A few days ago I stumbled over a fantastic website created by Nick Case:
The evolution of Trust - https://ncase.me/trust/
It explains interactively how trust can evolve based on a simple (but famous) example from game theory: the iterated prisoner’s dilemma. A few months ago I have spent some hours myself on this problem (a python/pytorch implementation including deep learning agents) so I was immediately fascinated.
One can experience in the included simulations that trust can evolve (in the sense, that a kind of agents dominates after some time, which prefers to start and maintain cooperative behavior), if ..
- Interactions happen repeatedly (if you meet a person only once, it is safe - and therefore effective - to cheat. This is not the case if you meet the person all the time)
- Win-wins are possible (non-zero-sum-game): cooperation is beneficial for both agents
- Low miscommunication (If the level of miscommunication is too high, trust breaks down)
Now I can’t resist to add some thoughts to this results:
- It's not really necessary for the game to be zero-sum to make trust break down. It’s - strictly speaking - only required that the game looks like a zero-sum game. If a large number of players believe that win-win is not possible, trust will break down. This might look like a subtle difference, but I believe it is essential to understand the current state of society.
- We see that with many repeated interactions, the tit-for-tat strategy is very hard to beat (I also could not manage to beat it, even with neural networks). Tit-for-tat agents always start with cooperation and then simply copy the interaction partners move (i.e. cooperate if the partner cooperates, defect if the partner defects). All well performing agents in this regime seem to be some variant of tit-for-tat.
- If there is also little miscommunication, this leads to a society where cooperation is the norm and non-cooperative behavior is very rare.
- But the tit-for-tat strategies ability to retaliate is still very important: if it's not present, even very few „always defecting“ agents could dominate the whole society after a short time. We could say this functionality implements the famous „Si vis pacem para bellum“ („If you want peace, prepare for war.“) quote.
This case corresponds well with what we know about the lives of our prehistoric ancestors, which were hunter gatherers.
- They lived in small communities (tribes). The number of interactions between the few members must have been - over the years - very high.
- It is also very clear that cooperation had huge benefits for these societies (hunting, defense against animals and enemies etc.). So the game was surely non-zero-sum.
- Because people were all from the same group, they knew each other extremely well and they also shared culture and language to 100%. Therefore miscommunication must have been rare.
- Therefore a tit-for-tat strategy must have evolved in our ancestors minds.
- But of course the real game of life is much more complex than the prisoner’s dilemma. Therefore there must be a huge variety of ways to defect (and, accordingly, to retaliate).
- But how then could people keep their „retaliation skills“ up to speed if non-cooperative behavior was so rare? The only possibility must have been to analyze the few cases of escalating conflicts with maximum rigor.
Therefore, our enormous interest in everything evil does not (!) mean that we are evil by nature. The contrary is true: it’s rather evidence that our ancestors must have operated almost all the time in a cooperative mode: if violence really was abundant at the time, such a strong interest would not be required.
The situation was somehow comparable to the European armies of the past decades: every day practicing for a war which never (until recently) happend. Where officers were analyzing wars far away all over the world to get clues for improving their weapons, in case they would be needed one day.
We meet men, who build remote controlled model tanks or play war with Airsoft guns in the forest, with great skepticism. But it’s actually exactly the psychological drive behind this kind of interests which is absolutely required to maintain predominantly cooperative behavior (like „peace“, among other good things).
Anyway: almost all of us have a thing for evil in one way or another. Some play with toy soldiers, others watch horror movies, war/action movies, read crime novels, practice martial arts or have a preference for (consensual) „pervert“ sex. All this is not evil (yes, even in the case of sex. Read my blog posts about this topic: 1 and 2)!
On the contrary, these are the things which are absolutely required to hold evil at bay.
But why then is evil growing wild everywhere these days?
- By owning land and natural resources we have introduced strong zero-sum elements into the game („if I own it, you can’t own it“). This is a result of the invention of agriculture. Note that owning is a cultural concept. Owning something is only possible if it's accepted by all the other members of society. Therefore the zero-sum element introduced (voluntarily by consensus) by tolerating owning are somehow not real, but they can fool us by making the game look zero-sum (as explained above).
- To make things much worse, we have invented the concept of intellectual property to make ideas and knowledge ownable. This is particularly damaging, because the interaction of sharing knowledge is actually highly non-zero-sum (sharing doubles the value of the knowledge of both the involved people at zero cost). By making knowledge ownable, this important mechanism becomes zero-sum too.
- Modern media select content to maximize emotional impact in the audience. This means that they serve their customers what interests them (for the harmless reasons described above) most: crime and violence. The problem is, that this leads to a distorted view of society: society gets perceived as much more non-cooperative than it actually is. This can, as we have seen above, make trust break down.
- Digital media also allow the automatization of communication (like TV, bulk mailings, websites etc.). As these means of communication are so cheap, they simplify 1:N communication tremendously: where some time ago you had to talk to many people individually, you can now address thousands or even millions with a few minutes of work. But -if- interactions result from these communication activities, they are very often single interactions (like a shop never sells to the same customer again). In this case - sadly - the most effective strategy is „always defect/cheat“. Because of this we get bombed now with all kinds of attempts to cheat us every day (ranging from ads over SPAM emails to fishing attempts). Therefore this extreme cheapening of global communication creates enormous incentives for non-cooperative behavior. Furthermore this also creates a false perception of society as consisting mostly of people who prefer non-cooperative behavior. And in a society which consists (or better: appears to consist) mostly of people who always cheat, it is the most effective strategy to always cheat too! Therefore it’s modern media which is a root cause of people really getting evil.
- Modern electronic communication and media also allow strangers from very different cultures to interact - mostly in languages like English which they speak only poorly. It’s clear that under such conditions the rate of miscommunication must increase drastically. This again makes dominantly cooperative behavior in the group much more difficult. Of course miscommunication happens mostly between groups with different languages and cultures. After a few failed interactions between members of such groups, the members of the other groups are perceived as predominantly non-cooperative.
But what to do now? It seems we have messed up badly.
But at least it's nobodys fault. And maybe we understand a bit better how and why.
What do you think?
Write me.
Foto: Pixabay, Artist: qimono (thank you!)
Follow me on X to get informed about new content on this blog.
