When (and How) to Ask a Crowd?

November 19, 2007 · · Posted by Jordan Frank

ImageIn "Wisdom of Crowds is Cowardice," Central Desktop points to a Ross Mayfield statement (on the Conferenza blog) about the benefits of making decision rights more participatory and decoupling information rights from decision rights. Central Desktop concludes by urging "Lets just try to keep a little perspective when we talk about this stuff." OK. Lets do that...

I'll agree with Mayfield's assertion that providing more open access to information is a good thing, but to a point, and for a purpose. More precisely, I strongly advocate moving to a "can know" from a "need to know" culture, but certainly can't see enterprises moving to a "must know" culture. More on that in Collaboration Tools - Are Information Silos a Problem?

I'll also stand by Central Desktop's assertion that decoupling information and decision rights "would be mass chaos, counter-productive and a waste of time" - but only some of the time.

Central Desktop claims that "Successful business is NOT a democracy - successful business is a culmination of experience, talent, wisdom, lessons and execution." I think the intention here is that managers make better decisions than masses.

I agree, and research backs up the claim. The Fall 2007 MIT Sloan Management Review article "Intuitive Decision Making" by Kurt Matzler, Franz Bailom and Todd Mooradian says that "90% of the differences between top performing and average performing senior executives can be explained by emotional intelligence." Emotional memory triggers faster than cognitive processes and, thus, can be used effectively to quickly assess a lot of information and make good decisions. Being able to manage, recognize and interpret emotions in a business context is a manager skill that reflects a master chess player's capability to recognize patterns at lightning speed. (The article also tells us that a chess player can recognize 50,000 board patterns when pieces are naturally configured - but can't beat a novice when the pieces are placed at random on a board).

Central Desktop also relates the argument against using Crowds for decision making against the backdrop that the "success of most collaborative systems rests with a tiny percentage of users" and notes a 2% contribution rate at Wikipedia and 0.16% at YouTube.

Central Desktop may be right about the participation rates but wrong about the assertion. When it comes to participatory decision making, a high lurker ratio may be a good thing. While the 98% of wikipedia "lurkers" aren't lurking, they are contributing to something, somewhere. They are gathering their own information and collaborating in their own communities. They are becoming experts in something, reading their own source information and gaining the skill to bring diversity to decision making process.

Besides, mathematically speaking, its impossible to achieve high contribution rates when we look at any one collaborative workspace. What if there were 100 people in 10 open communities. The likely outcome is that all 100 would visit all 10 but would be primary contributors to only 1 or 2 communities. Is that a failure to achieve 100% contribution, or an absolute success? Obviously, its a success. I talk further about how to extend participation in blog and wiki contribution in Beta Bloggers Need Not Lurk in the Enterprise

So, what are the implications of all this when it comes to making decisions. I feel like Dennis Kucinich when I say "I actually read The Wisdom of Crowds." (For context on this remark, fast forward to the 5 minute and 38 second mark in this YouTube slice of the CNN Democratic Debate).

To Ross's point, In some cases, equal access to all information and an equal vote with a simple average makes sense. This is the case if you are counting jelly beans. Only one random person in a class of 56 can outperform the average of the class. Many (but certainly not all) business decisions can be rounded down to a Jelly Bean counting type of exercise.

In the book, Surowiecki describes a scenario where a submarine was lost (somewhere in the Atlantic, if memory serves me right!). When it came time to locate the sub, a group of experts came up with their own best guesses and an average of the guesses led them straight to it. Given only the few known details about the sub, could a nation of people answer the question as well as a handful of experts? No. But it did take more than one expert.

When you want to use a crowd to beat an expert, make sure these four tenets all hold true:

  • Decentralized: No one dictates the answer.
  • Diverse: Different people bring different information to the table.
  • Capability of summarizing opinion: There is an ability to arrive at a verdict based on the crowd's answers.
  • Independent: Each person mostly pays attention to own information.

In the Jelly Bean scenario, each person has their own approach to counting and estimating the beans. If they shared their ideas before guessing, they would become less effective as a crowd. In the submarine example, the same held true but the difficulty of the decision demanded that the "crowd" be a small crowd of experts (more like a VP level management decision).

So, for a little perspective in this matter, it's important to understand the problem you want to solve, determine whether you need a crowd or an expert, then locate the right crowd or right expert.

Page Top