Big Data In, Big Threat Out

Doctor Lydia Robles was confused to the point of immobilization. She kept going over the results again and again and all the data pointed to an outbreak. A serious outbreak. An outbreak that had the potential to become a pandemic. When she made the recommendation to relocate critical medical resources and supplement vaccine stockpiles in the New York first responders command centers she had no idea that DHS would stick their nose into it and decide to move the entire southwest vaccine stockpile to New York.

But someone did.

She didn’t think her work was important enough to attract the attention of any surveillance programs, but based on what she’d heard from the Snowden leak, she guessed that what got watched by the NSA didn’t have that high a bar to go over any more. Someone must have looked “over her shoulder” and decided that it was worse then she had initially thought. But still, someone should have talked to her. There was something not quite right about it. Something that a big data analysis couldn’t see but something that her human intuition told her was just a bit off.

With a precision that rivals an Army Ranger strike force, as soon as the trucks were unloaded in New York, the pathogen started to make itself felt in San Diego. The only correlating event that they could trace so far was that many of the people clogging the emergency rooms and morgues were at a Chargers game the day before. Estimates said that it would take 18 hours just to get the vaccine stockpile back to that part of the country and another 12 hours to get in a position to administer it. This meant that they were already 2 days behind a virus that made HPAIV look tame. That was translating into another problem: where to put the bodies.

This thing was spreading fast. As fast as the model said it was supposed to spread in New York. But the data in New York was clearly wrong. Someone had salted the New York data in order to make the CDC think there was a problem. This was intentional and it was an attack. But how and who? And even more curious, why?

Whoever they were, they had significant resources and the patience of Job. Linda wondered how many computers had been secretly hacked to create the thousands upon thousands of web accesses and tweets over the past few months that were required to influence her analysis.

One thing was sure; Big Data turned a corner this week. A very dangerous and deadly corner. When she crawled out of bed this morning she no idea that her passion for data analysis would be the cause of so much misery and death. Garbage in, garbage out was the old rule. Not any more! Now, garbage in, Big Garbage out! Linda knew that her job would never be the same again. Big Data as a weapon. Shit. She was going to have to figure out a way to add a pedigree to her data in the future.

That is, assuming there was a future and she still had a job in it after this was all over. Or worse, if she was even still alive by then.

So I ask you, what went wrong here? Sometimes failures aren’t technical in nature. Yes, the technology can seem to fail us, as it might have here. But even the best technology can’t save us from poor reasoning and sloppy processes. Sometimes the assumptions we make are at the root of the failure we’re trying to analyze. As was the case here, sometimes the very behavior of the underlying technology can be used as a delivery method. The tech was just a bit player here. Also consider that the very behavior of the people and the processes they employed were targeted in very subtle ways.

And finally, what could have been done differently?


2 thoughts on “Big Data In, Big Threat Out

  1. A phone call is always a good way to start. We often find ourselves exchanging data in ways that decrease the richness of the information. In the scenario above, it would have been great for Lydia to call up a local hospital or the Public Health Department in NY to find out what they were seeing on the ground. This would at least have validated the need for the drugs before getting FedEx to come pick up the packages.

    My favorite line in here is “Garbage in, Big Garbage out!” Ignoring the malicious actors, this is still a critical concept to promote. In healthcare a nurse might always select the first drop down option when working up a patient because she feels it is an unimportant field. A future big data scientist without knowledge of her practice might find an astounding correlation between that junk data and incidents of cancer where no real causation relationship existed.

  2. A phone call would have been a good idea assuming that she was in control of how the data results were being used. Keep in mind that someone got between her data analysis and the eventual (wrong) conclusion. In many cases attackers are relying on poor communications habits and long latency times associated with human driven open loop processes.

Leave a Reply

Fill in your details below or click an icon to log in: Logo

You are commenting using your account. Log Out /  Change )

Google photo

You are commenting using your Google account. Log Out /  Change )

Twitter picture

You are commenting using your Twitter account. Log Out /  Change )

Facebook photo

You are commenting using your Facebook account. Log Out /  Change )

Connecting to %s