Big data does not help in preventing terror

Doesn’t rule out innocent people
The fight against terror is already being conducted using big data, and US intelligence leaders believe that big data offers terrorist control.

But just how much data should be gathered and what should be available for intelligence work are important questions for a democratic society.

Former NSA lawyer Stewart Baker has stated that if they just have enough metadata, it will provide everything they need to know about a person’s life. “Just having enough data” can evidently provide the answer to most things.

One possibility for liquidating suspected terrorists could be to use drones to aim at and shoot people’s mobile phones. The hope is that the person near the phone is the suspect that is being hunted down.

And this is a major problem.

Big data is lousy at ruling out innocent people. If innocent people happen to be sitting by the targeted phone, they can be killed too. This may end up creating even more terrorists.

Statistics and the terrorist hunt
Sætnan believes that big data is not suitable for tracking down terrorists before an attack takes place.

“Technology optimists who believe big data can solve most of our social problems need to learn basic statistical rules and recognize the obvious limitations,” she says.

Sætnan has long experience with large amounts of data and many years as a surveillance researcher behind her. She knows the problems that arise when we mix data sets, which you can read more about in the fact box.

Large datasets are usable when we want to improve the target percentage by a few percentage points. They can also be good at highlighting possible patterns to research further.

Big data is useful and can be used in medicine, marketing, product development and other areas where we’re not looking for a needle in a haystack, but trying to find clusters and patterns.

“But if we’re going to use big data to capture terrorists, we’ll need to redefine terrorism to apply to groups of people with a certain background, and we’ll have to expect a number of innocent people to become victims in the fight against terrorism.

Big data and the hunt for terrorism
Ann Rudinow Sætnan believes the following issues stand in the way of using big data in the terrorist hunt:

·  Accuracy is poor. It’s like trawling for fish when a harpoon is really the right tool.

·  All people have biases, so there is no such thing as raw and independent data. The creator of the questions influences the data and questions we ask.

·  Hacking, fake identities and errors in proofreading mean we cannot trust the data completely. The more data there is, and the faster it flows in, the harder it is to fact check.

·  Merging data sets changes the context and meaning of some information, making the information less accurate.

·  Increasing the amount of data does not help, because then the accuracy is reduced.

·  Merged datasets contain many coincidental patterns that do not make sense. There is even a separate website for such spurious correlations.

·  Data collected for one purpose cannot simply be used for another purpose, because it can end up being both misleading and illegal.

·  Big data can provide good answers when wrong answers have only minor consequences, but it is dangerous to make mistakes when searching for terrorists, which can result in either overlooking a terrorist or misidentifying and killing an innocent person.

— Read more in Ann Rudinow Sætnan, Ingrid Schneider, and Nicola Green, eds., The Politics and Policies of Big Data: Big Data, Big Brother?, 1st Edition (Routledge, 2018)

The article is published courtesy of Gemini, which publishes research news from NTNU: Norwegian University of Science and Technology and SINTEF