Overestimating and underestimating risk

To better assess the true level of risk, put short-term risk into a longer-term context

Share this:

There’s something strange going on with risk: it’s very hard for people to assess its true level. Seems like people tend to underestimate long-term risks and overestimate short-term risks.

From an evolutionary perspective, this can be easily explained. If you overestimate the short-term risk of a bear approaching you, you’re probably much better off than if you’d underestimate it. Better to run once too far: the false positive in the world of bears hunting you is exhausting at worst, but the false negative is really bad.

On the other hand, a risk that takes decades to emerge doesn’t hold you back from reproducing. Your reproduction is the only relevant point in evolutionary terms; whatever happens after you have produced offspring has already been passed on in your genes. So underestimating long-term risks is, evolutionarily, not likely to be a problem.

Have you ever wondered why people are still smoking? Why it’s so hard for young people to think about their pension? Or why climate change is so hard for some people to accept? Well, here it is, we tend to underestimate long-term risks. It gives us a focus on the news of today; we tend to be very concerned with what’s happening and see this as an enormous risk. In this context, I’ll refer to my favorite news bulletin of all time, the perhaps well-known BBC bulletin of April 18, 1930. The announcer said: “There is no news.” Then, 15 minutes of piano music was played. Isn’t that wonderful? I wasn’t born yet, and to be honest, I’m happy I wasn’t. Clearly, there was no news, but there were definitely long-term risks, as the devastating events that followed less than a decade later have showed. Now, everyone knows there were tremendous long-term risks back then, but they weren’t seen. But there wasn’t any news. We could learn from that.

With this in mind, what can we do? It’s really unnatural to try and start being less afraid of short-term risks. That’s how we’re wired. As a first, it’s probably a good idea to reduce the time on news sites and stop watching the evening news. Another thing we can do is place the short-term risk in a longer-term context. The approaching bear? Well, the longer-term context is that there will be no longer-term context, so running is a good idea (actually, it isn’t; bears are frighteningly quick). In many cases, many breaking news items are mere incidents in the larger scheme of things. They are definitely devastating for those who are concerned, but are not necessarily always the indication of the end of civilization.

On the flip side, one can also take advantage of the knowledge of how this works in people. This is why climate change should perhaps have been called “extreme weather.” Climate change is the real long-term risk, but extreme weather is observed more frequently. And will therefore be easier to accept as true.

Sometimes, long-term risks need to be regulated. Governments should indeed take steps to keep people from smoking, enforce building up a pension, or require companies to have their financial numbers audited.

The same holds true in a more mundane situation. When it comes to building software, the long-term risk is that software will become impossibly difficult to expand and maintain. This is difficult to explain to those who need that one feature tomorrow (or yesterday, preferably). But they will be receptive to metrics that tell them: what you just built will either speed up future development by x%, or slow you down by y%. People understand the weather (e.g. Is it raining?), but have difficulties understanding climate. Experts, who tend to understand the long-term risks better than laymen, should especially understand how this works. Rather than only banging on about the long-term impact, it’s important to also show the short-term impact. It may require some creative thinking, but it’s necessary to get your message across and reduce your own frustration that “They just don’t get it!

Related resources