Saturday, 21 April 2012

Web Bot Project

Web-Bot project intends to predict future events based on the information available on internet. Its creators sell predictions but keep algorithms a secret. The tool was developed to predict trends in stock market but later on they started to use to predict virtually everything.

This tool works based on the assumption that if you can sense 'the mood' of the public on internet, you can predict the future. What it does is-to monitor, quantify, assess, analyze and then predict based on various feeds into the internet stream, be they various chats, blog-posts, forums or articles.  

An excerpt from Wikipedia article on its methodology:

Internet bots monitor news articles, blogs, forums, and other forms of Internet chatter. Words in the lexicon are assigned numeric values for emotional quantifiers such as duration, impact, immediacy, intensity, and others. The lexicon is dynamic, and changes according to shifts in emotional tension, and how humans communicate those changes using the Internet. As of 2008, there were about 300,000 keywords in the lexicon, along with emotional context indicators which are fed into a computer-generated modelspace. They then use a technique called Asymmetric Language Trend Analysis (ALTA) developed by Clif High to generate a predictive report from a fully populated modelspace. These predictive reports are known as ALTA or Web Bot reports. The 2010/2011 report underwent a formatting change and is titled The Shape of Things to Come – Volume Zero.

It has been insinuated by experts that the predictions by Web-Bot consistently tend to be negative in nature. This was admitted by its creators as well. Why?

An excerpt from Wikipedia article:

Ure and High hypothesize that changes in language precede changes in behavior. This is the basis for attempting to use ALTA as a form of future viewing. The authors have repeatedly asserted in interviews that the predictions made in the ALTA report have an inherent bias toward events of a negative nature and tend to be framed in catastrophic terms. However, the reports have also indicated positive events such as the development of new technology and increased social awareness.

It's beyond my understanding how change in language over the web could predict natural catastrophes, even if you could prove that change in language is associated with change in behavior, because there is no proof that human behaviour affects nature to bring natural catastrophes. 

Even when Web-Bot has claimed some hits as their predictions, they have mostly been failures and pseudo-scientific.

Image Source: