This Nobel prize in Economy proves that self-organising teams are better and cheaper*

* More accurately

In 2009, Elinor Ostrom was the first woman to be awarded the "The Sveriges Riksbank Prize in Economic Sciences in Memory of Alfred Nobel" for her analysis of economic governance, especially the commons.

In her book "Governing the Commons," she describes examples of communities all over the world that manage a "common pool of resources" without a centralised authority, and have done so successfully for generations. Think of fishermen agreeing amongst each other what their quotas will be for the season; a farming village that builds and maintains an irrigation system which distributes a limited supply of water to everyone's fields; maybe a village of sheep herders who decide together how many sheep everyone can put on the common grazing field.

The prevailing economic theory says that this is impossible. After all, these systems are very sensitive to free-riding, everyone has an incentive to cheat a little to get a bit more. Overfishing, not doing your maintenance duties on the irrigation channels, overgrazing... It seems inevitable that over time the common resources will be exhausted, making everyone worse off. This is what is called "the tragedy of the commons".

What's the solution, how to avoid this tragedy? This is what we usually see: the government sets quotas, checks that fishermen don't catch too much and imposes fines. The common grazing field is privatised and the owner allows access for a fee, while taking care of long-term maintenance. According to the prevailing theory, only an external authority that sets rules and punishes violations can protect the commons.

In practice however, Ostrom discovered many counter-examples of communities self-organising and governing themselves without such an external authority. In these communities, the users themselves set and enforce the rules. She noticed that these counter-examples were similar to each other in many ways, which she summarised in 8 design principles. If you want to set up an environment where self-organisation can be successful, these design principles are a great source of inspiration!

In what follows, however, we will go a bit deeper. Just like Ostrom, we will use economic game theory to explain how self-organisation without external authority is possible. Even better: we will prove that a self-organising team is cheaper than hiring a boss!

TL;DR

We will use a somewhat abstract description of a team as an example to show that:

1. To influence behaviour, good transparency is more important than punishment;
2. Transparency is easier and cheaper to achieve between team members;
3. Working as a self-organised team is cheaper than hiring a boss.

Please don't let the numbers and tables discourage you. They are meant to help clarify the thinking - if they are confusing then please skip them!

Collaboration dilemma aka 'Prisoner's Dilemma':

Imagine this situation: a team is faced with a certain task that requires two people’s knowledge and skills.

The two team members can choose to collaborate to get the job done. Or they can be smarter and try to minimise their own effort while still getting credit for getting the job done. In other words, they hope that their team-mate will do the work. Of course, if both team members choose this, the job will not get done and no-one will get any credit. That would be the worse possible outcome for the team as a whole. However, if one team member can take the credit on their own, they are seen as the hero and get an extra reward. In that case, the other team member has lost their precious time without getting any reward for it. We could express “credit” or “reward” in arbitrary numbers, maybe even call them “storypoints”. Then we can present all possible scenarios and outcomes in this table:

If A is really smart, she will choose to go for the credit without doing the work. Why? In case B decides to do the work, A will get 10 if she does too. However, if she goes for the credit alone, she will get 11. In case B decides to go for the credit alone, A would get -1 if she puts in the work. If she also decides to go for the credit alone, she will get 0. So regardless of what B decides to do, A will always be better off by not doing the work and just going for the credit.

The same is true for B. So we are in a situation where both A and B have an incentive to not do the work. Obviously it would be better for each team member if they both do the work, they would get 10 + 10 instead of 0 + 0. But unfortunately the situation is such that the expected outcome is the worst possible outcome. This is how game theory represents the tragedy of the commons. In this particular scenario (also known as the prisoner's dilemma), there is what is called a "stable equilibrium" at "A gets 0, B gets 0".

This is what we should expect to see in real life: cheating, free-riding and the worst possible outcome. How can we motivate A and B to do the work and get the best possible outcome?

Centralise the coordination

Boss to the rescue! Let's introduce someone who is responsible for the team to collaborate and who can punish people for cheating. She will observe if the team members actually do the work and reduce the reward by 2 if she notices that they are just going for the credit. This provides a clear incentive for both A and B to do the work. The numbers in our table get modified slightly:

In this new situation, A will choose to collaborate. Why? In case B decides to do the work, A will get 10 if she does too. However, if she goes for the credit alone, she will get 9. In case B decides to go for the credit alone, A would get -1 if she puts in the work. If she also decides to go for the credit alone, she will get -2. So regardless of what B decides to do, A will always be better off by doing the work and not just going for the credit.

Again the same is true for B. Result: A and B will collaborate, the job will get done! Just knowing that they would be punished is enough, they don't even need to actually be punished. We have found the theoretical solution to our coordination problem!

