A couple of days ago I posted this from The Industries of the Future by Alec Ross:

“Whereas land was the raw material of the Agricultural Age and iron was the raw material of the Industrial Age, data is the raw material of the Information Age.”

Having read a bit further into the book, and not able to get off it off my mind, I want to revisit this. Because there is a very profound, if subtle, shift going on here. Rightly, Ross labels all of these inputs as raw materials. But that is where the similarity ends. Two of them are naturally occurring, requiring only some capital and ingenuity to commidify. They are also finite, and subject to the laws of diminishing returns. But the real barrier to obtaining and exploiting land and iron (etc.), is money. And at least in recent times, this has not been so hard to come by for someone with the necessarily skills and business model.

Data, on the other hand, is a whole different animal. It is an entirely human construct, largely the byproduct of our behaviour and actions. Often contributed freely. Datasets are highly differentiated, perfectly excludable and the benefits compound. And more than anything, the fields that lend themselves to data are subject to first mover advantages.

Let’s take Facebook for example. It is hard to imagine an entity possessing a better dataset on the human race. And the better Facebook’s data gets, the more it can customise and tailor the product, encouraging more use and more data to be contributed, and on, and on. The same can be said of something like driverless cars. Google has amassed an ungodly amount of data through strategic projects, like Google maps. Recently, we are seeing the likes of Daimler and BMW attempt to create a consortium to catch up (in the driverless car space), but its success looks doubtful. Google’s lead is incredible, its incentives aligned.

The massive advantage afforded by datasets is evidence in another area – artificial intelligence. Recently we have seen Facebook and Google (among others) open source certain aspects of their operations, notably designs for Ai infrastructure. Cooperation will certainly achieve benefits for the whole. But the data, the secret source, is being kept back. What would once have been incredibly valuable, technical designs requiring capital and ingenuity, is worthless compared to the information we freely and willingly hand over to these companies.

I guess what this boils down to is that land and iron (etc.) are commodities. Companies that operate in these industries are mostly operating in perfectly competitive markets, advantages (if any) usually deriving from scale. The world of data is one of natural monopolies. While BHP’s average cost for a tonne of iron goes down with scale, quality does not improve. Google search just keeps getting better and better.