Monday, June 04, 2007

Playing cards in class

I just joined the faculty of one business school. And I am impressed. They have this system where you use blue, green, yellow, and red cards to motivate student participation. The cards are distributed following this rule: blue (which is worth 100 points) is for active student with sensible comments (I know, you have problem already with this criterion -- hold on). Green and yellow cards follow with 85 and 70 points, respectively. Red is actually a punishment rather than a reward -- it is worth minus 10, i.e. for student who has no idea what he talks about, or who gives totally wrong answer to your question. To ensure that you don't give too many blues or too few yellows, every session is provided with 5 blues, 10 greens, 20 yellows, and 5 reds.

Sounds good as far as recognizing the importance of incentive. But I think it has some drawbacks (at least to me):
  • I'm always skeptical of forcing students how they should study. Just because you're active in class doesn't mean you grasp the course well. (I recall my fellow students who never spoke a word in class ended up magna cum laude -- while some who were talkative -- too active you lost your concentration -- flunked their exams).
  • This whole thing of distributing cards break my concentration in teaching. The books are thick already -- I don't really have much time playing the card game. And it's not easy to make fair decision in split second, e.g. when you see three hands at the same time. Not to mention the sudden protests like "I think my comment is good and deserves a blue. Why green, Sir?" -- you need at least 3 minutes to settle on that.
So in the first three classes I abandoned the game and instead trying to explain 3 chapters as clear as possible. And I got a memo from the Director saying that I should follow the system -- the card game.

OK, I'll think about a better system next time. You have an idea? (Some time ago, when Tyler Cowen asked how to solve the RSVP problem, I offer an idea -- that kind of idea: like some machine does all that card distribution thing for me, while I can concentrate with my powerpoints).


  1. whatever happen to old fashioned caning??

  2. Hi, T/S! If only some robot can cane for me :-)

  3. In business schools talking gets higher reward than studying...

  4. 'sthat right, Bleu? See, I'm such a novice :-)

  5. At the end of each class, list the names of everyone who gave his/her opinion in class. Let your students vote -1, 0, or, 1 for each name.

    Once the tally is counted, you can assign the cards based on your own card-distribution rule. This way, while you're teaching, you can act like Simon Cowell, sending signals on whether a particular comment is smart or stupid.

    To avoid collusion, you should state at the outset that the votes act as a guideline only -- you have the final say on who gets the cards.

  6. That's good idea, Arya. But maybe I should allocate 10-15 minutes at the end of each class doing that.

    Forgot to tell you, I have to input the attendance twice: before and after class! And in the second call, I have to key in any card earned by the students.

    I wish I got more time.

  7. Arya's suggestion would be good for one shot thing. .

    As Arya says, in the repeated interaction, one of the sub game perfect NE is some forms of (tacit) collusion. This collusion will be an equilibrium since your threat (Information) will not credible. Students will finally learn that they decision does affects the card distribution.

    If this is the case, the "incentive (?)" purpose of the card thing wouldn't be get.

    My suggestion is just do the card thing as they said. :D.

    Give all students a specific number (1 - 40). Every student that want to ask, has to mentioned his number. Based on your judgment, you can categorize the question (bad or good)in on small paper. Give point 1 for good and point 0 for the bad and -1 for annoying ones. I think it will not distract you from answering the question and back to your slides.


  8. In fact I had to do it 'as they said' last week, Mike. But then, in the remaining time, I rushed them with 3 chapters. And they raised flag, as for me, dehydrated :-)

    But thanks!

  9. Aco:
    I wonder whether collusion will indeed happen -- if the transaction cost to collude is too high, it might not happen.

    If you're up to it, this can be an interesting research: How many times should we repeat the game to reach an SNE? What say you? ;-)

  10. Arya,

    my hypothesis is that the collusion happens. Because, it is one of the SPNE in the this kind of repeated games. (at least theoretically).

    Yet, it could be a good thing to do class experiment in order to proof this. However, I am not sure the current set up could be used as a valid experiment set up. May be some experimental economist could give a comment on this.

    To aco, I am sorry to hear that.


  11. Mike:
    SPNEs in repeated games are often derived assuming infinite repetition. Such an outcome is, however, not guaranteed for finite games -- of which, Aco's case is one. Also, the number of players is large enough to make collusion slightly more difficult [typical analyses of SPNEs doesn't deal with many player games -- at least the ones that I can understand ;-)).

    That's why, it may be interesting to test what we constitute as "infinite" in this case.

    Of course, this is assuming that students value these cards enough to come up with efforts to collude. [I suspect, they might end up resorting to a rule of thumb or random strategy].

  12. Arya and Mike, this all is getting more and more interesting. I sure am interested in developing this into a real experiment :-) Yes, I'm gonna consult with experimental economist. Now I can't stop thinking about this whenever I am in the class :-) Accidentally, we had a faculty meeting last week on this issue. I of course raised the tradeoff issue between the card game and the obligation to 'stick with the fat syllabus'. As predicted, they wanted me to achieve both :-(

  13. i thought you were refferring to students who play cards around LP or canteen area. we used to have at least one set of 'remi' and 'gaple' card at each corner of them building. with Katjep chasing us away and witholding our KTM. oh yeah, those were the days.

    so this is not about that, yeah?

