Some years ago I was in Australia’s Northern Territory. The intrepid explorer that I was, I was croc-spotting from the comfortable heights of a bridge over the East Alligator River. The river derives its name because it is east of something. And because it’s croc-infested.
I was reading a story about a German tourist (it’s usually a German) who was attacked by a saltwater crocodile in the vicinity (1). The story concluded to reassure that one is more likely to be killed by a vending machine than a saltwater crocodile.
I imagined what the apotheosis of a left brain thinker, the data-driven Renaissance man, might have done with that statistic. Might he have peeked in to the East Alligator River looking for a vending machine and seeing none, jumped right in?
This empirical fact is useful if you suffer from croc-phobia and live in the Upper East Side of Manhattan, and the biggest voyage you ever plan to undertake is to the Hamptons. But it’s not terribly useful, and marginally harmful, if you’re deciding whether to kayak rivers in Northern Australia.
The vending machine has reared its deadly head again. It seems that more Americans have been killed by vending machines than have died from Ebola. Well let’s head to Liberia for the winter, because there are fewer vending machines there.
Sorry, I jest. But this is not a joke. Some actually think this is a relevant statistic to put Ebola in perspective. And some are actually reassured by it!
Look, I get it. There’s an irresistible smugness in putting such an empirical fact to a millenarian, usually extreme right of center, Nervous Nellie’; and admittedly the hysteria has reached comical proportions (2). That’s fine so long as one immediately says to oneself ‘just kidding, knock on the wood but I hope this Ebola thingie is controlled.’
This piece makes the numerator/ denominator point rather wittily (3). It’s like Where’s Waldo and asks you to identify the Ebola positive patients from 310 million dots. But this, like arguments such as ‘more Americans die each year from high-fructose corn syrup than Ebola, so should we ban flights to USA?’ would be classified under ‘naïve empiricism’ by Nassim Taleb, the author of The Black Swan and Antifragile.
(Actually Taleb now calls it ‘the empiricism of idiots’ (4). I think it may be Frequentitis (5), an excessive belief in Frequentist statistics with a characteristic meme: ‘where’s the data?’)
The difference between Vending Machines and Ebola
As Taleb explains, the problem with Ebola is that it is multiplicative: i.e. it multiplies.
Vending machines, unless they are secret Transformers, don’t.
This means that the growth of the disease is non-linear, geometric or exponential. Here is the point to appreciate. If one make an error, the error also multiplies, and one looks like an exponential fool not a linear fool. The error is not just a deviation from a straight line; y=mx +c +error. The counterfactual is another world.
Why should we care about the error? Aren’t our experts top notch? Surely error is neither inevitable nor certain.
Because of the nature of the event, the pay off. Getting it wrong about Ebola is catastrophic for populations and economies. Getting it wrong about breast cancer, for example, is bad, but will not lead to a global catastrophe.
The most important question in decision making under uncertainty is not ‘what’s the right answer (point estimates, central tendencies)?’ But ‘what’s the worse error?’ It’s like Pascal’s wager. Pascal decided to believe in God because the cost of being wrong if God existed was greater than the cost of being wrong if God didn’t exist (6).
In the case of Ebola the error is severely asymmetric. There is a cost of falsely over-reacting (embarrassment, political incorrectness in profiling Africans disproportionately, Americans appearing parochial again). There is a cost of falsely under-reacting. I need not enumerate this error further.
In fact, Taleb says that when the pay-off of an event is as catastrophic as Ebola’s, probabilities can no longer guide decision-making, particularly if the probabilities were computed in the same manner as the chances of getting two sixes and a five in a dice throw (Ludic Fallacy). This makes us more ‘fragile’ (susceptible to low probability, highly consequential events) because we believe we’re in control when, in reality, we might not be (7).
For a non-linear process with significant uncertainty, risk (numerator/ denominator) is meaningless. It’s not wrong. It’s not even wrong. It’s useless. What’s important is the uncertainty associated with the parameters, and with Ebola there’s plenty of uncertainty (8).
So should we all get non-linearly hysterical now?
No, of course not. I can’t believe you’ve asked me that.
Hysteria, as far as I’m aware, will not produce antibodies against Ebola.
Hysteria will make us more fragile – i.e. make the situation worse. It will force politicians to offer cosmetic solutions such as Ebola Czars. It will politicize the issue by creating anti-hysteria hysteria. This will be a rather useless dialectic which, instead of synthesizing a solution, will increase the viewership of Fox News and MSNBC.
Is there a middle ground between ‘Hollywoodification’ of Ebola and ‘Disneyfication’ of Ebola?
