Skip to Content

Virtual World Study Reveals the Origin of Good and Bad Behavior Patterns

What do the patterns of good and bad behaviors in an online world reveal about the nature of humanity? Computer scientists have begun to tease out an answer.

The way patterns of behaviour emerge and spread through society is the subject of intense research at the moment. That’s partly because of the huge advances made in modelling the interactions between individuals in computer models and partly because of the sudden availability of huge datasets that capture certain aspects of human behaviour, such as mobile phone, email and trade datasets.

However, none of these techniques fully capture the way behaviours spread in society. That’s because behaviours spread from one network to another, for example, an angry phone conversation can affect the next email you write.

That’s an important limitation since it’s reasonable to imagine that when networks are superimposed on the same set of nodes (ie people), behaviours can easily transfer from one network to another.

Study this might work if it were somehow possible to synchronise datasets from different networks, but the prospects for this seem remote given concerns about anonymity. At least in the real world.

Today, Stefan Thurner at the Santa Fe Institute in New Mexico and a couple of pals show how to get around this limitation. Their idea is to study the patterns of behaviour that emerge in a virtual world where every interaction is recorded for posterity.

The world they’ve chosen is a massive multiplayer online game called Pardus, which started in 2004 and today has some 380,00 players.

Thurner and co studied eight basic actions in which players to interact with each other. These are: communication, trade, establishing or breaking friendships and enmities, attack and punishment. They simply recorded the stream of actions that each player performs and then looked for patterns that occur more often than expected.

Their conclusions are straightforward to state. Thurner and co found that positive behaviour intensifies after an individual receives a positive action.

However, they also found a far more dramatic increase in negative behaviour immediately after an individual receives a negative action. “The probability of acting out negative actions is about 10 times higher if a person received a negative action at the previous timestep than if she received a positive action,” they say.

Negative action is also more likely to be repeated than merely reciprocated, which is why it spreads more effectively.

So negative actions seem to be more infectious than positive ones.

However, players with a high fraction of negative actions tend to have shorter lives. Thurner and co speculate that there may be two reasons for this: “First because they are hunted down by others and give up playing, second because they are unable to maintain a social life and quit the game because of loneliness or frustration.”

So the bottom line is that the society tends towards positive behaviour.

That’s interesting because it opens a new front in the study of the human condition and the origin of good and bad behaviour.

On this matter, Thurner and co come to a firm conclusion: “We interpret these findings as empirical evidence for self organization towards reciprocal, good conduct within a human society,” they say.

In other words, humanity is fundamentally good.

Maybe. More interesting will be a next generation of studies that examine how small changes in environmental conditions can lead to big changes in behaviour. There’s not shortage of examples of this in real life, where entire societies have suddenyl become murderous mobs.

What is that needs to change to make humanity tend towards bad behaviour? Studies like this may help answer this question.

It may turn that the forces at work are finely balanced, which would certainly help to explain some of the more unsavoury episodes in the history of humankind.

But if that’s possible, Thurner and co may have inadvertently stumbled across a new way to study historical behaviour.

That’ll be worth watching in more detail.

Ref: Emergence Of Good Conduct, Scaling And Zipf Laws In Human Behavioral Sequences In An Online World

Keep Reading

Most Popular

Large language models can do jaw-dropping things. But nobody knows exactly why.

And that's a problem. Figuring it out is one of the biggest scientific puzzles of our time and a crucial step towards controlling more powerful future models.

The problem with plug-in hybrids? Their drivers.

Plug-in hybrids are often sold as a transition to EVs, but new data from Europe shows we’re still underestimating the emissions they produce.

Google DeepMind’s new generative model makes Super Mario–like games from scratch

Genie learns how to control games by watching hours and hours of video. It could help train next-gen robots too.

How scientists traced a mysterious covid case back to six toilets

When wastewater surveillance turns into a hunt for a single infected individual, the ethics get tricky.

Stay connected

Illustration by Rose Wong

Get the latest updates from
MIT Technology Review

Discover special offers, top stories, upcoming events, and more.

Thank you for submitting your email!

Explore more newsletters

It looks like something went wrong.

We’re having trouble saving your preferences. Try refreshing this page and updating them one more time. If you continue to get this message, reach out to us at with a list of newsletters you’d like to receive.