The blog, Midas Oracle, looks at prediction markets in the news and issues surrounding their value and usage and contains many current articles about different markets and their social utility.  The author examines the utility of prediction markets, and determines that it is in fact their ability to synthesize facts, expectations and beliefs gathered by a diverse audience of users much faster than traditional media outlets can that makes them a valuable means of forecasting events.  It is arguably this efficiency that should be focused on by developers and those looking to harness the collective intelligence of prediction markets in order to develop new ways of making prediction markets more effective.

    Putting Suroweicki's qualifications of a diverse, decentralized and independant crowd in the context of their ability to amass this information most efficienctly is a useful method of understanding their significance.  It seems to make sense why these characteristics would be useful for prediction markets.  Another interesting aspect of this blog is that it does not hold the crowds on a pedestal for their supreme intelligence, but rather values their speed and popularity as their defining characteristic.  This follows more in the line of conventional wisdom, and is an important concept when examining the role of these markets in the foreseeable future.

    This article examines the Hollywood Stock Exchange and gives basic information about its usage and what happens with the collected data.  In the case of HSX, the value lies not in the prediction of the future, but rather the accumulation of the preferences (tied to the demograpic information) of the average users.  That data is then sold to production companies who can adjust their upcoming films, and determine more intelligent money allocation based on their consumers.  This model works two ways, predicting the value of actors, producers and movies and also informing the film industry about what the average consumer desires. 

    While the Hollywood Stock Exchange does not release their demographic information on their website, they operate an entire research end that enables studios to selectively purchase certain types of data collected.  In this case it is important for the average person to join in on the trading, not only to get data on the target audience but also to give reason to attract experts, who see opportunity to do well.  These experts' opinions are important and might not be harnessed in other more traditional methods.

    This article looks critically at the state of current regulation on insider trading, and its traditional role of encouraging investment through assurance that no one is up against someone with special inside knowledge.  However, when looking at the role of insider trading in prediction markets, regulation could be inhibiting a very valuable addition to reaching useful accurate results in these markets.  The task of managing important corporate information has traditionally been kept to the very elite few, but technologies enabled by the internet have begun spreading the idea that this process actually stifles innovation.  It is quite evident in this case that regulation about insider trading conflicts with any attempts to harness collective intelligence, even in a way beneficial for both the business and investors.

    This article explores another facet of prediction market communities that I believe has been kept largely under wraps.  It is unclear how much insider trading goes on in prediction markets, though it is plain to see according to this article that it probably shouldn't be discouraged.  Especially on play-money markets where the only benefit is the social value of predicting the future, these annoymous markets should be an opportunity for those with any bit of information to come forward with if they choose for the benefit of the whole.

This paper examines five significant issues with prediction markets and attempts to solve them in order to determine if the current level of enthusiasm for their potential is appropriate or not. One of the points argued is that prediction markets depend heavily on the "uninformed trader" who is ironically the hardest type of person to attract. They recognize that play-money markets have been found to predict as effectively as real-money markets, but argue that the issue of attracting this type of person still exists. Finally they conclude that offering sports or entertainment betting, subsidizing the investment and personal career concerns have been chief effective motivators in the past for this sector of trader.

While this article also examines other issues with prediction markets, the question of the "uninformed trader" is one crucial to the argument of the diverse crowd and choosing the right incentives. They develop an interesting conclusion on this issue, that is not really repeated in any other literature on the subject. However, since in the case of the markets I am studying, the entertainment value already exists, and there is no way to purposefully boost career concerns, it seems that subsidizing the investment is the only plausable incentive they seem to offer to reach more diverse crowds. My question from this article is why they would even worry about conducting the market in real money if play-money games with prizes seemed to work just as effectively?

This article challenges the conventional belief amongst economists that "markets where traders risk their own money should produce better forecasts than markets where traders run no financial risk." In actuality, real money markets reflect more than past predicitive performance, they relfect an entire mass of social factors behind individual wealth, like financial status and willingness to take risks. Play-money prediction markets are based solely on track record and previous predictive performance and many systems use this model in addition to prize incentives based on rank to ensure players continue to buy and sell. The primary research of this paper is a study conducted to determine the level of accuracy sacrificed when using play-money compared to real-money prediction markets. The conclusion reached was that both markets were almost identical in accuracy.

It appears that though the predicitve power of these two real and play money markets are about equal, that the play-money one would actualy average a closer fit to Surowiecki's opinions about what constitutes a good market by eliminating the discouraging financial factor. This study is crucial in the examination of incentives as it attacks a noticible divide within the prediction market world.

