The Science of Crowds

Jonathan Cooke

If you’re a Premier League Football team, you must follow a simple formula for winning the title. First, you win at home against other teams. Then, against lower league opposition you win at their ground as well.

The ability of a crowd to motivate players has long been established as a factor that rival managers consider when they visit a home ground. The fan base is often described as a ‘12th man’, spurring their team on to do incredible things. But do crowds really have that effect? And can crowds have more palpable effects on the world than simply winning football games?

crowd pixnio.jpg

Image Credit: Pixnio

The concept of ‘crowd wisdom’ can be traced all the way back to Aristotle, but the most quoted example comes from 1906 and Francis Galton, a famous statistician. He observed that the median and mean guesses of a crowd that were attempting to guess the weight of a cow came astonishingly close to the actual value –  within 1%.  

It doesn’t really matter whether your crowd is made up of Nobel Prize winners or the people from your local pub (who may or may not be prize winners). Imagine you are asking several people how many jellybeans are in a jar. You’ll get a few people overestimating and you’ll get a few people underestimating. Whilst you would imagine this would skew the results away from the actual number of jellybeans, it doesn’t really. In fact, you tend to get a similar number of each type (overestimates and underestimates) of poor guess, so they cancel each other out when it comes to working out your average.

To put it another way, as an individual you may never win a ‘guess how many of x is in this jar’, but get a big enough crowd together and you’ll get an average which is close to the actual answer.

It’s for this exact reason that the votes of Dixville Notch in New Hampshire U.S.A are broadcast ahead of the rest of the country. Prior to the 2016 election, this small village had successfully predicted the eventual winner of the presidential election eight out of fourteen times the results were broadcast, (to those interested, they didn’t predict Donald Trump, but then again who did?) and is generally seen as a predictor of how the rest of election will go.

Collectively pooling our intelligence has helped us in many numerous ways. Wikipedia is the greatest example of them all. An encyclopaedia that is generated, and edited, by an internet population of 3.2 billion people. Now of course, as your professor will tell you, never source directly from Wikipedia. Many pages, particularly small ones that are very specific and dull to the clear majority of the population, are trustworthy; they are maintained only by those who care about it.

Popular pages, on the other hand, such as those of celebrities or politicians, are regularly accessed and edited by individuals with agendas, or just those who want a laugh. In doing so, they can rapidly spread false information which quickly becomes group intelligence, which can have disastrous consequences.

“Crowd crushes” were for a long time seen as the main result of large crowds working towards a singular goal (indeed, for a long time, crowds have been seen as being more stupid than individuals). A crowd can act as a ‘wave’, which can ebb and flow and is a danger to itself. It’s been observed that if you put more than 7 people within a square meter, then when you apply force to them then they’ll act as more of a wave than individuals. This can result in people literally being crushed to death as the pressure of crowds forces the air out of their lungs, or industrial strength fencing being bent by the sheer weight being put on them.

Thankfully, modern studies of crowd behaviour focus on modelling natural patterns when groups of people walk in a similar direction. For instance, much like geese, it’s been observed that when a group of people, three or larger, walk in a similar direction then they’ll naturally form a V-shaped formation, which forces crowds to part around them and facilitates faster movement. This information is being used to better coordinate fire exits.

The key ingredient for crowd intelligence to work is that all decisions taken by the individuals must be independent of one another. To take the jellybean example, the best way to ensure you get a more accurate estimate is to hide participant’s answers from one another. Individuals when given more information tended to group towards a similar set of answers, which means you get a narrower range of answers, but the accuracy can get worse.

The stock market crash of 2008 was largely down to this problem. Many banks followed the example of a few, loaning out unsecure loans and mortgages, falsely believing that because others hadn’t failed, neither would they. This incorrect consensus amongst banks meant none saw the inevitable crash coming and so therefore were unable to respond to it.

‘The wisdom of the crowds’ is routinely used by politicians and those who do not wish to listen to experts on subjects. Whilst crowds can be smart, they are only useful for numerical decisions that have one simple answer. However, studying crowds and crowd behaviour will allow us to ensure the safety of every individual caught within the group-think.

#Crowd #JonathanCooke

0 views0 comments