ThoughtStorms Wiki

Many interesting political/economic debates come down to collective action problems.

Collective action problems appear in many familiar guises in political and economic theory, for instance - PublicGoodsProblems, CommonResources, ParadoxOfVoting, VariousDilemmas etc.

I would like to develop a TypologyOfCollectiveActionProblems

In general, they all fit a very broad characterisation as so:

  • There are some rational agents.
  • The agents form a 'group' or 'collective' - in the very broad sense that actions taken by members of the group effect the others.
  • You could say, as theorists do, that each agent has a 'decision problem' about what action it will take.
  • You could say, as some do, that the whole group has a 'decision problem'. But here the result of the decision, the group's action, is the combination of all the actions taken by all the individual members.
  • Groups can decide on the group action in many different ways. Each individual may decide and act in isolation, or they can confer, deliberate, vote, interact in many different ways in markets or fora, etc.
  • You could also think about a 'group outcome', which is the result or effect of the group action for every group member.
  • Suppose that there is such a thing as a good or even a 'best' outcome for the group. Whatever this may mean. (Utopia, pareto-optimal allocation or something better, a situation where 'everyone is better off', or the majority are, or the ones who deserve to be ... Of course there are big problems with all kinds of SocialOrderings.)
  • A Collective Action Problem (or CAP) exists when there is a possible good outcome, but the group decides on an action that leads to a worse outcome.
  • It may (or may not) be possible for the group to 'solve' the CAP by using a different kind of decision process. (Although - adopting a decision process and sticking to it is itself a group action and may itself be problematic.)

Many political and economic institutions can be examined as proposals for SolvingCollectiveActionProblems. Which is where I think it gets interesting.


(See also : TheLeftTheoryOfDisagreement)

Some collective action problems/models:

NOTE: The starting point for serious general discussion of the issue is Mancur Olson's 'The Logic of Collective Action' (1965). The key message is - 'The existence of a large group with a common interest does not automatically give rise to a common interest.' Eg. contra Marx, just because the working class has an interest in uniting to overthrow its exploiters, doesn't mean it will.

"Collective action problems appear in many familiar guises in political and economic theory


"The agents form a 'group' or 'collective' - in the very broad sense that actions taken by members of the group effect the others."

It seems to me that there are probably far less times when an action by one person doesn't have an effect on somebody else - much of this time the effect will be intangible, but it's definitely more than nothing. All the products I use, for instance, have been created elsewhere, so I am linked to others through the labour required for my environment, or through the waste produced by this manufacturing process that will probably have an effect somewhere else.

Call me sceptical, but "Collective Action Problems" just seem to be a way of saying "having to live with everyone else". Of course, there are vast, vast fields within this (of which natural, animalistic group behaviour seems like as good a starting point as any), but CAP seems a bit too vague and abstract to me to be solvable in any direct way.


Yes this is a very vague and abstract notion. And, indeed, the vague and abstract notion of 'collective action problem' is not solvable in any direct way. However, particular kinds of collective action problems may or may not be solvable in various ways. Grouping them together under this general label will make sense if, and only if, different species of problem and their solutions have interesting things in common. (More on this soon.)

That said, I don't think that all issues of 'having to live with everyone else' fit the description above. For instance, it may not make sense to think about all social issues using rational choice assumptions. Or in many cases there may be no such thing as a 'good group outcome', whatever that means. Or if there is, it may be that the best group outcome is in fact achieved.

For instance, if you believe neoclassical economics, competitive markets lead to the best possible outcome. It is only when there are 'market failures' that 'collective action problems' can arise.


I think Collective Action Problems are more like Collective Decision Making problems (something you probably hear a lot about at Runtime :-) That seems to me to cover a wide range of problems, but is still a sub-set of "living together" problems. Collective Decision Making requires a group to assert a goal (or set of goals) and agree to some structured way of moving towards it. This is a very difficult problem. So difficult that it's often easier to avoid it by allowing decisions to emerge from the aggrogate of individual decisions. I guess that's one of the advantages to markets.

What I'm now wondering is whether collective action problems are a kind of information compression. Basically, if everyone does what they want, that's n bits of information. To get a collective decision rather than a mere aggrogate of individual decisions, you are trying to reduce the amount of information in the system so that more of the actions of people are co-ordinated with fewer degrees of freedom.

Now there are plenty of ways of reducing freedom and information, but many are clearly bad. Which might be what happens when an individual dictator forces everyone else to do what he wants. So the challange of collective action is to find a way of reducing the diversity of intentions, without getting a bad results. Rather like reducing the size of a file without losing the important signal.

(Dunno, maybe this is this isn't a very useful analogy ... but clearly I win BonusFreakyConnectionPoints for getting to InformationTheory).

Where this might be useful is in making a certain kind of assessment of different kinds of collective decision making.

Name Explain Comment Informational Issues
DirectDemocracy Everyone votes on every decision which is executed centrally Usually considered bad because people aren't trusted to understand the complexity. Maybe bigger problem is what happens when there are constraints on decision by another, but people vote for incompatible things (eg. low-taxes and welfare) Because it leaves information reduction until the end, many degrees of freedom at point of vote, informationally expensive
Pure Representative Hierarchy Society divided into clear nested groups, at bottom, everyone votes group's preferred representative, then representatives of the group vote the way they were instructed at the next level up, repeat Acknowledged problem, what if you're stuck under a corrupt, (un)representative kept in power by another faction Is information being lost prematurely at lower levels. (eg. don't get optimal compression with run-length-encoding)
Multi-level representative democracy Everyone votes for a number of representatives at different scalar-levels. Worst system except all the others - Churchill
Decentralized Anarchy Small goups vote for what they want in limited area. No central decisions are made How do you balance competing claims between groups? Doesn't throw much information away, but may not compress very much.


I like the 'signal' analogy. Makes me think of electronics.

The general form of a group decision can be:

  1. Input signal - n agents with n different objectives, beliefs etc.
  1. Circuit - decision-making process
  1. Output - n actions of n agents
  • 4. Outcome - n 'payoffs' to n agents

And some kinds of decision-making process will involve 'feedback'.

Though not all. I would tend to think of co-ordinated and unco-ordinated decision-making processes as just different types of collective decision-making. Co-ordination isn't inherently desirable. In many situations co-ordination may improve group outcomes ('solve' a CAP), but other times unco-ordinated decision-making may work better.

Though MarcusAurelius says:

'We are made for co-operation, like the upper and lower rows of teeth.' (Or something like that.)

See also :

Two things I wonder about, and what references may be good to check out:

  • Direct Voting along the lines of ClusterVoting. Basically every vote is open to every citizen, using ProxyVoting can be given to people for certain votes and not others. If you knew of and trusted someone on subject A, then you could assign your vote through ProxyVoting.
  • Collective Action Problems can be looked at in a different view, more like the TheoryOfInfiniteStruggle view, things are the same, always:

** BrainStorm, come up with a PlanToWin, EvaluateOutcome, repeat, forever. What is needed is a common understanding of the nuts and bolts of the system in which we live, as well as understanding how to work together. This is what I find so exciting about wiki, there can be a page for the information I just found out about today:

***I was talking to the middle eastern guy at the gas station I go to today. I ask him if the station will be closed on Monday - Memorial Day, he replies "no." Then he tells me the guy that signed the lease for the station's name, would rather close at 9pm every day - because it is a lot of work to keep a business open 24/7. But that is what the lease calls for. As most of the other gas station leases. Interesting level of control...

  • With the idea of [CommunityWiki]:WikiProliferation], [[CommunityWiki:WikiProliferation],] it is an exciting idea of organizing the work of a society on wiki.

– Best, MarkDilley

See also :