Our presence in the Middle East has always been questionable — but recent evidence has surfaced making it even harder to justify what we’re doing. Thanks to the Snowden documents, new information has come out regarding the NSA’s SKYNET program: that it’s been used to deploy drones over Pakistan, and gather information on potential terrorists. The drones use a machine learning algorithm to parse the cell phone records of over 55 million Pakistani citizens. It then uses this information to run calculations on a person’s likelihood of being a terrorist, based on information of confirmed terrorists, fed to the machine beforehand. If someone comes up with a high enough likelihood, they’re labeled a terrorist, and become a potential target for assassination.

This is ludicrous enough in it’s own right, but even more ludicrous is the fact that this “terrorist algorithm” might not even work right, with data science expert Patrick Ball referring to the algorithm as “completely bullsh*t”. The simple explanation is that we don’t have enough specific information on enough terrorists to teach the computer to accurately single out potential terrorists. The estimated success rate of the algorithm is currently around 50%, and while attempts are made to compensate for it, the number of false positives is still massive.

So, what does this mean? The amount of people America has killed in Pakistan, with airstrikes or otherwise, is massive. With this new information on SKYNET and its drones, we now know how the USA decides where to strike, and as a result, have more accurate estimate of how many innocent people we may have killed — and that estimate is in the thousands. The war on terror has been hotly debated since it began, and it will be interesting to see how much the fervency of this debate will increase based on this new information.