I recently saw a claim that data scientists should be statisticians, experts in machine learning, coders and business people – all rolled into one. A little humility please. Unless data scientists are going to be extraordinarily bad statisticians, machine learning experts, coders and business people, then it would be wise to remember some traits associated with our status as mortals. Becoming expert in any one, or at most two of these things is a lifetime’s work and should not be dismissed so easily.
Some years ago CSC issued a paper titled ‘CEOs are from Mars and CIOs are from Pluto’. It encapsulated the domain gap that existed between IT and business very well, and was obviously the inspiration for this piece. Since this paper was published things have only got worse. We’re on the verge of using functional languages, originated from Lambda calculus and certainly totally meaningless to a business person. But if we are ever going to get a grip on coding complexity then functional languages represent the best chance of achieving this aim. They are extremely formal and formality is the complexity killer – the object oriented paradigm really doesn’t stand a chance – but I digress.
We should spare a moment’s thought for the business managers. Half a century ago their job was fairly straightforward – buy stuff, sell stuff, account for stuff and so on. And each department would have a number of clerks whose job it was to make sure that records were kept in neat filing cabinets. And then computers came onto the scene which required specialists, who might have very little interest in the business, but a great deal of interest in coding, hardware maintenance, and of course their careers. It was, and still is a love-hate relationship. Business managers need IT to reduce head count and make their operations more efficient, but since they generally have little understanding of the technology they tend to be mistrustful and not a little resentful. Yes I know we are not supposed to say these things – but this is the reality.
As if having an IT department that speaks a different language and has its own agenda is not enough, the poor beleaguered business manager now sees a new spectre on the horizon – the data scientist. IT people may talk about polymorphism, encapsulation, ACID transactions and the like, but data scientists talk about support vector machines, additive models, feature engineering and other gobbledygook. The difference and the commonality between IT and data science is that one is largely concerned with data technology and the other with data (although I think the term information scientist might be more apt). Traditional IT was largely concerned with electronic filing cabinets and glorified calculating machines, while data science is concerned with patterns of behaviour and ultimately, improvements in decision making in the face of uncertainty.
The business will no longer just have one department that it desperately needs but hardly understands, namely the IT department, but two, with a new set of people who speak a foreign language. And the languages of IT and data science are largely different too. These are early days for data science, and very early for the data science department – or whatever it is to be called. There will undoubtedly be different roles – some nearer to the business and some quite isolated. The technology will also differentiate with some tools friendly enough for business users not to feel totally isolated, but others that need a PhD+ in order to extract hitherto unseen patterns. Just as the IT department has its business facing side and the inward looking technical side, so the same will inevitably happen with data science.
All of this raises a fairly fundamental and uncomfortable question. Which part of the beast is the dog and which part the tail? The new breed of large, profitable corporations tend to be information based, and even the simplest business is now keenly interested in social data, a web presence, analytics and intelligence. All businesses are becoming information businesses to a lesser or greater extent. So does this imply that the business manager needs to be information savvy? Well, yes of course it does, and the more information savvy the better. The tail is truly starting to wag the dog, and in the process is becoming the dog.
Mars is a long way from Betelgeuse and the new troika of business-IT-data science is the shape of things to come, with the inevitable tensions and misunderstandings that come from disparate interests and agendas. But this evolution in business will not stop here, and the next development, which will be cybernetic in nature, is not even a glint in the research lab’s eye. And maybe it is just as well.