The End of the Big Data Fad ~ Introducing Data Flow
If I were a venture capitalist, which one day I hope to be, I wouldn't fund companies and start-ups that process reams and reams of Big Data or Dark Data. Big Data, as we know it, is a flash in the pan, and it will disappear just like the Atari.
Yes we will have the internet of everything generating more data than anyone can handle. Yes we will have data generated by ever single one of the billions of humans inhabiting this planet. Yes we will have data generated by the trillions of devices that are or will be connected. But gathering huge lots of data and then batch processing it, is an unsustainable model.
When I was just starting out adult life, one of my neighbors was a draft dodger from the Vietnam War. He was/is a pacifistic in the positive sense. He didn't trust the motives of the US government and the McCarthy Communist witch hunts, and his buddies dying in foreign jungles for an unfathomable reason, in a war that they couldn't win. So he came to Canada. He has a degree in civil engineering, but he landed in Silicon Valley North and started working for a start-up. It was an exciting time. The computer was becoming ubiquitous, and almost every industry was crying for some sort of computerization in an age where there weren't any off-the-shelf software packages.
He joined a start-up and his job was to write the software for a data digestor for a shoe manufacturing company. The company would do a run of shoe components (soles, uppers, large, small, all sizes, ladies, men's, children, brown, black, purple) and they didn't know how many parts of what. They generally ran a line until they ran out of component pieces, and then laboriously switched over to another make, color and size.
It was the perfect application for a data digestor. Every time a component batch was made, someone would swipe a pre-programmed card and the result would go to a collector, and then to a database that could be queried, and management could do some actual planning by matching what components they had and knowing the manufacturing run limit. Big data wasn't very big then, but it was still an issue.
After the data digestor was delivered, the start-up ran through all of their money including the shoe company money, leaving my friend as the last employee. His wife was a registered nurse so he lived off her income while he refined the data digestor. He worked for shares of the company abandoned by the other employees. He was paid his putative salary in shares and the shares were valued at ten cents. After a year of refinement and frugal living, the company was bought for its data digestor, and my old neighbor is wealthy to this day.
All this to say, is that Big Data is going to grow exponentially, and we can't handle it like we are doing now, with old paradigms. Big Data has to be digested, mined and made sense of when it is created. It can't be allowed to accumulate otherwise when we get around to it, the effort will be akin to emptying the ocean with a teaspoon.
Because of these old paradigms still in play, in the long term, I would short the stock of SAP and SAS and all of the old-school stuff. A better way will come along, and just like the tire-cord industry disappeared when Michelin invented the radial tire, and we will have a new something-else.
So if I were a venture capitalist, I would fund and bet my money on creative, innovative paradigms that for men and machines that made sense of data as it was created. If we don't develop a sure-fire universal way of doing that, in a few short years, even if every molecule of silicon in every mountain in the world was transmuted to semi-conductor memory, it still wouldn't be enough.
A bigger philosophical question is, "How much of this data is valuable?". That question will be answered by clever minds who can monetize pieces of it. The economic world is very Darwinian.
As for Big Data, we will bury that sucker. Its child will live on though, and we will call it Data Flow. Data Flow software will be ubiquitous and highly necessary,