranked player have proposed such a swap? The students certainly think so!
RM expressed that the currently 5th ranked player [with 4 rounds remaining] should try look at several rounds ahead so as to see what direction the game is heading in. KAR agreed, emphasizing that players should always aim to have better ranking at the end of the rounds; how far you are from the end of the game should be considered when making decisions. HB brought up the notion of other ``scheming'' players that carefully watch your moves, and will try to prevent you from doing well. However, most of the class thought that players would be better off concentrating on improving their rank, as opposed to deliberately targeting highly ranked players for mistreatment.
SK suggested making swaps that benefit not just your player, but also other players -- this provides more incentive for these other players to help your own player. KAR agreed, and added that proposed swaps have a higher chance of being accepted if they improve the other player's ranking too. DK wanted to swap with as many other players as possible. This way, everyone is happy. However, KAR reminded him that there will exist constraints in the game that will prevent such a thing: a limit on the number of rounds, not all players have card(s) that you might want, etc.
NN addressed the conflict issue that might arise when the same card(s) are needed for the hands targeted by multiple players. Such conflicts potentially hurt both conflicting players. KAR added that being aware of the directions that other players are playing towards will be of help when deciding what cards to play. SP reckoned that it is much easier to realise a projected hand with a smaller number of rounds remaining due to the fact that hands might change radically over the course of a large number of rounds. This might be so, but KAR reasoned that once certain players have attained their target hand, they might be content enough to refuse any more changes, leading to convergence rather than constant change. There might also be an on-going cycle for a player who stays-put shifting from highly-ranked to low-ranked simply because other players change their hands so radically. In this situation, players must not only have a preferred hand they play towards, they must also aim to stay ahead of the crowd.
IW noted that the number of players in the game is important. KAR followed up on this: the number of players is directly proportional to the number of cards available for swapping -- the more cards that remain undealt, the less choice each player has for swapping. For games with few players, there is a lower chance of players getting the hands they want, particularly the higher-ranked ones, like four-of-a-kind and royal-flush. DK proposes that players use some sort of look-ahead heuristics: with some sort of AI, a player might be able to estimate how much other players would want to swap with it.
KAR mentioned some architectural issues that students will need to consider. How should players be structured? How do you measure how much other players might want your cards? How do you estimate the likelihood of you getting cards from other players? MH suggested that players attempt to capture, and hang on to, high-value cards early on in the game. This way, such players actually determine the likelihood of other players getting these cards from them. This will prove useful, especially if these cards in question are crucial for these other players to help form their targeted hands.
KAR then divided the students into 10 groups, one for each player as shown on the screens, and distributed the cards that GSKC had dealt to each group. At first, KAR said the game would last 10 rounds. The first round passed with only three swaps being agreed to. Most swaps were rejected by players holding out for better deals. For example, two teams each had an 8 and a J. One team proposed to the other to give up the 8 in exchange for receiving a J. This deal was rejected, because the other team was holding out for a deal in which they received the higher ranking pair.
The first round took a fair amount of time, so KAR shortened the game to 2 rounds. As a result, students were making more aggressive swap proposals since this was the final swap opportunity. At the end of the game, most of the groups ended up with fairly good hands. Two groups even had four of a kind. This was quite something for a 2-round game.
An interesting observation that SK made during the game is that some cards may have a high value that can be used in an indirect way. For example, suppose that player 1 has three jacks. Player 2 has a jack, but does not need any card that player 1 is holding. Player 2 could swap the jack to player 3 for a card that player 2 wants. Player 3 values the jack highly because he knows that player 1 will almost certainly want it; perhaps player 1 has a card that player 3 desires.
At the end of the game, KAR requested that students start work on the projects right away, that they work on the implementation and raise implementation issues away from class. This will give more time for discussing the actual problem in class. He suggested that the students initially try to come up with a simple but original player, perhaps by modifying one of the given players. The most important thing is that the students' players actually work with the simulator.
Groups for project 1
| Group | Members | |||
| 1 | JW | MH | DA | |
| 2 | RM | BH | BGB | |
| 3 | AC | JRF | IW | |
| 4 | NN | RB | SM | |
| 5 | AM | HB | JDG | HBB |
| 6 | SK | BPS | SJY | SP |
| 7 | BG | RG | MD | DK |