Big Data Is Just a Big Scam

The newly emergent "big data" meme has never been defined in any meaningful and definitive way. It's the most amorphous new buzz-term that I've seen for a decade. It's one of those "eye of the beholder" terms used to liven up a seminar and eventually soak investors.

Let's start by asking what does big data mean? Lots of data? More data than you can handle? Amorphous data? Out of control data? Useful data for analysis? Useless data? Information overload?

If you read enough about big data, it is all of the above and more. The key is not the data, but the challenge of how to handle the data and what to do with the data itself.

In other words, how can we make this huge pile of data, that we have managed to accumulate, be useful in new and profitable ways? The data pools can come from anywhere via various computing mechanisms such as Facebook posts, NSA logs, mailing lists, customers, etc.

I would argue that most readers of this column are themselves repositories of big data. I just bought a 3 Terabyte drive for backup. I have a lot of data to back up! Big data!

A useful tool I can imagine would be a big data analysis tool that the police could use to find you guilty of some random crime by going through your files. Or at least find something that would embarrass you. Big data!

This all harkens back to a comment made by a former CEO of American Express, who told an audience that if the company wanted to (and this applies to all credit card companies) it could use your personal buying habits and tendencies to put together a complete dossier and definitely tell if you are having an affair or not. Big data!

From what I can tell, this is exactly what big data does best, spy on individuals.

The popular commentary by the industry is that somehow huge databases that contain information about you will be used for your benefit. Yeah, you know, so you'll get advertisements for things you are actually interested in. Targeted advertising. This benefits everyone!

Here is how this works. I'm interested in buying a new camera and hear about the new Bogus One. I read an article on PCMag.com about the camera, then I go to Amazon and check out the prices and user reviews. It appears as if nobody likes the camera. It's an obvious piece of crap. The reviewers all shunned the camera. The buyers hate the camera. This is not a camera I want. But the big data analysis software decides to give me hundreds of ads, on every webpage I visit, all showing the Bogus One camera. Big data!

It's worse than that. If things went in the other direction and I decided I liked and wanted the camera and actually purchased the camera I'd still get ads encouraging me to buy the camera. Big data!

Targeted "big data" advertising is the biggest load of baloney, ever.

OK, so let's forget that fiasco and move to using big data for trends. Big data might find various needles in various haystacks. So you'd get something like this: Forty percent of all Subaru buyers are vegetarian/vegan and 80 percent of them are Democrats. But that number increased to 60 percent and 90 percent, respectively, if we are talking about a Subaru Outback. Furthermore, it increases to 99 percent vegetarian/vegan and 99 percent Democrat if the car is painted green.

Big data theorists would have a field day at the seminar with this information. But it ignores all the variables that can change the calculations and trends, which are endless. How long will any of this particular big data hold up if the CEO of Subaru drives up to a gathering in a green Outback and declares that "All whales must be annihilated!"?

It doesn't even take that much to change the dynamics of big data conclusions. Despite what the American Express guy says, data is actually blind. That's why you get ads for products you already purchased.

Big data would have you jump to conclusions.

If someone buys ten books on how to make a bomb, does it mean he wants to blow up the Federal Building? What if he was a bomb disposal expert and that wasn't included in his job title as police consultant? Big data cannot know everything, yet. It cannot know what you are thinking. I'm thinking, for example, that the Bogus One camera sucks. Big data cannot know that unless I'm interrogated. And I might lie about it.

None of this sounds good or healthy for the society. Jumping to conclusions, making assumptions, acting on false assumptions. Hounding the public with useless advertisements.

Big data as a high concept will never fully define itself and die a miserable death. I hope.