Ideology is not a word heard often in the IT industry. And yet, understood as the “common sense” of any kind of social organisation – the fabric of ideas underlying its activities, its mental horizon – it is illuminating and productive.
Donald Farmer, formerly, and famously, of Microsoft and Qlik, is currently working on how the concept of “ideology” can be used to light up the ways in which data analytics both shapes, and is shaped by, the businesses that use it. “The distinction between business and IT is more entangled with our way of thinking about the subject than with any practical distinction,” he says. In a word, it is ideological – and not in a good way.
Farmer makes this point in a set of notes he has shared with Computer Weekly, and our sister publication SearchDataManagement.com. He is working on a book on the topic, alongside the advisory work he does as principal of his own firm, TreeHive Strategy. He presented his ideas at the recent 2017 Pacific Northwest BI & Analytics Summit.
The way a company does its accounting is “driven by an understanding of business which is ideological”, he says. “For instance, cooperatives have a different ideology to public companies. And genuinely innovative companies, like Google and Facebook, have changed the way they talk about their business by turning their users, and what they do, into the product.”
Farmer sees an analogy in business ideology with political ideology, which “tries to stop people thinking any other way”. He adds: “We run our businesses on ideologies, but people are not aware of that, so it is a step forward to just make them aware. Once we realise the way we think about our business is not set in stone, it opens up other possibilities.
“In a crisis, you’d think that would introduce space for new ideologies, but what actually happens is you fall back, you go back to basics, to core principles. What wins in those situations is the story that is comfortable, not the one that shows a way forward that is radically new.”
Farmer is interested in how forms of notation shape artistic and cultural practices – and business is a cultural practice. He gives the example of musical notation to illustrate the effects of how businesses “notate” their activities by way of different, chosen accounting methods.
Many guitarists learn to play the instrument using tablature, or “tab”, but you can also write music for the guitar using stave notation. Farmer says: “Although folk music has beat, rhythm and a key, notating music with a key or time signature enables possibilities which are unlikely to develop in the folk genre. A Bach fugue, The Nutcracker or The Rite of Spring would be impossible without notation. And jazz, Indian classical music, electronics or birdsong can be all but impossible to notate.”
“Analytics notates the business”
The same point applies to different languages. Every language is a culture, and each has its strengths and weaknesses, says Farmer. As a Scots Gaelic speaker, he notes what he calls the language’s “tremendous descriptive power, especially of natural things, and its powerful ambiguity”.
He adds: “Gaelic is often quite elliptical, not complete. It has proverbs for every occasion, but it is not necessary to spell them out. There is [in Gaelic-speaking communities in Scotland and Ireland] a common understanding that does not need to be said out loud.”
It is the same with business, says Farmer: “Analytics notates the business.” And a company will evolve forms of understanding that are embedded in its culture that do not have to be said out loud, that go without saying.
Beyond business as usual
Farmer says companies typically have four streams of development. “There is bug fixing, the roadmap of what you are going to do, the feedback channel from customers, but also what you are doing that is innovative,” he says. “If you are not doing that, you will get left behind. Someone else will outpace you.”
In that fourth dimension, Farmer does see value in Artificial Intelligence (AI) and machine learning (ML), about which there has been so much hype in the last year or so. “It is very important to cut through the hype,” he says. “Most AI has a lot of human intelligence in it. These systems are being built by the smartest computer scientists in the world with great resources behind them. AI does not have a mind of its own; it is powerfully leveraged human intelligence.”
This autumn, Farmer will speak at IRM UK’s Enterprise Data Conference in London on “The new literacy: the skills and insights you need in the information economy”.
He says that although we can all learn, as individuals, to be data literate – which he distinguishes from mere numeracy to include understanding graphs and other visualisations – the real value kicks in when the business as a whole becomes more sophisticated in its understanding of data. That means it can communicate its understanding to individual employees who, in turn, understand what is being conveyed. “It’s a collective endeavour,” says Farmer.
And being genuinely data-driven means understanding that facts have to be interpreted, he says. There is always a level of ambiguity that will become even more important to apprehend when AI or ML systems are more ingrained in organisations. “We always want the facts, but predictions always have ambiguity,” he says. “Take the US election. It was said that [Hillary] Clinton had an 80% chance of winning. She lost, but she still had an 80% chance.”
Read more about the ideas behind and around data analytics
- Heedless data collection in business analytics programmes is breeding moral hazards, according to Frank Buytendijk, author of Socrates Reloaded: The Case for Ethics in Business & Technology.
- Applications vendors have some soul-searching to do over their approach to ‘big data’ ethics and information privacy, according to experts.
- In this SearchCIO Q&A, author and data scientist Cathy O’Neil talks about the danger of trusting algorithms too much.
In Farmer’s view, a sophisticatedly data-driven business would “not be thinking in absolute numbers, but in relative terms”. He says: “For example, hitting 60% of a sales target might be fair enough in context. It’s about becoming more humane in our understanding of data. We are not Mr Spock or the grimly utilitarian Gradgrind from Dickens’ Hard Times.
“And so the role of the analyst becomes less about the numbers and more about what they mean.”
In the field of data analytics, Farmer thinks there is a common wisdom that is fundamentally mistaken – that accuracy and precision are the be-all and end-all, when “debate can be more important than clarity”.
“We all talk about collaboration, but no one says that we need space for conflict and argument,” he says. “When two departments have different numbers, the traditional business intelligence answer is to impose a centralised version of the truth, and then we are all supposed to work happily together. But isn’t it better to have the two conflicting departments have it out? Let’s use that creative dissension about what the truth is to generate a better understanding of the business. That’s how you get creativity and innovation.”