Game on Rawls’s Second Principle

by Harry on March 18, 2004

For the first time in ages I am not teaching a course in which I’m going to use this game, but this morning I got a plaintive email from a former grad student saying that she had lost her copy, and wanted to use it in a class she is teaching. So I thought I’d post it just so that it is available to anyone who hasn’t already made it up themselves. The game is heavily adapted from a series of experiments done by Frohlich and Oppenheimer, and I’ve refined it over the years. It always seems to work well, even with lower-division students in low-discussion classes. I tend to use it prior to asking them to read any Rawls, and spend about 10-15 minutes at the beginning of the game explaining each of the principles and insisting that I really mean it that they can make up their own principles. I assume lots of people have done something like this, so if your version has features you think I should incorporate please let me know.
(Note, as you can see I do not distinguish the two aspects of the second principle — when I have done that it has confused the students and made the game work less well — this way, in fact, they frequently distinguish the issues themselves in the class-wide discussion we have after the small groups have worked it through).

Game on Rawls’s Second Principle.

You are bargaining on behalf of your client in an Original Position. You have to come up with a principle of distributive justice which your client will be able to live under satisfactorily. You know nothing of your client’s talents, abilities, gender, religious views, sexual orientation etc., except that you do know that she or he is not disabled. Fortunately, the other people in your group are in exactly the same situation. You will have to think about how you will justify your chice to your client when you find out who he or she is. Note: a prior OP has already agreed that basic liberties — freedom of conscience, religion, expression, association, and the rights to participate in public and political life, as well as the liberties associated with the psychological and physical integrity of the person — must be guaranteed to all, so you need not worry about violations of people’s basic liberty-rights.

You have six principles from which to choose. Remember that these principles will govern your client’s entire life, and he/she will not be able to escape.

1. Laissez Faire: Markets will operate without government intervention, except to protect private property (including intellectual property through patenting and copyrighting legislation) and to place modest limits on the emergence of oligopolistic and monopolistic markets.

2. Equality of Resources: People will have roughly equal resources available to them over their full lives.

3. Sufficiency: Everyone will have a `basic needs’ safety net guaranteed. Above that level, markets will determine rewards, except as the democratically elected legislature chooses to constrain them. (If you select this principle, be prepared to explain what count as basic needs and why).

4. Maximin: Inequalities of resources will be arranged so that the least advantaged will better off than they would be under any other arrangement.

5. Equality of Welfare: Resources will be distributed so that everyone is more or less equally happy.

6. You may, if you choose, formulate a compromise between these principles, or formulate an entirely different principle. If so, it must be precise and you must be prepared to defend it.

Appoint a scribe for your group. Also appoint a chair. Discuss the principles for 45 minutes. Then vote on the principles. You are aiming to achieve consensus. (In the real OP you would have eternity, which we cannot reproduce here).



Chris Bertram 03.18.04 at 9:00 pm

That’s great and very well designed. I particularly like the justification to client part. When I’ve attempted a similar exercise I’ve not had that feature. Instead I’ve given them each an envelope containing an identity (“You are Bill Gates”/”You are a disabled single-parent on welfare” and so on) which they get to open at the end. But your way of doing it is so much better.


mark 03.18.04 at 9:36 pm

Thanks for the exercise! This is quite fortuitious for me, as I’m teaching Rawls next week (for the first time). I think you could also combine the way you do it with Chris’ suggestion as well, and somebody (From another group that agreed to a different principle) could play the client of the other group after the decisions have been made, which should get the juices flowing…


Tom Runnacles 03.18.04 at 9:38 pm


Damn, that’s a good one.

Almost makes me wish I was back in my previous incarnation teaching pol phil so I could try it out :)

If you’ve been running this game for a reasonable length of time, do you detect any patterns in students’ responses? What principles of their own do they generate? And does rerunning the game post-Rawls make a difference to what they say?


