The first is tracking flu trends through the analysis of Google search query data. CDC and Google researchers discovered that there is a direct correlation between flu related search queries in Google and an eventual flu outbreak. Even more interesting, the query data precedes the data collected by health care providers, meaning outbreaks can be detected in an area before they are realized by the healthcare system.
The second use of aggregating human thought is the correlation between twitter and stock performances. Economists at the Technical University of Munich (Technische Universitaet Muenchen, TUM) discovered that twitter buzz is an accurate predictor of stock performance, with twitter buzz leading the market by about a day. The scientists have created a website where you can go to see what their model is predicting based on real time information.
These devices have important implications for how knowledge is realized. The internet is making the process less formal, and more accurate. Take the declaration of a flu outbreak. The traditional system is for hospitals to aggregate information on people coming through their doors. This skews the statistics by only capturing the people who are sick enough to bring themselves into hospitals. Information is then collected from hospitals by the CDC and an outbreak is declared. This declaration requires a huge formal structure and the involvement of multiple experts. Contrast that with Google Flu Trends, where the process covers a wider data set, is real-time and automatic.
The twitter case is interesting for the same reasons as Google Flu Trends, but it also includes a tacit voting system, in which the volume of information is also a statement on the quality of information. Users that give better tips are more likely to be repeated by other users. Unlike the Google Flu Trends system, this information doesn't precede the traditional sources, it incorporates them. Information that comes from outside sources, such as financial media or stock tips from friends, makes its way back into the system through twitter users where it is put on an equal footing with all other data.
Unfortunately these systems are vulnerable to misuse, and the Twitter system be less useful the more popular it becomes. While there is little advantage for anybody to flood Google to create a false declaration of a flu outbreak, the incentive to manipulate the Twitter stock pick system will rise as the system becomes more popular. The more popular the service, the more effectively the system can move the market. It is easy to see how someone could benefit by creating a false move in a stock by injecting bad information through zombie twitter accounts.
In both cases better decisions can be made using systems that rely on what the mass of individuals are thinking, rather than the declarations of a few experts.
The twitter case is interesting for the same reasons as Google Flu Trends, but it also includes a tacit voting system, in which the volume of information is also a statement on the quality of information. Users that give better tips are more likely to be repeated by other users. Unlike the Google Flu Trends system, this information doesn't precede the traditional sources, it incorporates them. Information that comes from outside sources, such as financial media or stock tips from friends, makes its way back into the system through twitter users where it is put on an equal footing with all other data.
Unfortunately these systems are vulnerable to misuse, and the Twitter system be less useful the more popular it becomes. While there is little advantage for anybody to flood Google to create a false declaration of a flu outbreak, the incentive to manipulate the Twitter stock pick system will rise as the system becomes more popular. The more popular the service, the more effectively the system can move the market. It is easy to see how someone could benefit by creating a false move in a stock by injecting bad information through zombie twitter accounts.
In both cases better decisions can be made using systems that rely on what the mass of individuals are thinking, rather than the declarations of a few experts.
No comments:
Post a Comment