IF YOU thought you could protect your privacy on the web by lying about your personal details, think again. In online communities at least, entering fake details such as a bogus name or age may no longer prevent others from working out exactly who you are.
That is the spectre raised by new research conducted by Microsoft. The computing giant is developing software that could accurately guess your name, age, gender and potentially even your location, by analysing telltale patterns in your web browsing history. But experts say the idea is a clear threat to privacy - and may be illegal in some places.
Previous studies show there are strong correlations between the sites that people visit and their personal characteristics, says software engineer Jian Hu from Microsoft's research lab in Beijing, China. For example, 74 per cent of women seek health and medical information online, while only 58 per cent of men do. And 34 per cent of women surf the internet for information about religion, whereas 25 per cent of men do the same.
While each offers only a fairly crude insight, analytical software could use a vast range of such profiles to perform a probabilistic analysis of a person's browsing history. From that it could make a good guess about their identity, Hu and his colleagues last week told the World Wide Web 2007 conference in Banff, Canada.
Hu's colleague Hua-Jun Zeng says the software could get its raw information from a number of sources, including a new type of "cookie" program that records the pages visited. Alternatively, it could use your PC's own cache of web pages, or proxy servers could maintain records of sites visited. So far it can only guess gender and age with any accuracy, but the team say they expect to be able to "refine the profiles which contain bogus demographic information", and one day predict your occupation, level of qualifications, and perhaps your location. "Because of its hierarchical structure - language, country, region, city - we may need to design algorithms to better discriminate between user locations," Zeng says.
However, Ross Anderson, a computer security engineer at the University of Cambridge, thinks the idea could land Microsoft in legal trouble. "I'd consider it somewhat pernicious if Microsoft were to deploy such software widely," he told New Scientist. "They are arguably committing offences in a number of countries under a number of different laws if they make available software that defeats the security procedures internet users deploy to protect their privacy - from export control laws to anti-hacking laws."