Matthew Yglesias 03.18.04 at 9:41 pm

I’m interested in the fact that you don’t use Rawls’ actual formulation of the second principle, but instead the alternative he calls Natural Aristocracy. It’s a relatively minor point in the grand scheme of things, but I’ve always felt that N.A. was preferable to the Democratic Equality (i.e., maximin constrained by fair equality of opportunity) that Rawls espouses. How did including the opportunity principle “confuse” the students?


Jacob T. Levy 03.18.04 at 10:00 pm

Why maximin as the only maximization principle? Maximizing aggregate, mean, or median welfare/ wealth/ utility are all fairly commonly-thought-of options. And I’d imagine students without a prior background won’t immediately intuit the difference between equality of resources and equality of welfare– at least, they won’t be able to quickly describe what the latter looks like.


Bob McGrew 03.18.04 at 10:29 pm

My favorite maximization principle that no one seems to use in these Rawls debates is minimax regret.

A person’s regret is their loss from not living in their favorite society (the society which gives them as much as possible. Note that this will depend on their endowments: if you give as much as possible to someone who is less endowed, there’s not much to go around to motivated the better-endowed who would be producing most of the society’s value.)

Then you organize your society to minimize the regret of the maximally regretful person.

As far as justifications for the clients go, this is the simplest justification of all. The client is going to be unhappy with the bargainer insofar as the bargainer didn’t get the best deal for the client. The bargainer can say: “Look, if I’d made any other choice, you could have been more unhappy with me than you are now.”

I have no idea what a minimax regret society would look like, but I think it’s a very natural decision rule to use.


Bob McGrew 03.18.04 at 10:47 pm

Another thing I’ve always wondered: do people who have never heard of Rawls really come up with the maxmin distribution rule? Is that robust to what society you use? I’ve never read Frolich and Oppenheimer (or heard of it before), but it sounds relevant.

Harry – what usually happens in your class?


harry 03.19.04 at 1:02 am

Thanks for the nice comments, and anyone should feel free to use this or variants. I recommend Frohlich and Oppenheimer. Their experiments are interesting though they ask, from a Rawlsian point of view, the wrong questions, because they allow the subjects to incorporate their own views about justice into the answers. I reviewed it in Economics and Philosophy 1993, and its one of the better things I’ve written (wasted in a review).

Responses: I think in future I’ll incorporate Chris’s envelope idea after the fact and make them write up their justification to the person they got as a client in the envelope.

The answer to Tom’s q sort of answers jacob’s too — there are only three real contenders, and its the same three whatever others I include: maximin, sufficiency, and laissez faire. I vary the others, and used always to include max ave utility, but it makes no difference to the discussions the students have. I did it twice in England and those times even laissez faire didn’t get a look-in. I do two things that matter in the pre-game talk — explain the difference between welfare and resources (so that those who don’t intuit the difference get it) and explain the difference between maximin and equality (which I am sure skews the discussions against equality).

Sufficiency is by far the most popular response, and interpretation of it tends to be plausible and generous (though the last time I did it, one group interpretated sufficiency as meaning ‘enough to stay barely alive’).

Bob — how different do you think your rule is from the maximin decision rule? I’ve been rolling it round in my head and can’t figure it out.

Matthew — Matthew Clayton has a terrific paper arguing for Natural Aristocracy in the Croatian Journal of Philosophy, and it has prompted me to rethink my own views pretty radically, in your direction. Rawls, interestingly, admits in Justice as Fairness that he doesn’t know any more what sort of priority, if any, fair equality of opportunity should have. He says:

bq. Some think that the lexical priority of fair equality of opportunity over the difference principle is too strong, and that either a weaker priority or a weaker from of the opportunity principle would be better, and indeed more in accord with fundamental ideas of justice as fairness itself. At present I do not know what is best here, and simply register my uncertainty (p163, note 44)

The students find the exercise confusing when equality of opportunity is put into the mix because a) they find it hard to deal with something as complex as R’s full second principle plus a bunch of variants and b) they already assume that the Liberty Principle protects some basic non-discrimination principle, which most of them have not yet distinguished from anything as strong as fair equality of opportunity. If I had much longer class time (I get 75 minutes, so much longer would be double that) I would try to do it the more complex way again.

