This in-depth and fascinating article on the development of and current inner-workings of Google’s auction system: AdWords (the links that show up on the right side of your google search results). The auction for the ads is instantaneous, self-serviced by companies, using a variation on second-price auctions, indiscriminate of company size, rating the quality of the ads, and also earned Google $21 billion last year. Google also uses the vast amounts of data collected to constantly improve searches as well as advertising campaigns.
Auctions draw upon collective intelligence since each bid represents an individual’s value placed on the item and winner is the person who values it the most, therefore attempting to maximize producer surplus and capture consumer surplus. In this scenario, bidders worry their bid could be much higher than all others’ and would be inclined to bid lower than their actual value placement. However, Google’s auction system sells the item (ad) to the highest bidder but at 1 cent above the second-highest price, thus attaining truer values in customers’ bids.
The working paper explores 250 examples of diverse uses of “Web enabled collective intelligence.” The workings of Google, Wikipedia, Threadless, Linux, Digg, Amazon, and blogs (like this one) are studied and broken down into answers to four key questions: Who, Why, What, & How.
With traditional organizational structures, individuals wrote articles, designed shirts, etc because they were paid to do so, but many of these companies, such as Wikipedia, run on people’s unpaid efforts, instead motivated by what the researchers call “Love” or “Glory.”
I found the example of a website called Marketocracy particularly interesting because every investor (they have 85,000) can make one or more portfolio(s) (100,000) and they select the top 100 performing ones as their m100 Index. According to their website, “the m100 Index has beaten the S&P 500 Index in 8 of the 11 quarters since inception, with a beta of 0.53 compared to the S&P 500’s beta of 1.00.”
The working paper can be downloaded for free.
MIT’s Center for Collective Intelligence defines Collective Intelligence (CI) broadly as “groups of individuals doing things collectively that seem intelligent.” This differs from groupthink (i.e. voting along party lines, investors getting caught up in speculative bubbles) because the individuals following groupthink are not making decisions based on personal knowledge and research, but instead on others‘ actions and rhetoric.
Here is a background article on swarm theory by the National Geographic: http://ngm.nationalgeographic.com/2007/07/swarms/miller-text
This site is great for aggregate information on the distributive behavioral model (including applications to physics, mathematics, biology, etc): http://www.red3d.com/cwr/boids/
MIT’s Center for Collective Intelligence is doing interesting research on current, and potential, organizations that draw upon collective intelligence: http://cci.mit.edu/