  14. Arya,

    About the two person game. Actually, I think, analysis on collusion using supergame indeed considers multi individual/firms (not just two individual) in infinitely repeated game. (this could be one-stage or two-stage game. And this is a SPNE. :))

    Yet, It is indeed true that the collusion will be difficult to be a SPNE if the period played is finite . The main difference could be the discount factor and the credibility of punishment. It happens because all people know that at certain point of time, the game will stop.

    Some experimental studies on collusion showed that the price distribution is indeed higher in than the non-cooperative nash price. (the experiment is multi person and multi period)

    Aco, it is gonna be interesting to see the results of the experiments.


  15. I might be wrong, but did I detect extra enthusiasm, certain glee and satisfaction of the fact that the subjects in question are MBA students?

  16. Mike,

    I’m sorry, I’m not really sure about your idea. Let's use Arya’s idea. Suppose I’m one of those students. If I talk something in the class and my argument is sensible (in my fellows’ views), I may get a point from my fellows. But if my fellows think that my argument is not sensible then I may get minus point. Regarding Arya’s suggestion, the problem here is that I do not know my fellows’ views regarding my idea (whether my idea is sensible or not in my fellows’ views). In repeated games, I can learn how to argue sensibly and it comes when I am really ready.

    Now suppose there is finite repeated game. Those who think “Oh my God I’m not ready to say something, I’m not sure about my idea and my friends’ view on it” will not active in the class while those who think “Ok guys this idea is really brilliant and you should buy it” will active in class. Then since my fellows think in the same way as I think, I know that those students who are active must have good ideas. What happens next is separating equilibrium. Those who do not have good ideas (or not sure whether their ideas are really good) will be silent while those who are smart will show up.

    Pointing other issues, first regarding SPNE and repeated games, I agree with Arya that the outcome maybe different, but a repeated game which has a unique NE has a unique sub perfect equilibrium. It means that SPNE may exist in finite repeated games.

    Second, regarding collusion, even if the cost of collusion is zero, do I have any reason to collude in this finite game? I think the answer is no unless I know for sure that other will cooperate and agree to vote positive point for every student in the class which is very, very hard. Even if the students agree to decide which students who are gonna talk today-“ok guys, today is your turn, you, you and you should speak up in the class”, it is, still, really hard to make sure that others will not deviate. My point here is that the cost of collusion is independent with the strategies of which each student chooses.

    But in the playing card as Aco does, it is surely different game since he is also player (I’m still thinking what sort of game Aco does). Let’s keep the game simple mate :).

  17. Sorry guys, i made serious mistake. I should drop the first sentence of second paragraph out ("Now suppose there is finite repeated game" i didn't realize when i wrote it so i came up with separating equilibrium) because i don't think that arya's game is kind of repeated game since the fellows student observe the argument first and then vote.

  18. Mike:
    Of course you're right: SPNEs can (and are often used to) solve multi-player games. I guess I need to refresh my game theory. ;-)

    In an infinitely repeated game, here's how I'd collude.

    Suppose there are 10 students in the class. I'd recruit five students to form a coalition, and tell them, "Let's randomly assign (with equal frequency) who would comment in each class session. Then, irrespective of the comments, let's give each other a +1, and give commentators not in our group -1."

    Assuming that Aco tallies the vote and give blue (the high-valued card) to the highest-ranking students in terms of the number of votes, group members who comment should always receive blue cards in every session.

    The problem is to avoid defectors. One way would be to play a grim strategy: Should any member fail to get the blue-card when s/he comments in a session, from then on all members will give -1 to all members.

    While this could work for infinitely repeated game, in theory, it's harder to make it work for a finite game: People who have received the blue-card in the 'last round' (of the group's comment-cycle) no longer has the incentive not to defect.

    However, in practice, arisans show that there are out-of-the-game mechanisms (e.g., social pressure) to ensure that most (if not all) of them play on till the end.

  19. aco, here's a thing... why don't you cut some of "importance material"... say around two or three sub chapter... they do not need to know the whole game man... that should save you some time and some breath...

    cards anyone? are there any pink and brown ones? there supposedly trendy... somehow.

  20. 1. The idea behind this whole cards game is that the institution wants to produce a complete
    manager. a person who master the art of speaking in public as well as the course. To achieve
    the first goal, the director invent the card system. I bet, he was inspired by one of his
    studying-experience abroad.
    His professor once said that he failed teaching the class because no one ask or give comment
    during the course (they just nod). He said that "if u understand, then talk or ask something
    about it, that way i know u understand"
    2. Students won't care whether his/her opinion deserves blue or green. It's all the instructor
    prerogative. if the insructor thinks it deserves green, then green it is. We won't bother the instructor with such question.
    As far as i see, no student ever asked why the instructor gives blue or green (as long as
    it is not red). I myself will studying more if i haven't get the blue card. ;)
    btw, it forces the instructor to explain all the material efficiently, given the time. :)