Friday, December 15, 2006

Inkling Presents at Yahoo's Prediction Market "Confab" and Additional Thoughts

A couple nights ago I was asked to give a 10 minute presentation as part of a Yahoo "mini-conference" on prediction markets. The event was held in a classroom on the Yahoo campus and I believe I heard close to 300 people showed up. Yahoo did a great job of putting it together. Unlike other conferences or large meetings I have spoken at in the past in my previous life as a consultant, I was actually told where I needed to be at a certain time, there were no snafus in the logistics and there was plenty of free food afterwards...So thank you David Pennock, Chris Plasser, and others from Yahoo for putting together a great event.

The agenda for the evening can be seen here. There is also a replay of the entire event on video.

In the 10 minutes I had, I mainly focused on what we had learned watching people try to run their own markets, both in the public space and among our client-base. Here were the top 5 reasons markets fail:

  • question bias
  • asking question with no verifiable outcome to cash out the market
  • market description is poor or confusing with lots of loopholes; outcomes are not well defined
  • timeframe is unrealistic
  • no information is available for traders to make a reasonable judgment

It was great to meet the people we hear so much about, including James Suroweicki himself, who gave away 75 books at the event and was a very down to earth, nice guy.

After the event, I also had an opportunity to meet many of our users and people who are currently running markets on Inkling, along with a few of our clients I had never actually met in person - an added bonus.

As I was thinking more about the conference on my flight back to Chicago, I was thinking about a particular question that was asked where time didn't warrant enough discussion. The question was in relation to running prediction markets in companies and who should participate in the markets that are set up.

In many of the scenarios presented at the conference and other case studies we've all read about, a market has usually been run as an experiment with a control group. And usually when we first talk to potential clients they want to run their markets in a similar fashion: invite a group of people from a particular division, set up a market for them, have them trade, and look at the results. Part of this is because you need to have people with knowledge of the market to make informed trading decisions. But it's also, I would argue, because that's how business is already structured: lets give this to the "product group" to play with, or the "finance division."

But this is constrictive to the confines of old organizational thinking. For one, the diversity of knowledge inherent across the entire employee base is lost. Prediction markets can be a catalyst to break down stovepipes and hierarchy to simply try and get the best answer possible. Second, Suroweicki said in one of his comments that prediction markets have moved too slowly in the corporate space. We completely agree. Running prediction markets at all is a good first step, but hopefully just a baby step towards a much larger role for prediction markets in the workplace.

All companies now have email as a productivity tool given to their employees, along with an internal portal, collaboration software, etc. Prediction market software should be as prevalent and an integral part of the business processes of any organization. Everyone in the company should be a trader. Everyone in the company should have the ability to run a market. It's why we rant and rave about our "do it yourself" model and stress so much about the usability of our software! This scares a lot of people because of what may be unearthed when people can ask any question they want, but I would encourage anyone to go back and listen to the excellent Microsoft presentation by Todd Proebsting at the confab where knowing that everyone thought a project was going to be late was scary, but also allowed them to actually address the problem and cut features to make the date. No one got fired, no one got yelled at; the problem was identified and the talented professionals there worked to fix it.

Management 101 says you tell your people to let you know about problems as early as possible before they become unmanageable. Risk mitigation is key to successful projects and that's exactly what one of the roles a robust marketplace should play for a company in addition to making forecasts: serve as an early warning indicator so problems can be solved before they become detrimental to the performance of the business.

In Alcoholics Anonymous, you have to be able to admit to and talk about the problem you have before you solve it. Isn't this also what companies need to be able to do more of as well to be more successful executing?

No comments: