Flying Seraph
Why Data Won’t Get You Very Far

W. Edwards Deming was a remarkable man. Often called the ‘original data scientist’.

So it feels right, in one sense, to see so many people post, ‘like’, Tweet and otherwise share the following (attributed to him):


In this highly ‘sciencified’ era, we feel compelled to measure everything. Rushing to prove we’re all logical, rational folk. We’re smart. We’re grounded. We think. We don’t rush to judgement.

For those promoting data and the scientific approach, Deming’s words provide great support and comfort.

Yet I can’t help believing that publishing and promoting this quote is well-meaning, but only a half-truth (i.e. not good enough for a court).

You see Deming also said this:

“The most important things cannot be measured.”

And this:

“The most important figures that one needs for management are unknown or unknowable….”

If I wanted to be mischievous – which I do enjoy – I’d say, on the evidence, he is saying that data is necessary but insufficient.

I worry far less about justification.

I’m far more interested in anticipation and imagination – the possibilities…

I see data far more as an after-the-fact explanation. Like most research I’m involved with, most data I’m exposed to is a rear view mirror.

Don’t get me wrong. I love a clear explanation of ‘what has been’ or ‘what is’.

It’s just that I care far more about ‘what could be’.

That is the domain of imagination and creativity. It’s the realm of gut feel, intuition and possibility. Data will play a part (Informed gut feel, perhaps.). But compared to the value of conjecture and speculation, its role in making genuine breakthrough – as opposed to tinkering and optimising – is pretty small. Sometimes downright obstructive.

Any doubt? Here is but one example:

Akio Morita ran Sony. He was told that people in Sony Walkman focus groups couldn’t possibly imagine why they’d want to carry their music around with them (data). No need for this type of product. Thankfully, he trusted his gut. Told his people to press on. Gave the world the Walkman and, indirectly the iPod.

In the non-commercial sphere, I often wonder why – when we can analyse music and reveal its mathematical heart – has data science not given us a new Beethoven? When we can subject text to rigorous scientific scrutiny – why don’t we have a machine that writes with the elegance and beauty of Shakespeare? When we can deconstruct Picasso scientifically, where is our Picasso algorithm?

My suggestion is this:

If you really like the Deming saying about data and opinion, that’s fine. As long as you balance it – as he did – with his ‘important things can’t be measured’ saying.

I’d far prefer you follow this Bill Bernbach quote:

“We are so busy listening to statistics we forget we can create them.”

Or this – also from Bernbach:

“Advertising is fundamentally persuasion and persuasion happens to be not a science, but an art.”

Sir Martin Sorrell got it right many years ago. In a paper for the IPA called ‘Beans and Pearls’ he explained that he hired extraordinarily imaginative people to produce pearls. He confined himself to counting the resulting beans.

And that, folks, is the sequence in which to use the 2 Deming quotes: first imagine, then measure. Have an opinion, then test.

Whatever you do, don’t fool yourself into believing data takes precedence over imagination, gut feel or opinion.

Don’t dismiss opinion, imagination, intuition, pet theory merely because the data hasn’t got there yet.

Please follow me on Twitter: @MarkSareff

Follow my series in here

There are no comments

Add yours