August 23, 2011
How to do a Crowd
I work for a company that makes crowds for a living. Here are some of the things I’ve discovered (with the help of my colleagues) along the way: 1. Crowds are only useful when the quality of the crowd output improves as the size of the crowd improves. It is easy to create big groups of people, but far more difficult to get them to scale usefully. The key thing seems to be providing some mechanism that allows crowd members to observe the behaviour and actions of other participants as part of the core activity of the crowd. 2. Crowds can very easily turn into mobs. A mob, in this context, is any crowd which no longer performs the function for which it was originally created. It goes off instead on some tangent which may or may not be actually destructive, but at best isn’t creating much value. You almost always get a mob when you fail to build in some control function that lets crowd participants evaluate each other against some common standard of behaviour. The control function also has to have some way of advertising to everyone how well each individual participant is complying with the standard of behaviour. 3. Crowds have cultures and traditions. They’re set by the very earliest members who join. Whatever that first small group of people does, is what the whole crowd is going to from then on. It is usually simpler to throw away your crowd and start again if you want to change these established cultures and traditions. 4. Crowds have a magic number of participants. Below the number, they’re just discussions. Above the number, useful behaviours start to emerge as group dynamics begin to work. The magic number is usually much larger than you’d expect. And getting a crowd to that number is very hard work. 5. If you’re building a crowd, recruitment of members before the magic number is the number one challenge you’ll face. Until group dynamics start, crowd participants get little value from staying around, so you have to do things to make them stay. 6. Crowds are surprising. They’ll hardly ever do what’s expected. If you want predictable, you don’t want a crowd. 7. Crowds aggregate and average. Where there is a distribution of possible outcomes, they’ll take the position closest to the mean. Also, if data can be aggregated to a higher level of abstraction, they’ll tend to do that too. Individuals won’t display this behaviour, of course, but the crowd does it automatically.
8. Crowds are only capable of doing three different things: recruiting new members (virality), going to the same place as everyone else (herding), and creating new knowledge from trials or observation (emergent wisdom). 9. Whatever you’re designing, you really want to have the crowd do all three things if you want to maximise your chances of getting the outcome you want. 10. Crowds are living, breathing things. They require care and feeding, like a garden. The constant tending is expensive and time consuming, but necessary. Fail at this, and your crowd will either die from attrition or turn into a mob from lack of attention.
James Gardner works for a company that makes crowds for a living. In this article, he shares the things he discovered.
Working for a company that makes crowds for a living, James Gardner shares what he's discovered:
1. Crowds are only useful when the quality of the crowd output improves as the size of the crowd improves.
2. Crowds can very easily turn into mobs.
3. Crowds have cultures and traditions.
4. Crowds have a magic number of participants.
5. If you’re building a crowd, recruitment of members before the magic number is the number one challenge you’ll face.
6. Crowds are surprising.
7. Crowds aggregate and average.
Read the rest of this article to know more...