From Thinking, Fast and Slow:
Experts who acknowledge the full extent of their ignorance may expect to be replaced by more confident competitors who are better able to gain the trust of clients.
I believe Hanlon’s razor applies here: ignorance is a better explanation than dishonesty. I imagine most overconfident predictions are sincere. Unfortunately, sincere ignorance is often rewarded.
14 thoughts on “Overconfidence pays”
I think it is a bit like overclocking your CPU. The benefits are obvious and it is often quite safe. However, go too far and you’ll crash. Crashing could mean being exposed. It could also mean falling prey to depression.
In science, there are huge incentives to be overconfident. You’re only rewarded for publishing, not for being correct. There’s no penalty for being wrong. If you discover your own error, you get another paper. If someone else discovers you were wrong and writes about it, you get a citation!
Politics is even worse. If you can confidently tell people what they want to hear, you’ll go far.
As damning as Kahneman’s perspective is (e.g. with stock brokers, and then there’s John’s additional examples), there are huge legitimate advantages to being overconfident in many fields, especially where criteria for success are ambiguous or truly unknown. Many areas of engineering and business come to mind, where if one had full understanding of the particulars of the specific domain at hand, the rational course of action would be to find another domain. Ignorance leads to overconfidence there, and usually to failure, but often enough to incredible success if the gradually humbled can find a way to struggle out of the mess she’s jumped into blindly.
Chas: I’d say that there are often legitimate advantages to being decisive but not to being overconfident. Sometimes it’s better to carry out mediocre decision quickly and with gusto than to spend time looking for a better decision. Emergency medicine and military combat come to mind. But you can be honest and rational about it.
A Civil Engineer, say, gives the client confidence in the project being funded not by by challenge avoidance, self delusion etc. He or she tells the client these are the foreseeable risks in costs, politics, regulation, safety, ground, weather, skill shortages, inflation, unknown unknowns etc. This is how probable the risks are, and these are the contingency methods I will use should they arise. There is no chance to get the project right second time around. The Civil Engineer has no certainty of which risk will actually occur but the client is quietly given full confidence that should they any arise they will be fully planned for and covered. That is how an honest professional works as opposed to a snake oil salesperson.
Someone not able to foresee such risks and manage them to a successful outcome is soon shown the door.
Robert: Unlike many experts, civil engineers actually know what they’re doing. They deal with measurable quantities and physical laws, and they have centuries of experience to draw from. Unlike, for example, management consultants.
Robert: Of course, there are tons of domains where ignorant overconfidence spells certain doom (mortgage-backed security analysis comes to mind). I think we’re in violent agreement due to the ‘engineer’ nomenclature; software and product design of all sorts (both physical and ephemeral) sometimes benefits greatly from ignorant overconfidence that can provide the cover one needs to carry on with a task that otherwise would be cast aside.
John: An example that comes to mind is Paypal. I don’t know if you know its backstory (Founders at Work has a great interview that covers it), but if the founders knew what they were actually building (ostensibly the most sophisticated fraud-detection system to that date, rather than a payments processor), they probably would have passed. It’s the unknown unknowns that can get you into trouble, because you can’t be honest or rational about them; all you can do is carry on, and believe you’ll be able to handle them.
I’m afraid I’ve hijacked the discussion at this point… :-/
Chas: Maybe that’s an example of a little knowledge being a dangerous thing. If the Paypal founders only knew that they were in for a lot of hard work, maybe they would have backed out. But if they knew they were building the highly successful company that Paypal has become, presumably they’d glad to do it.
“Experts who acknowledge the full extent of their ignorance may expect to be replaced by more confident competitors who are better able to gain the trust of clients.”
Civil Engineers are no different from Management Consultants in that they can draw on the past for some information. But MCs have a little black book of “proven solutions” that are applied ready or not to “problems”, fictitious or not. Clients essentially pay them to write reports to support decisions already made, but to take the blame if things go wrong. No client would ever trust an MC to actually implement any decisions or to run an actual business. However, confident they appear they are not trusted. The one who will not get hired is the one who is so confident he thinks he can actually solve the companies problems better than the CEO can.
A Civil Engineer employed to build the NASA launch station had little to draw on; neither has the Engineer who has to propose flood protection to New York, say, with unknown future rises in sea levels. The political, economic, legal, skills, finance and regulatory situation changes during a long project can only be guessed at.
Who would you employ for something vital?
Robert: You may be even more cynical about management consultants than I am. ;)
The civil engineer hired to build a NASA launch system has Newton’s laws, and finite element software to apply those laws to complex situations. And he has experience from analogous projects. A launch pad isn’t entirely different from more common structures like bridges and towers.
The engineer hired to propose flood protection for New York has meteorological history to draw on. The most terrifying predictions of climate change don’t predict completely unprecedented weather, or even dramatic change over the course of decades. Nobody expects New York to face challenges over the next 50 years that the Netherlands haven’t faced for centuries.
No one is doing anything from ground zero.
Yes you have the FE software but what you have not got is 100% confidence that the data input is right, nor quite what the answers might mean for novel structures. Black boxes make validity judgements far harder than working things out simply from basics. The welter of junk science coming out daily from the misuse of black box statistics packages also shows that. It is not unknown for USA Engineers to make basic errors of imputing data in feet for buildings that are to be built overseas in metres – real spans on site are 3 times bigger than their models.
The point I am making is that real experts can candidly express unknowns in terms of risk management and contingency. Bluffers cannot. Clients can be reassured by the former. A bluffer just says everything will be OK – trust me – ignorance is bliss. The war in Iraq seemingly started on that basis and got worse and worse without any contingencies made when the population drifted into total anarchy.
The engineer for New York has not got the future meteorological data over the next 100 years or so. Polar ice melts are expected to raise sea level by 30 to 90 feet, super depressions raise the sea level during the storms by another few feet. Weather events are predicted to get more severe and frequent – one event therefore is far more likely to be extreme. Netherlands cannot cope with that. It could not cope in 1953 with relatively tiny surges. If NYC is not concerned it ought to be.
@Robert: I’m pretty sure you’re off by orders of magnitude on the 30-90 feet thing: IPCC says more like 30-90 cm a factor of 30 times smaller.
This is the first thing I could google up about what IPCC says on sea lavel rise:
Not that IPCC couldn’t be wrong by a factor of 30… But just that at the moment, Civil Engineers are not really planning for 30-90 feet of sea level rise.
The IPCC forecasts are from computer models which predict rises only 50% of what has actually happened to date. NSF have funded research to look at t over 1000 inland fossil beaches which actually occurred in the Pliocene era when CO2 levels were last at the same levels we now expect. These beaches have a height above current sea level of 38 to 295 feet. Depending on the rate of CO2 etc growth they now expect sea level rises of 12 to 18 feet this century, then accelerating level rises next Century on.
Any multi $B expenditure that falls short of full protection based on these huge ranges in available data would tend to indicate “Overconfidence does not Pay”. But NYC might really want to believe the scamster who says he is certain it will only be something less than 1 foot, with a 20 year personal guarantee written from the Cayman islands.