The tragic death of Harambe and what a 200kg gorilla can teach us about logic and risk.
What can a 200kg Gorilla can teach us about risk?
In Cincinnati this week, a 200kg gorilla was put to death after it had grabbed a four-year old child who had fallen over the wall and into the enclosure. At this stage, nobody is actually sure how the child came to fall the 4.5m down into moat – what we do know is that by the time the decision was made to shoot Harambe, the child had been in his presence for 10 minutes entirely unharmed.
Obviously, people are outraged. An endangered species was killed because of a combination of parental neglect and a failure on the Zoo’s part to adequately account for how situations like this, despite their rarity, could be resolved without lethal intervention. By all accounts, Harambe had done little more than drag the child out of the moat and away from the screaming spectators. Indeed, University of New England animal behavior expert Gisela Kaplan is of the opinion that the gorilla was actually investigating the child’s encroachment into its territory, and would have realised it was of no threat to the troop. Others who have watched the footage go as far as suggesting that it appears as though the gorilla was holding the child’s hand shortly before it was shot.
Was Harambe acting aggressively towards the child? No.
Was he likely to see the child as a threat? No.
Should he have been shot? Absolutely.
Evaluating the decision process which lead to Harambe being shot
It’s about understanding the nature of risk and rational decision making.
One of the things I have found really poignant about this incident is the way in which both the decision to kill Harambe and the subsequent outcry illustrates the difficulty people have with understanding risk and distinguishing emotion from rationality in decision-making.
I understand why people are upset – shooting an animal so close to being human they exude soulful complexity and empathy seems tantamount to murder. The fact that such an animal is kept locked in a cage in a zoo for our amusement invites comparisons to the barbarous travelling freak shows of the 19th Century. Points of this nature are strong, but they have no relevance to whether or not Harambe should have been shot. Ultimately, the child is human, the gorilla was not. The debate about keeping gorillas locked up for public amusement is one we need to have, but it does not belong in the decision-making process that led to the shooting of this beautiful animal.
As an exercise, place yourself in the situation of the zookeepers and handlers at the zoo.
A child has fallen into the enclosure. The clock starts now.
As a zookeeper, you have an intimate understanding of both the species generally and the animals individually. What you know is that there is a child in the enclosure and the troop leader has taken a hold of it. You are making decisions in the moment – there is no benefit of retrospect to rely on – you can only respond based on what you know and what you see in front of you.
There are two obvious considerations:
- The child cannot be allowed to be harmed.
- The animal holding the child is capable of killing it in an instant.
What do you do? The first thing you notice is that the crowd is loud – it is screaming because a child is being held by a Gorilla. They aren’t going to be silenced – watching helplessly as a child is put in imminent danger unleashes our strongest emotions and most primitive of inclinations. You know the noise is likely to make the Gorillas agitated – they are deeply sensitive creatures, and you know that this particular gorilla could tear the child limb from limb upon a sudden impulse; such is their immense power.
Obviously, you try to engage the gorilla – but they are not human, they don’t necessarily respond or understand verbal commands, and you are conscious of the fact that any action could quickly antagonize the gorilla into being violent.
Remember, in hands strong enough to crush a skull, it holds a human child. One moment is all that it takes for this to end horrifically.
You consider tranquilizing the gorilla. But again, the sedative effects of the tranquilizer are not immediate – that five minute window would slow its cognitive function, making it more prone to acting on impulse. In essence, tranquilizing Harambe means you shoot a dart into a 200kg gorilla and hope that it doesn’t startle for the millisecond it would take for it to kill the child. You are now talking about a gamble.
Remember, it is holding a human child; when are the odds of occurrence low enough to permit you to take the gamble? Oh, and the clock is still ticking – we aren’t in an air-conditioned room considering the issue with the benefit of hindsight, we are standing on the edge of an enclosure looking at an adult gorilla holding a child whilst people scream and cry in anguish.
You are in the pressure cooker.
Do you keep waiting for an amicably satisfactory resolution? Do you hope at some point the Harambe loses interest and leaves the child alone? One moment, one false move, one mistake, and that child is dead. Sure, it might leave the child alone – or it could tear it apart and mutilate its carcass to demonstrate control of its territory – or, not being conscious of its own strength, it kills the child by accident.
At some point, the difficult decision has to be contemplated – do you shoot the gorilla? The longer you wait, the more likely the gorilla is going to take that decision out of you hands.
Killing the gorilla saves the child. A properly aimed bullet from a big enough rifle kills the gorilla instantly, and painlessly. It presents the lowest risk to the life of the child.
Sadly, that is what had to happen to Harambe. He died, the child lived. Discussions about gorilla’s rights and the cruelty of keeping fellow primates as a public spectacle belong in another time, another place. Not the Gorilla Enclosure of the Cincinnati Zoo when a child is at risk.
Lessons about risk and logic from Harambe
The other aspect which has struck me about the outcry over Harambe’s death has been the way in which people have criticized the Zoo’s actions, and relied upon logical fallacies to do so.
What are logical fallacies? Simply, faulty or defective reasoning which is used to develop an argument; it is an error in thinking as opposed to an error upon the facts. The balance of this article will address some of the logical fallacies people have fallen into.
The importance of relevance and separating rationality from emotion.
