James Surowiecki - The Wisdom of Crowds @ KMWorld

I’m looking forward to the opening session here. James Surowiecki who’s book Wisdom of Crowds I really enjoyed is the opening speaker. Hope to hear more stories on why “all of us” are smarter than anyone of us! No slides he’s now center stage presenting without notes.

Starts with a story from the book about a famous expert who was walking though a fair and came acroos a contest where people were trying to guess the weight of the ox after it had been slaughterd. It was a diverse crowd, with experts and just those that had no insider knowledge. Afterwards.. he just took the average of the groups guesses. He thought that the experts would get it right. The crowds judgement turned out to be essentially perfect.

Under the right conditions groups of people can be remarkably accurate. For most the widom of crowds can seem counter intuitive. What’s a smart crowd? How do you tell the difference vs a dumb crowd.

Examples.

• Jelly Beans: Guess the number No one person outperforms the group.
• Who wants to be a millionaire. 1) phone a fried or 2) poll the audience. Experts two thirds of the time. Audience gets the answer right 91% of the time. Collective Intelligence becomes more powerful.
• Google: key to success if finding the right pages at the top of the ranking. The page rank algorithm. A kind of voting system. Survey’s the internet, aggregates votes / links and comes out with an answer. Hidden order underneath the surface tapping the wisdom of crowds as a whole.
• Future: Who goes to the racetrack? The prediction machine in action. The odd predict almost perfectly. In 1970 they looked at every race run with 7 horses. The favorites will win 33% of the time. Actually won 34% etc. The crowd of bettors had an almost perfect forecast. People bet for all sorts of reasons. Somehow collectively they come up with this great forecast.

The benefits of collective intelligence becomes more impressive.

• Can we use markets inside to predict outcomes. HP set this up on printer forecasts. They were more accurate than the internal forecasts
• Google set up an internal market predicting over 200 different events. Mapped almost perfectly to outcomes.
• BestBuy tried this for giftcards opened it up to all employees. It was 95% accurate.

Prediction markets can be very valuable. They get around beaurcracies and hierarchieys Orders get carried down etc. They make it harder to get info fromw wher ie it is to where it needs to be. EG Hoarding of Information. Information is power. Tell boss what they want to hear. Set up incentive to enable people to let then know what they want to hear. People gain systems to own benefit. The W0C model places the only premium on being right. Doesn’t matter where you are in the organization. These are tools that allow it to arise to the surface.

This only works under certain circumstances. How do we make the crowd smarter: Strategies for Organizations.

1) Aggregation: need a system for aggregating lots of individual judgments into one collective judgment. What you really want is some way of aggregating their judgment. There are more sophisticated voting methods etc. You want lots of individual judgments that you can bring together.

2) Diversity: The real case for diversity is not so much sociological.. what you really need is cognitive diversity, different heuristics, different tools to solve the problem. Think hard about building a diverse organization. It also makes it less likely that everyone on the team will make the same mistake in the same direction. It also helps you avoid problems with smaller groups. First is when the group is homogenious the more they talk to each other the dumber they become. That conversation may not make them more reasonable and intelligence. Unfortunatley they become more righteous more certain that they are right. So how do you disrupt group think. The devil’s advocate was created by the Catholic chrurch when someone was going to be appointed a saint. The problem is you don’t want the same person to always be the Devil’s Advocate. So the group listens.. then does it anyway. Need a diverse time from a start. Avoid the problems of peer pressure.

3) Independence: Enable people to rely as much as possible relying on their own sources of information. You want them to give you their thoughts rather than the thoughts of others. Nobody really wants to wake up and say they are conforming today. We tend to be concerned about our reputations. One way to be reasonable is to look at people around you and do something sort of similar.. Pretty close to everyone else. Going out on a limb and you are wrong; these people are punished more harshly. We tend to herd in conditions of uncertainty. There is a paradox here. You should try and stay away from the center of the groups. Groups are smartest when they act as much as possible like individuals. How do we enable genuine opinions. One of the things we spend a lot of time striving for consensus. The best group decisions arise out of disagreement. Disagreements is at the heart of collective wisdom. I like his street cleaning story. Imitation may make sense sometimes. If most are imitating the there is no reason to think the group will be smart. You have to do some work to get them to think independently.

Need to look out for talkative people. People who post a lot tend to have more influence etc. Opinions become a hub. It would be okay if talkative people were smarter. There is no connection vs what is said and wisdom. We tend to overestimate in general who we should be listening to. We think in one way or another experts will identify themselves. Information is often located in places you would not expect to look. You don’t know who should be asking. In that way you will get more input. Individual experts will make mistakes.

What’s the Promise?
A sub disappeared in 1968. Made a last transmission and the vanished. So when they went to llok for the sub… started in a circle. Finally a guy came forward with an idea. Craven assembled a diverse team of salvage, submariners etc. In consultation with them he came up with a set of scenarios. He then asked the groups to bet on these things. He then took all these bets he then had effectively a map. It wasn’t near where they were looking… They came up with a collective judgment It was 220 yards from where they said it would be. This group had little information. The group as a whole knew them all.

Blogged with Flock

Tags: jamessurowiecki, wisdom, collectiveintelligence, crowds, smartmobs, kmw07