There is, I believe. It’s called honesty. Here is a bold suggestion. Treat people as if they are mature individuals. Tell them ‘we don’t know your risk of contracting Ebola.’
Yep, ‘we don’t know but what we do know is that your fretting incessantly about Ebola will not help matters.’
Give them something to fret about. Yellowstone is due an eruption. And there’s that piece of rock Cumbre Vieja, precariously perched over the Atlantic Ocean. If that drops there will be mega-tsunamis affecting the entire eastern seaboard.
I find that when I’m inundated by ways the world can end it is paradoxically reassuring. It has served me well as an inoculum: I’m now permanently in the chill zone (9).
Should we ban flights to Western Africa?
This is the crux of the issue isn’t it? This is the great divide between those who want an insulated country with sealed borders and those who think an American is a citizen of the globe.
Well I have news for you. To paraphrase Trotsky, you might not be interested in the rest of the world but the rest of the world is interested in you. No American will be safe from Ebola until the epidemics are doused in the countries in which they are currently festering. Liberia’s Ebola problem is not just a Liberian problem but an American problem. Sorry, that comes with globalization. Remember globalization? It got us cheap goods at Walmart and customer service from Rajeev (aka Robert) from Bangalore. It has made us more fragile.
But surely banning flights will reduce the risk, even if it does not eliminate it and surely non-elimination of risk doesn’t negate its reduction, you may counter. It’s reasonable logic. But remember we are dealing with a non-linear process. We could easily be back to square one.
A more pragmatic question is this. Were flights banned how do we ensure that medical personnel and help gets to the afflicted areas with the same, or even greater, fluency as presently?
However the decision with regards to banning flights is made, it can’t be made by empiricism, despite pundits saying there is no evidence flight bans work. Remember the n? It’s one.
Making the CDC more Antifragile
The CDC is a national treasure. Take this as a compliment from a small-government libertarian. And those who see in Ebola a political opportunity to bring to earth another government agency, be careful. You could be scoring the biggest own goal. Or cutting your nose to spite your face.
Having said that the CDC is prone to fragility.
To reduce the fragility here are some suggestions.
- Insulate the CDC from democracy. By that I mean let them get on with their jobs unencumbered by what the people, media or politicians think. Democracy wants answers and solutions which, when made in haste because of media-induced hysteria, are almost certainly going to be simplistic (Bruce Willis or Cinderella). And wrong.
- Stop wasting Dr. Frieden’s time by being a monkey around his neck. He has the most important job on the planet presently. He’s not our agony aunt. He shouldn’t have to waste time informing politicians so that they can make futile prescriptions. Leave that to the Czar of Ebola. There, I just found a role for the Czar!
- The Director of CDC must learn to say ‘We’re not sure if Ebola will never become airborne, but we’re doing our best to face this contingency.’ That is ‘we don’t know.’ Folks, please learn to internalize ‘we don’t know.’ ‘We don’t know’ doesn’t mean incompetence. It means sincerity.
- Let me put this as delicately as I can. Perhaps the CDC should focus on microbiology not macrobiology – i.e. stick to fighting non-linear infectious diseases.
- Fund the CDC. Over fund the CDC. Bloat them during times of peace. So that during times of threat they can scale their operations to gargantuan proportions.
The folks at the CDC are the true James Bond of medicine. If the Ebola crisis will be solved, it will be solved by them.
The Ebola crisis is unique. It cannot be solved by empiricism or evidence-based medicine. Policy wonks on this occasion, just this occasion, embrace Taleb’s wisdom not Gauss (10).
The fight against Ebola needs judgment, uncommon bipartisanship and insulation of our most important foot soldiers from the sensitivities of the masses.
Instead of being hysterical about Ebola, respect it. And respect even more the people who are working to save us from it.
- The tourist was actually bathing in a billabong in Kakadu National Park.
- Link courtesy @skepticalscalpel (I would never have found it)
- The Washington Post link is really clever. But wrong.
- To be exposed to these quotable gems follow @nntaleb
- Please don’t take that to mean that I think evidence-based medicine is useless. Just not terribly useful in policy making with non-linear catastrophic events.
- I’m not suggesting you take Pascal’s wager. It’s just a framework for thinking under uncertainty.
- Read Black Swan and Antifragile. By Black Swan I don’t mean the movie starring Natalie Portman, which is worth watching though.
- For the mathematically inclined, Taleb explains the effect of uncertainty.
- The particular inoculum for me was Guide to the End of the World, by Bill McGuire. For maximum impact read it with Cormac McCarthy’s The Road. After initial bout of depression you will emerge delightfully cheerful. For a more mathematical treatment of Doomsday see Gott.
- I’m not suggesting a mutiny against Gauss. Just a temporary, strategic defection.