Is it going to rain today? It’s great to know the answer to this question when deciding how to dress for the day.

When I was still living in Timisoara, Romania, I used to use the Yesterday’s Weather heuristic: today’s weather is going to be similar to yesterday’s weather. And it worked about 70% of the time. Not bad for a heuristic.

When I came to Berne, Switzerland I continued to use this heuristic for a while, but my prediction rate dropped dramatically. Why? Because the weather in Berne just does not want to comply with this heuristic.

One possibility to improve the situation would have been to complain to some higher authority to regulate better the Berne weather so that I can go on with my nice little system. The other solution was to understand that heuristics are just approximations of our understanding of a particular situation or system.

In the case of my heuristic, it took me a while to understand that I should not predict "weather", but "weather in Timisoara", and "weather in Berne". In other words, the location has a significant influence on the weather, and this influence should be explicitly taken into account. For that matter, there are many other factors that need to be taken into account, but location provides a reasonably good example for the discussion here.

It is not because a heuristic worked in one situation, that it is directly applicable to others. This idea stayed at the basis of my very first scientific paper which was about how we should first check our heuristic against the historical data of a system, and to apply it to the current situation only if it proved effective in the past.

As a consequence, I am not using that heuristic when in Berne (except perhaps in summer when the rain is not that cold).

Is it going to rain today in Berne? Maybe, according to goingtorain.com. goingtorain.com is slightly different than other weather prediction sites. Instead of saying "there are 60% chances of precipitation" they say "maybe". And in big letters.

I do not know what heuristics they use, but I do like their presentation. It reflects honesty. They can certainly not go wrong with a "maybe", but they are not afraid of saying that they just don’t know and that the responsibility is yours. However, when they do know something they say "yes" or "no". No matter the answer, they put it in big letters. That is, they stand behind their heuristic and they recognize that it is not a perfect one.

Good heuristics are hard to find, and when we find them they improve decision making. The only important thing to remember is that no matter how good, heuristics are not laws. They are still just approximations.