Good processes and bad outcomes
Even good processes can lead to tragedies — we should not let hindsight bias make us believe otherwise
When you think about it, your personal life is actually quite complex. You’re faced with a huge number of possibilities pretty much every waking minute of the day. How on earth can you, every time, make a choice that is in your best interest?
Every morning I need to decide what shirt I am going to wear. I have not really counted them, but there are dozens in my wardrobe — T-shirts, polos, formal shirts, casual shirts. Of course I don’t really weigh up the pros and cons of wearing each one of them, every morning. I use a process. I decide what category is most appropriate, and within that category, I simply take the shirt that is on top of the pile. (Just in case, I perform a quick mental check to verify that it is appropriate for the day ahead, and in the unlikely event that it isn’t, I swap it with the next one.) Even with this small addition, my approach vastly simplifies what would otherwise be a daunting task.
We also use processes to help us overcome potentially detrimental influences, for example the way we sequence our meals. The dessert is generally the least healthy course, and most of us leave that for last. This widespread convention helps us have the most nutritious input first, and protects us from the temptation to stuff ourselves with pudding or cake too early on.
What works for individuals can also be beneficial for society as a whole. Jurisprudence — laws and the way they are implemented — acts as an intricate process to help ensure that all citizens are treated in the same way, without arbitrariness or preferential treatment. On occasion such a process can go against our individual self-interest. When we are caught out parking where we shouldn’t or driving too fast, wouldn’t it be good if a friendly civil servant of our acquaintance could make our ticket disappear, and save us a few bob? But we accept that, on the whole, as members of society we are better off with a process that is blind to who we are, or whom we know.
Unfortunately, even good processes don’t always lead to good outcomes.
When things go wrong
Imagine I take the top T-shirt from the pile to wear to a garden party one afternoon. Pretty standard issue, no offensive slogan on the front, so it easily passes my ‘is it OK?’ test. A few hours later, helpful as I am, I am carrying a tray full of glasses outside. Then the sleeve of my T-shirt gets caught on a door handle. Unable to keep my balance, I stumble, and I let go of the tray. Mayhem ensues: I fall to the floor amidst the breaking glass, suffering a large cut on my arm, and worst of all, breaking my phone.
The T-shirt I was wearing — the choice my process had produced — had the widest sleeves of all the shirts in my wardrobe. With any other T-shirt, I most likely would not have caught the handle with my sleeve.
It’s tempting to engage in counterfactual thinking after the event. What if I had chosen to wear a polo shirt instead? I knew this shirt had comfortably wide sleeves — why didn’t I consider that in my mental check? What if I had given more thought to deciding what to wear, instead of this stupid process of picking the one on top?
It is also easy to be taken in by hindsight bias. Once an event has happened for certain, it seems straightforward to trace it back to a decisive point in the past, when we could have known the disastrous consequence ahead.
So it is with processes in society. Last Tuesday, a man attacked and killed two police officers and a member of the public in the Belgian city of Liège. (Later it would emerge he had murdered another person the night before.) The killer was a petty criminal, serving a sentence for drug offences and burglary, and on temporary release from jail ahead of his forthcoming release. It later also emerged that he had converted to the Muslim faith in 2012, and that he was named in two national security files.
Unsurprisingly, in the hours and days following the murders, there was plenty of blame flying around. Newspaper commentators wondered whether the killer should have been able to benefit from day release — “it was known that he was not a reformed citizen, that he had converted to Islam, and that he associated with would-be terrorists in goal — could this tragedy have been averted?” “How could this man walk freely?” Someone clearly made a mistake. For the opposition MPs there was no doubt: the responsibility of this tragedy lay with the government and in particular with the Justice minister, who should of course immediately resign.
But it is not the ill-considered attribution of blame that is the biggest problem with hindsight bias. The risk is that hindsight bias becomes instrumental in the decisions made in the aftermath of a tragedy like this.
Of course, it is crucial to investigate the exact execution of the process that led to the release of murderer. Every single fact that was known about the perpetrator, at the time of every single step in the process, should be reviewed and reconsidered. Did those who made certain decisions have access to all the facts? Were the facts properly evaluated? These are all important, valid and necessary questions.
Tragedies: triggers for our thinking (but no more than that)
There is only one fact that should not figure in this enquiry: the fact that the person who was on temporary release committed four murders. Because he was not the murderer at the time of the release — he was the eventual murderer. We all know that now with certainty, but nobody knew this at the time the decision was being made. Nobody could therefore have taken this fact into account at the time.
When we look back in time, we see only certainty. We know that four people were murdered. There is no doubt about that. But hindsight bias makes us believe that we could also have known with certainty that earlier signs would inevitably lead to the tragedy. We see the path to what happened in reverse. We overlook the fact that, when we look forward to the uncertainty of the future and not to the certainty of the past, at every juncture the decision was a matter of probabilities.
It is not because the murders actually happened, that they were necessarily and inescapably the result of the process that allowed the day release of the perpetrator. Perhaps the criteria were not applied correctly, or perhaps the criteria themselves should be revisited and changed. The enquiry should provide clarity and guidance for possible changes.
Tragedies should trigger our thinking about how we can better try to avoid them. But they also seem to accentuate our desire for simple solutions, and hindsight bias appears to tell us that such simple solutions are easy to get. But hindsight bias fools us by presenting what is a probability as a certainty.
This illusion of certainty is a very poor guide to decision making. We had better be on our guard for hindsight bias.
Originally published at koenfucius.wordpress.com on June 1, 2018.
Thanks for reading this piece. If you enjoyed it, please consider telling me (and all Medium visitors) by clicking or tapping the applause button 👏, once, twice or many times. And of course, do share it in your social network: use the Twitter and Facebook buttons the good folks of Medium provide to make your life easy, or — only slightly more effortful — just copy and paste the link wherever you see fit. Thank you!