Collective Intelligence - Regressing to the Mean?

In a recent article for Tech Central Station, Robert McHenry, former Encyclopædia Britannica Editor in Chief, writes a sharply reasoned take-down of the Wikipedia phenomenon:

"However closely a Wikipedia article may at some point in its life attain to reliability, it is forever open to the uninformed or semiliterate meddler. It is true, unfortunately, that many encyclopedia users, like many encyclopedia reviewers, have low expectations. They are satisfied to find an answer to their questions. I would argue that more serious users, however, have two requirements: first, an answer to their questions; second, that those answers be correct."

I'm reminded of the old adage that free advice is worth exactly what one pays for it. Historians and other experts are clearly better sources for established facts. The major flaw of the Wikipedia approach is a lack of any incentive - aside from finite goodwill - for those who truly do have the right answers. Why continue offering them (with an inherent cost in time and aggravation) against a relentless tide of misinformed, if not malicious 'contributions' and 'corrections'?  It all has the flavor of an argument with a two-year-old - or a teenager... [Recently, some young men in my daughter's honors history class 'revised' Wikipedia's entire history of World War II in the course of one mischievous afternoon before anyone noticed and corrected the material.]

But Mr. McHenry's reasoning fails to acknowledge that elitist approaches are often equally inadequate, (and sometimes systematically misleading) in assessing the probabilities of future events.  And amazingly, the lines between past and future often blur - especially in technology market forecasts. I know. I used to create them for two of the leading firms in that industry. You've probably seen the charts in the media: static 'data' about the next three years (e.g., pen computing, always up and to the right), seamlessly blended with data from of the last three years. What's usually absent is any mention of the massive uncertainties and fragile assumptions used to generate the rosy numbers. Furthermore, the subtle career incentive in that industry is towards creating forecasts that justify the ambitions of technology vendor clients. (When was the last time you saw a public technology market forecast that realistically acknowledged any speed bumps, much less the potential for a downhill slide?)

Where prediction markets differ from Wikipedia's regression to a messy mean, and traditional experts' forecasts of perpetual growth and sunshine is that they reward those with objectively right answers in proportion to the confidence that each one has in his or her knowledge. Even more importantly, over time, prediction markets marginalize those whose opinions prove incorrect. Or in simpler terms, they force people to put their money where their mouth is, naturally amplifying the influence of those who have a better handle on the truth.