Data mining (also known as pattern recognition) is an anti-terrorist intelligence strategy. Data mining means bringing together different kinds of databases, linking them, and trying to identify suspicious patterns of individual behavior. The purpose is to prevent terrorist attacks. Suspicious behavior may indicate that such an attack is imminent, and data mining has been defended as the most important prevention tool.
If a Muslim chemistry graduate takes an ill-paid job at a farm-supplies store what does it signify? Is he just earning extra cash, or getting closer to a supply of potassium nitrate (used in fertilizer and explosives)? What if apparent strangers with Arabic names have wired him money? What if he has taken air flights with one of those men, with separate reservations and different seats, paid in cash? What if his credit-card records show purchases of gadgets such as timing devices? (source)
Intelligence services are routinely bringing together different data-bases such as
- credit card and payment data
- travel data, flight reservations, hotel reservations
- census data such as race, religion, occupation
- data on internet use, email and phone use
- police information such as convictions, known associates, fines for illegal photography
- CCTV data (e.g. from cameras situated close to extremist mosques)
When all these data bases are linked, the value of the information they contain increases significantly. When intelligence services learn that someone travels to Afghanistan or Pakistan, it may not ring a bell. And anyway, there are too many people travelling to these places. But if they filter on those people who also read extremist websites, visit extremist mosques, have extremists associates, travel in the plane as other suspicious persons, have suspicious payments etc., then they may be on to something.
However, data mining may lose some of its benefits to the extent that terrorists are aware that this is happening and fine-tune their behavior. They know that much of this data mining is, by necessity, automated in computer programs which assign a “suspicion value” to certain activities or combinations of activities (the data bases are too big to do it any other way). So they can lower their suspicion score by regularly visiting very non-Muslim websites such as porn sites, or call telephone numbers of brothels.
Data mining may lose it effectiveness and has also been criticized as an invasion of privacy and a criminalization of behavior that is perfectly legal even if sometimes somewhat strange. Normally, invasions of privacy such as phone-tapping require a prior suspicion. Data mining means spying on people without such as prior suspicion, because suspicion can only be established as a result of mining, not beforehand.
It means spying on people in two different ways:
- The data bases used contain individual data that were not intended for use by intelligence services and that were often handed over by individuals on the assumption that the data would be treated confidentially.
- When a suspicion is established on the basis of data mining, more traditional means of surveillance come into play (phone tapping, observation etc.), which also violate people’s right to privacy.
Of course, privacy is not an absolute value and different types of rights need to be balanced – in this case the right to privacy of some and the right to physical integrity and security of others. And an assessment has to be made of the priority of one right compared to another. But it seems to me that treating every citizen as a possible terrorist, and looking at his or her individual and private data as a matter of routine, is way over the top. I would like to see some information of the number of terrorist plots foiled by data mining.
But privacy is not the only victim. If someone is labeled “suspicious” as a result of data mining, he or she may end up on a “watch-list” and may find it difficult to travel or find a job in certain places deemed risky (hospitals, fertilizer producers, government agencies etc.).