What are the chances that the new restaurant we’re opening will succeed? We’re going to have the best lasagne that anyone has ever tasted, plus some fancy cocktails with creative names. We can’t fail, right? Well, once we start thinking probabilistically, we have to acknowledge that failure is indeed one possible outcome. But how can we assign a percentage to it? Do we just guess?
One good place to start is often with the base rate. The base rate is how frequently something occurs in general, before we take our specific circumstances into account. So in this case the base rate might be the statistic that 80% of new restaurants fail within the first five years. That’s a useful reality check! And it gives us a starting point that we can adjust up or down based on what we think are our specific advantages and disadvantages. Our lasagne might be spectacular, but it’s still only one variable out of the thousands that determine success, and the base rate can help us keep things in perspective. Maybe the lasagne shifts our expectation of failure from 80% down to 79%, but we probably need to look at a lot of other variables before we start celebrating our impending success.
It may sounds pretty obvious to consider the base rate in the restaurant example, but we should probably be doing this in so many other contexts, as well. Will there be a revolution in Iran sometime in the next ten years? It’s easy to come up with stories that make convincing predictions in either direction. “Recent protests are a symptom of massive unrest! The population is approaching its breaking point, and the status quo cannot last!” Or, “Government control over the population is now tighter than ever! Revolution is impossible for the foreseeable future!” When fleshed out with details, stories become extremely convincing, and even experts succumb to the persuasive power of their own narratives. In fact, their vast knowledge can exacerbate the problem and make it easier for them to construct plausible narratives that ring true. But humans are extremely susceptible to assigning way too much credence to these narratives, causing a rash of overconfident predictions. Wikipedia has a great example to show how our intuitions can be wildly wrong if we don’t try to ground them from a base rate.
A group of police officers have breathalyzers displaying false drunkenness in 5% of the cases in which the driver is sober. However, the breathalyzers never fail to detect a truly drunk person. One in a thousand drivers is driving drunk. Suppose the police officers then stop a driver at random, and force the driver to take a breathalyzer test. It indicates that the driver is drunk. We assume you don’t know anything else about him or her. How high is the probability he or she really is drunk?
Many would answer as high as 95%, but the correct probability is about 2%.
The outside view is another concept I like to group together with the base rate. Humans are biased towards optimistic predictions about the outcomes of our plans. We consistently underestimate costs and risks. So if you’re picking stocks, it’s easy to figure out a clever new method to determine what next year’s big winners will be. You might think there’s a 95% chance that your strategies are right. But when you read the research about how even professional stock pickers mostly fail to outperform a simple S&P 500 index fund, then you might need to start asking yourself if you’re missing something. Using the outside view means estimating based on a class of similar cases, rather than assuming that you have special advantages and insights that will make you more successful than others. The others who have tried, most likely felt the same way. (But if you really do have special advantages, you should of course account for that in your estimate)