Google spiders crawl publicly available web information and our index of historic, cached web content. Using a mashup of numerous factors such as recurrence plots, fuzzy measure analysis, online betting odds and the weather forecast from the iGoogle weather gadget, we can create a sophisticated model of what the internet will look like 24 hours from now.
We can use this technique to predict almost anything on the web – tomorrow’s share price movements, sports results or news events. Plus, using language regression analysis, Google can even predict the actual wording of blogs and newspaper columns, 24 hours before they’re written!
To rank these future pages in order of relevance, gDay™ uses a statistical extrapolation of a page’s future PageRank, called SageRank.
The technology behind Gday is called Mate, which stands for Machine Automated Temporal Extrapolation.