All thanks to the all-mighty, all-seeing boss.

Bosses are humans too

What if the boss doesn’t see everything that is really going on, or makes mistakes? When transparency isn't perfect, she can’t always punish correctly. Imagine she misses 30% of the time. This means that 70% of the time she would punish for not collaborating, but 30% of the time she would accidentally punish when somebody collaborates. Again, this can be represented by modifying our table. The table now represents the average case, taking into account 30% or 70% of the punishments (the punishment itself is still 2):

Now we’re back in the original situation: the tragedy has returned. Both team members will always get more by just going for the credit (either 9.6 vs 9.4 or -1.4 vs -1.6). Again, they are incentivised to not do the work, even though this results in -1.4 instead of 9.4 for each of them.

From this example we can conclude the following:

If what is really happening is not very transparent, punishment fails to

influence behaviour.

Fortunately, this is a very unlikely situation. Any boss being wrong 30% of the time is clearly not a very good boss. However, it is in the team member’s advantage that they don’t get punished, to be seen as collaborative. How can the poor boss really know what is going on if the information coming from the team members themselves might be biased to avoid punishment? Can she really trust their green-amber-red reporting? How detailed and objective can any report really be, can it ever provide perfect transparency?

Reporting and coordination don’t come for free

What’s more: the boss is also part of this system. The time and energy she spends on finding out what’s going on also has a cost. And probably the team members have to spend time and energy to keep the boss informed. Taking these costs into account can only make things worse.

So we find that a real life boss comes with downsides too.

Unless we find a boss who's always right, who understands every aspect of the work and who works for free, they might not be the optimal solution to our coordination problem.

Let's start again from the original situation and see if we can do better.

Deciding to collaborate beforehand: sharing the reward

Team members know that, in the original situation, the most likely outcome is that the task doesn’t get done and none of them gain anything. They realise they can do better if only they can find a way to avoid the tragedy. There actually is a way to do this. All it takes is giving up individual rewards. If the team members decide beforehand to always share the credit no matter what, the table would look like this:

In this situation, there is no reason not to do the work. A has nothing to gain by just going for the credit, her reward would be the same or would be lower. In other words: agreeing to share the credit removes the incentive to cheat. The task will get done, both members will get 10. This is better than the original situation, where they would both get 0. So they both have a good reason to agree to sharing the credit.

If, during the work, they find out that the other is not collaborating after all, they can still decide to disagree to sharing the credit (and stop collaborating, meaning the task wouldn’t get done). This is not worse than the original situation, so they still have no reason to disagree to sharing the credit. We can represent this reasoning in a new table, if we are willing to quadruple the size and make a table containing 4 tables:

After all, we know that the expected outcome of the original table is that nobody gets anything. So 3 out of 4 tables get reduced to "A get 0, B gets 0". And we know that when they both agree to sharing the credit, that tables reduces to "A gets 10, B gets 10". In other words, the table collapses to:

This means that if the team members really want to optimise their reward, they should both agree to the idea of sharing the credit.

Team members are humans too

Of course, this means the team members will have to spend some of their time coordinating the work, instead of executing the work. What is the effect of that extra cost?

Say it costs them 30% of their time to do the coordination, meaning they can only do 70% of the work. If that would result in getting only 70% of the credit, we can make one final table:

When taking into account the cost of self-organisation, the expected outcome is still that the team members agree to sharing the credit and that (at least part of) the work gets done. Final score: 7+7 credit. The final score with a human boss was -1.4 - 1.4, which is a lot lower.

In real life, a self-organising team without a boss, makes perfect economic sense.

Q.E.D.

Back to real life

Obviously it's not as clear-cut as this. This game-theoretic model is a simplification. The numbers chosen here are purely arbitrary. And we know that all models are wrong, yet some are useful. What this model highlights is that there at least exists some logic, some circumstances where self-organisation is a better choice than centralised authority.

And if it's not clear-cut one way, it's also not clear-cut the other way: people who say that self-organisation can never work are wrong. This model, plenty of counter-examples and a Nobel Prize prove as much!

"People who say that self-organisation can never work are wrong." - Nobel Prize Committee

Further reading/watching

Ostrom accepting her Nobel prize: https://www.youtube.com/watch?v=T6OgRki5SgM

A great course about Game Theory: https://www.youtube.com/watch?v=NSVmOC_5zrE&list=PLKI1h_nAkaQoDzI4xDIXzx6U2ergFmedo

Iterated Prisoners dilemma: Tit-For-Tat! https://www.sci.brooklyn.cuny.edu/~sklar/teaching/f05/alife/notes/azhar-ipd-Oct19th.pdf