That way, I get to see the speakers who are up before me, get a sense of how the audience is reacting to them, and as a consequence get to adjust what I plan to say accordingly.
A couple of years ago, I arrived early at the hotel where an enterprise conference was happening. I slid in at the back of the hall and listened to the Q&A that followed the second speech of the day. It was all very civilised until a question came from the audience about the lack of credit from banks for small businesses, which was an issue even then.
At that point things hotted up a little and, sitting at the back of the hall, I could see audience members sitting up straighter as the topic was explored.
Then a business speaker on the platform let fly. He ate the face off the man sitting beside him — a banker from the bank sponsoring the conference. The banks, he said, were disenfranchising, de-skilling and disempowering their own bank managers to the detriment of the local and national economy, by introducing new practices.
In the past, he said, if a bank manager was approached by a local business for a loan, the manager could give or deny that loan. But they would give or deny based on deep knowledge of the individual in front of them. The manager would know if the person came from a long line of wasters or a long line of entrepreneurs, that their understanding of finance was clear and informed or recklessly ill-informed. Because of all this local knowledge, he opined, the judgement made by the local bank manager in previous decades would always have been a solidly acceptable judgement.
However, the speaker went on, his voice rising along with his passion for his theme, the banks had now pulled that sort of decision away from branch banks and centralised the decision-making process.
To hell with local knowledge and human insight. It was now all going to be done by algorithms and that, the speaker said, was utterly ridiculous, ineffective, dangerous and a serious disincentive to entrepreneurs.
The man was cheered to the rafters.
“Algorithms,” people hissed at each other, as if it was a new swear word.
To me, it was one of those words you don’t feel you have to learn because you’re pretty sure you’re never going to use it. I hadn’t a clue what an algorithm was or is, so I looked it up when I got home.
An algorithm/algorism, the encyclopedia told me, in modern mathematics, is a set of operations reduced to a uniform procedure for solving a specific type of problem, for example, the algorism of continued fractions. Clear? Me neither. I kind of grasped that the speaker had been advancing the case for human intuition, as against the case for mechanical formulas, when it came to predicting outcomes.
But, according to Nobel laureate Daniel Kahneman’s new book, entitled Thinking, Fast and Slow, the speaker was likely to be profoundly wrong. Kahneman introduces his reader to a psychologist named Paul Meehl, who reviewed 20 studies examining whether clinical predictions based on the subjective impressions of trained professionals were better than the subjective impressions of clinicians.
“In a typical study,” Kahneman writes, “trained counsellors predicted the grades of freshmen at the end of the school year. The counsellors interviewed each student for 45 minutes. They also had access to high school grades, several aptitude tests and a four-page personal statement.”
THE algorithm against which the counsellors were to be measured was, by contrast, poorly equipped, with only high school grades and one aptitude test at its disposal.
Yet, despite this drawback, the mathematical formula was more accurate than the vast majority of the counsellors.
Satisfied that the algorithm, not human instinct, was king when it came to predicting academic results, Meehl moved on to examining the same kind of comparison, applied to working out how successful people being trained to be pilots would turn out to be, how many prisoners released from jail would reoffend and how many prisoners granted parole would violate its terms.
You will not be surprised to learn that the results he came up with matched the results for the school-goers. You will not be surprised, either, to learn that the popularity of the psychologist, following the publication of his findings, was somewhere between that of bubonic plague and mad cow disease. Dozens of studies began in order to prove him wrong.
“The number of studies reporting comparisons of clinical and statistical predictions has increased to roughly 200, but the score in the contest between algorithms and humans has not changed. About 60% of the studies have shown significantly better accuracy for the algorithms. The other comparisons scored a draw in accuracy, but a tie is tantamount to a win for the statistical rules, which are normally much less expensive to use than expert judgement. No exception has been convincingly documented.”
In other words, across a spectrum of human interest ranging from the longevity of cancer patients to the assessment of the suitability of adoptive parents, from the diagnosis of cardiac disease to the future price of Bordeaux wine, this annoying swear word of a mathematical tool is better at predicting outcomes than are educated, informed and involved human experts.
At the time, Meehl’s findings were described as controversial. Thirty years later, he mildly observed that “there is no controversy in social science which shows such a large body of qualitatively diverse studies coming out so uniformly in the same direction as this one”.
Why all of this is newly relevant is that we’re currently engaged in an argument about the availability of credit from banks to small businesses, with the banks saying they’re open for business on this front and small businesses saying “like hell you are”.
Kahneman says even when algorithms show the chances of a small business surviving for five years (in the US) are about 35%, “the individuals who open such businesses do not believe that the statistics apply to them”. That’s the problem, isn’t it? Small businesses, like the children around Lake Wobegon, as seen by their parents, are — all of them — above average. They always have great prospects and a glowing future.
Entrepreneurs, according to popular myth, are the ones who see the light at the end of the tunnel. Algorithms suggest what they’re seeing is not the exit/freedom/escape, but the lights of an oncoming train. Kahneman suggests they’re suffering from “entrepreneurial delusions”.
You pay your money and you take your choice. Or maybe, if you’re a bank, you hold on to your money.
* Terry Prone’s new book The Fear Factor is available in bookshops (€14.99)