The Age of Unintelligence
Just because you have a data lake measured in yottabytes, it does not mean you have anything of quality, writes David Brennan
I’ve been merrily quoting Arianna Huffington in recent months (“we are drowning in intelligence but starved of wisdom”) in order to explain BE Insight’s approach to the role of insight in the digital age. Unfortunately, at its heart, there is a serious flaw in the sentiment.
The quote assumes all data can be categorised as ‘intelligence’ but there is a real dichotomy at play here; not all data can be classified as ‘intelligence’ and not all intelligence consists of what we would normally call ‘data’.
Indeed, the notion that all of the data swirling around us can be considered as intelligence is one of the formative factors behind the emergence of what I’m going to refer to as The Age of Unintelligence.
Which is where we are right now.
We can see it in the political ‘debate’ around Brexit, or gun control in the US, or Russian influence on social media…or, indeed, any debate where emotions get fired up and people seek their own evidence to reinforce and validate their latent biases.
Facts get thrown around, regardless of their veracity (or even the slightest evidence supporting them). The political becomes personal. Expert opinion is derided if it doesn’t fit the desired narrative. Complex issues are reduced to simplistic choices.
Intelligence becomes unintelligence.
It’s a trend that I have also seen reflected in the debate about the role and value of digital marketing within the overall mix. Despite (or because of) the huge amounts of data we have to base our decisions on, actions do not appear to relate to the evidence.
[advert position=”left”]
So, what do I mean by unintelligence?
In my opinion, it is our collective willingness to relinquish our capacity for critical judgement based on verifiable facts in order to benefit from the superior decision-making prowess of a ‘higher being’. In the wider world, that might be a religion or a concept (“take back control”) or a populist leader like Trump or Putin. In our industry it is the lure of big data, AI and the almighty algorithm.
It is trusting the machine to make the correct decisions, without being sufficiently concerned with how those decisions are made.
Part of the confusion stems from the contemporary definitions of the word ‘intelligence’. The established usage of the word relates to our capacity to “think, reason and understand” and to have “a good mental capacity”. All positive attributes, right?
The more contemporary meaning has referred to “the gathering of information”, traditionally in a military or political context, but more recently within whatever fields relevant data can be accumulated. As we know from recent examples of ‘intelligence’ gathered from the days of Czech schmoozing of the Labour left, this kind of intelligence can be the exact opposite.
Unintelligence. Dodgy data. Fake news.
It’s understandable. The speed of disruption and the accompanying avalanche of data makes it challenging enough to deal with as it is, even at a surface level.
Plus, there are the psychological factors.
I’ve just read Richard Shotton’s excellent book identifying the 25 behavioural biases we all need to understand if we are to be successful in understanding consumer decision-making; I can think of at least eight of them that might help explain our industry’s embrace of unintelligence.
But, there is a sense that the status quo is being challenged. The recent debates around consumer privacy, transparency, data quality, advertising effectiveness and trading practices point to the increasing importance of applying real, human intelligence to our use of ‘intelligence’.
So, what needs to be done?
There are a few obvious restoratives;
Never mind the width, feel the quality
Just because you have several billion data points, or a data lake measured in yottabytes does not equate to quality.
When data was scarce, it was treated with respect. There was a sense – certainly within the Insight community (or Research, as it was then) that all published data needed to be checked and evaluated. That is now rarely the case, even across the data that is used for trading or to make strategic planning decisions.
A case in point has been the recent travails of Facebook; from massively overestimating video viewing time based on a simple calculation error to embarrassingly claiming the platform reached more individuals than actually exist, the platform has been beset with scandals around the veracity of its data. These were allowed to continue for two or more years, without proper scrutiny by the industry at large.
How much more of the data that fuels advertiser spend is similarly not fit for purpose? Without the answers to such questions, our intelligence is fuelling unintelligence.
If data is the new oil, we need to understand the refining process
There is such a purity about numbers, especially when they come out in sexy spreadsheets, graphics and charts. The problem is, we understand very little about how they have been refined and made ‘fit for purpose’.
So we assume they are what they say they are.
There are many areas in the refining process where unintelligence can breed; for example, algorithms and artificial intelligence.
Algorithms have been described as “opinions embedded in code” according to Kathy O’Neil, in her book ‘Weapons of Math Destruction’. At their worst, they can unwittingly spiral into prejudice, irrationality and fakery; in large part because, according to an article in Scientific American, “artificial intelligence picks up bias from its creators – not cold, hard logic”.
Unintelligence takes those refined numbers at face value. Intelligence questions which biases are built into the model.
A good example of intelligence in action is the recent Radiocentre/Ebiquity study into advertising effectiveness across ten media platforms and twelve different measures of effectiveness. The intelligence comes not just from the breadth and depth of the study but also from, the comparisons with industry perceptions. Some of the consistent disparities in ranking are disturbing. The fact that they are likely to continue being applied is unintelligence at its most stark.
Intelligence goes beyond the data
Behavioural economics defines availability bias as the power of recent experience or exposure to information to influence consequent decision-making.
Digital data is constantly available, causing the analytics dashboards to continuously flicker in our peripheral vision. So, of course it’s often top of mind. But, of course, it’s not everything.
The most obvious consequence has been the shift to the short-term. That has often resulted in diminishing marginal gains and loss in longer-term profitability as the power of branding is subsumed by the need to constantly activate short-term demand.
Code is binary; People are not
One thing I have learned in four decades in the marketing business is that – as far as people are concerned – nothing is ever 100%. There are always shades of opinion, meaningful segmentations of consumer behaviour and multiple potential outcomes to any stimulus.
Intelligence – and, indeed, unintelligence – is often delivered in binary form. Up or down, in or out, winners and losers, dead or alive. We have seen it in the constant stream of “Media X is dead!” headlines and micro-targetting cul-de-sacs.
It often takes a more human-based perspective to understand the nuances and identify the deeper influences behind the constantly shifting data-stream.
There are many more ‘cures’ for unintelligence; for example accounting for the roles of context, or emotion, or implicit attitudes behind the clicks, shares and purchases we track. But until we find better ways to integrate all of this learning into the data flow, we can never be sure we are accumulating genuine intelligence; much less real business wisdom.
David Brennan is co-founder of BE Insight