    This Wired Magazine article examines the new ways that companies have begun using the internet to harness a type of collective intelligence.  Entitled 'crowdsourcing', this method is employed by companies large and small to do tasks better and more efficiently through harnessing a large amount of people.  The key is that for the most part, it works out well for consumers as well.  An amateur photographer who might never get a dime for his photos can now continue to take them on the side and potentially make a couple bucks from printers looking for cheap photos through sites like iStockphoto.  The consumer in this case has large enough incentives to participate, and the corporation has even larger reasons to turn to the same model. 

    While this article describes a very different process from prediction markets, it is an interesting comparison as to how corporations (and consumers) can harness and profit from these aggregators of collective intelligence.  Of course the models described in this article are much easier to adopt because they don't necessarily undermine the expertise of others, they simply look to "fill in gaps" that the market has been unable to reach.

    This compilation of research based at the Center for Collective Intelligence at MIT sets up basic information about collective intelligence, its history, how it's used and what factors facilitate and inhibit it from functioning to its best potential.  Collective intelligence is harnessed and used by business organizations, computer science and AI, nature and prediction markets.  Factors that facilitate it are diversity of opinion, informal structure, shared vocabulary and infrastructure, intrinsic motivation and monetary incentives.  Its greatest inhibiting factors are biases, the bandwagon effect, homogeneity, polarization of the group and cultural boundaries that produce outlying data. 

      This site attempts to harness collective intelligence to accumulate the best info and research on the topic, and for that reason I felt it was an important addition to a paper on harnessing collective intelligence.  Besides a very detailed aggregation of information the Handbook provides interesting facilitating and inhibiting factors that serve as a contrast for those provided by Suroweicki.

    Collective Intelligence is described here as a tool to be harnessed, since crowds can also have negative or difficult characteristics in most other contexts and are generally not preferred to deal with "directly".  The formed collective can almost be thought of as a almost a distinct individual or expert according to Watkins.  Prediction markets here are examined as "sophistocated aggregation tools" bringing together communities of self-selected individuals who already perhaps have an emotional investment in the issues.  Watkins is also concerned with issues of trust and how to cultivate public trust in collective intelligence as a reliable source of information.

    Watkins touches on the characteristics of particular communities and the individuals who are drawn to prediction markets, an area that I plan on focusing on more intesively through the study of forums and demographic data on each of the particular sites.  In additon, the notion of trusting this collective prediction over that of the experts interacts interestingly with Surowiecki's theory that they experts often partake in the predicting.

James Surowiecki outlines through cultural examples the basic idea of collective intelligence and that time after time it is shown that the group estimate to a given problem is more accurate than most of all of the individual guesses. He then lays out a key set of characteristics for what makes a wise crowd. They are: diversity of opinion, meaning that users come from many different intellectual backgrounds; independence, that they do not rely heavily on the opinions of others; decentralized, meaning people can draw upon local knowledge, and a means for aggregating their opinions. He states that while the average is often mediocre in most cases, in decision making it is most often the best. Though he also states that collective intelligence is not always perfect, citing certain examples where experts certainly know better than the crowd.

Surowiecki lays out the particular parameters I will esxamine as a starting point in my research project in looking at the particular incentives of marketing techniques used to maintain a flourishing prediction market. The implications for data gathered from prediction markets will examine the future role of experts, building off of Surowiecki's comments. In addition this work will mark the branching off point from where I examine other definitions of "wise crowds" and how important his characteristics actually are.

    Information systems like the internet provide a medium for enhancement of the group dynamics that have been present in complex and lossly organized groups for hundreds of years.  Over time the average "social group" size has increased dramatically, allowing for a higher level of diversity of opinion.  The internet allows for the integration of knowledge in many different forms, has remarkable capacity for storage and can transmit data with low data loss, if any.  These qualities improve upon the merely human knowledge exchange and aggregation systems formed organically through social groups.  This article also examines the possible damaging factors to large knowledge systems, citing that many individual "wrong" choices would not greatly affect the outcome, but a randomly selected "leader" would be incredibly detrimental to the collective result.

    This article examines the very relevant characteristics of the internet in enabling collective intelligence to grow and thrive.  This is important in putting the study as well as prediction markets themselves in context, since they do not arise simply on the internet, but are created when certain conditions are optimal.  In addition, this article looks at the important inhibiting factors to an accurate concensus like the presence of a leader and potential for the bandwagon problem.