Zoos are evil, Harambe shouldn’t have been kept in a zoo.
I’m not here to debate the pros and cons of modern zoos; I do, however, empathise with the argument that some animals should not be kept in zoos. I understand that if we didn’t keep gorilla’s in zoos, the child would never have been in a position where it’s life was in danger.
The issue, however, is that Harambe was kept in a zoo, and the child was in danger. Criticising the zoo for making the decision it did on the basis of something of no immediate relevance to the issue in play is a logical fallacy because it conflates two non-related issues. They are separate discussions, both of which deserve attention. If an argument is not relevant, it is not rational. An argument not grounded in rationality is one made on the basis of emotional sentiment. By conflating the zoos and the shooting, you weaken the rational merits of both arguments.
Why? Because the problem with emotion is that it has a habit of muddying the clear waters of rational debate; its role is largely an obstinate one because people whose opinions are grounded in emotion are unlikely to recognise the merits of an opponent’s arguments. It is much more difficult to convince someone that your position is correct when the discussion is an emotional one, and you cannot win an argument without demonstrating that a position is either right or wrong.
In essence, keep emotional considerations as separate as possible from the rational arguments you can make in support of your position because it clears the mind of the audience you are trying to convince. Keeping emotion separate, you can ‘walk’ a person through why zoos are wrong, and have a better chance of them agreeing with you.
Avoiding the “gambler’s fallacy” when accounting for risk.
Harambe had the child for ten minutes and didn’t hurt him – therefore the child was not going to be harmed.
This an enormous mistake in risk analysis, because it fails to account for the fact that observations of past conduct are not a sufficiently reliable indicator of future actions. The ‘Gambler’s Fallacy’ is essentially the assumption people make that a history of outcomes can affect future outcomes. Transposed onto the Harambe situation, it is the notion that because harm was yet to come to the child, harm would therefore not occur.
The issue is that the ‘sample size’ used to deduct this conclusion is an observation lasting ten minutes with one gorilla. Using data to predict future behavior can be done, but it requires huge data-sets and enormously complex tools of statistical analysis to account for margins of error. Quite simply, this is not the case with Harambe – any argument that he was not going to harm the child because he hadn’t harmed the child is underpinned by a dangerous logical fallacy.
The only logical conclusion we draw from our observation is that, at the time of his death, Harambe had not harmed the child. No conclusion can be drawn about whether or not this indicates no risk of future harm.
Avoiding the ‘Black Swan’ Fallacy
The ‘Black Swan’ Fallacy is a concept of risk assessment advocated by one of the most outstanding and provocative free-thinkers I have ever come across, Nassim Nicholas Taleb. Amongst other things, Taleb is a risk assessment expert and financial trader who predicted the collapse of the American debt markets which sparked the Global Financial Crisis. He spends a lot of time writing, in layman’s terms, about how modern approaches to risk do not adequately account for outliers and unknown quantities. He uses the ‘Black Swan’ analogy:
“No amount of observations of white swans can allow the inference that all swans are white; but the observation of a single black swan is sufficient to refute that conclusion.”
The analogy has a strong historical resonance – Europeans had never seen a black swan and therefore there was no such thing as a black swan; all swans are white is the logical deduction you take from this observation. Such an observation, however, cannot account for facts which exist beyond the realms of the observers’ knowledge. Transposed to risk taking, a person in 1696 could bet a huge amount of money on very low odds that all swans were white. They would be extremely confident that such an outcome was true – after all, nobody in the West had ever seen one. Yet the rational deduction of all swans being white stood for nothing when Willem de Vlamingh observed black swans in Western Australia in 1697.
This has traditionally been viewed as a problem with induction, explored by the likes of John Stuart Mill. Yet Taleb takes the idea further to develop ‘robustness’ in risk systems to cope with or even take advantage of unexpected events, arguing that normal (symmetric) distribution models are defective because they fail to account for the consequences of outlying risks. Transposed onto banking and finance pre-GFC, he labelled the traditional approach to risk as essentially ‘picking up pennies in front of a steam roller’. Simply, we make small gains at the risk of enormous consequences because we assume that unidentified risks are non-existent on the basis that such risks have not been identified and quantified.
What has this got to do with Harambe?
Well, to paraphrase:
No amount of time observing Harambe not kill the child can allow the inference that Harambe will not kill the child; but a single observation of Harambe killing the child is sufficient to refute that conclusion.
Quite simply, the risk is asymmetric.
If you view his behaviour and assume that he is not going to harm the child on this basis, you are ignoring the enormous risks involved if something were to go wrong. On one hand, the positive is that both survive, on the other is almost certain death for the child and the gorilla, who would almost certainly be shot whilst killing the child. If we accept that primary concern of the zoo in the incident is the safe retrieval of the child, the event is an asymmetric risk profile because the consequences of him harming the child, means that reward (the safety of both) is massively outweighed.
This is further exaggerated by the fact that, once violence began, in all likelihood any rescue could not occur before the child was either seriously harmed or killed. The strength of the gorilla makes the child’s death a split second occurrence. Unlike ‘normal risk’ there is no incremental differentiation between serious harm and safety. Consequently, shooting the gorilla becomes the only logical conclusion because it is the only outcome which guarantees the child will survive.