BTW, no-one has said that Rawls’s justification of the maximin decision rule in TJ doesn’t work. I think it doesn’t work. And I find the revision in Justice as Fairness (pp97-99) less than fully enlightening.


Bob McGrew 03.19.04 at 6:51 am

Minimax regret is a very common decision rule in computer science. The idea is that, if I don’t know something about my payoffs, I should try take the action which minimizes the “regret” I will have once I learn it. (It’s used, for instance, when you have to delete information now without knowing what information you’ll need later. You delete the information you’ll least regret having lost.) In this case, the thing I don’t know when picking a society is my preferences – so I should choose a society so that, when I learn my preferences, I have the least regret about the decision I made.

As we know from economics, to get an efficient society (one with maximum social product) you need to reward the people who are productive. So a society that rewards productive people will probably give the best possible payoffs for the best endowed. A maximin society (as Rawls describes) will give the best payoffs for the least endowed.

Consider the regret of people in a society where everyone gets everything equally (the straw man Rawls sets up.) The regret of a highly-endowed person would be much greater than the regret of the less-endowed person, because the highly endowed person would stand to gain much more from living in his best society than the lowly-endowed person would from living in his. In a maximin society, the least endowed person would have no regret, but the highly-endowed people would have lots of regret.

A minimax regret society would end up giving more to the highly-endowed people than a maximin society would. In practice, I’m not too sure what a society that was formed from this decision rule would look like. My point is basically that there’s a bunch of reasonable decision rules to use in that situation – and minimax regret seems just as justified as maximin. (Some days I think it’s even more justified).

Along those lines, why do you say that maximin doesn’t work? I’d say it works, it’s just non-unique.


Dianne 03.19.04 at 10:49 am


Thanks for posting the game for me.

I used the game yesterday in class, and after the groups decided the principles they wished to choose, I handed out candy based on their choices.

This is similar to Chris’ idea about handing out an envelope to see who they might be, but the candy seemed to solidify the consequences of their choices very effectively and tangibly.

From a former grad student (who happened to be writing a lesson plan for Rawls the night before…)


Jacob T. Levy 03.19.04 at 8:41 pm

there are only three real contenders, and its the same three whatever others I include: maximin, sufficiency, and laissez faire. I vary the others, and used always to include max ave utility, but it makes no difference to the discussions the students have.

I’m surprised, I guess for two reasons. One is that laissez-faire isn’t the same kind of thing as the others. The “system of natural liberty” might be preferred for First Principle-style reasons, but if natural economic liberty isn’t thought of as liberty of the sort that Veiled choosers would consider part of the first principle, there’d be no reason for it to occur to them in the second stage. It’s not a distributive principle at all. It might be the institutional system recommended by a distributive principle (e.g. maximize mean or median welfare/ resources), but it doesn’t really belong in the list any more than Rawls’ own “property-owning democracy” (the institutional arrangements that he guesses will approximate the difference principle) does. At second-principle time one would opt for a distributive principle, empirics-to-be-inserted-later-and-subject-to-revision-if-the-system-doesn’t-match-the-principle. I’d have guessed that, unprompted, more would come up with something Benthamite than with laissez-faire, unless they’re already convinced that laissez-faire is utilitarian-best.

Second reason I’m surprised is because of the character of student objections to the DP I remember hearing, which were sometimes egalitarian (DP=too much inequality) but otherwise aggregative or utilitarian- that among between the distribution [1, 2, 2, 2] and the distribution [.9, 100, 100, 1000, 1000000] it’d be crazy to demand the former ex ante, unless the difference between .9 and 1 was the difference between starvation and otherwise. Some degree of “when in doubt, maximize expected value” occurs to them pretty quickly, in my experience.


clew 03.19.04 at 10:59 pm

Can this be modeled in Sims Online yet?

Comments on this entry are closed.