• Mark Monfort

Statistical fallacies and Survivorship Bias

This article relates to the following video. Whilst it runs for a good 25 minutes, it's pretty captivating and I have some interesting takeaways from it.


The Possibility of Life and the Fallacy

Finished? So what I found interesting was how a lot of what Professor David Kipping talks about can relate to business and data, not just the chances of life in the universe.


There is a lot of mentioning in science and popular culture about how easy life was to get started here so it must be true that there is life elsewhere. When we consider that there are so many planets rotating around other suns (across different definitions of habitable zones) that we can observe and the commonality across space of chemicals that what we know to be the ingredients/building blocks of life.


There is a famous formula, known as the Drake equation, which is often used to gauge the probability of life existing elsewhere in the universe. Because of the vastness of the universe, low probability numbers actually produce large results. But, all of this assumes that life getting started given a set of ingredients is a possibility and we must insert a conjecture here that this could be false.

DRAKE IMAGE

DRAKE EXAMPLE


The problem with all this is that a very important part of the equation is assumed but has not ever been proven. This assumes the possibility of life getting started is easy but various experiments to get life started from the building block ingredients have not worked.

GOOGLE


Our observation of extremophiles also fuels our speculation but is also a fallacy that is known as survivorship bias. We have no idea how extremely difficult it is to see life get started so just because to us it seemed quick (in universal terms), we do not know if many other planets, given the same set of starting conditions, failed.


Professor David sums it up well when he says “We tend to think our experience is typical … we think that the rules present here are similar to rules that are present elsewhere and that may very well be misguiding us when it comes to rules elsewhere” and for this we need to be mindful.


we think that the rules present here are similar to rules that are present elsewhere and that may very well be misguiding us when it comes to rules elsewhere


Application to business and data

What we can learn from this is to test our assumptions. Just because something is observable to us in our situation does not mean it applies to a whole group.


We might see that, 3 times out of 10, our customers will act in a certain way. That does not mean that this is true of all customers in all similar situations with similar business interactions. There might be something unique to our observable universe


In the world of data, we see the world growing and teaming with new sets of raw assets that can be used to give us new insights into unknown worlds. We need to be careful not to assume things to be true and imply them without also considering that our assumptions could be due to special circumstances and can even be false.


We can overcome this with the collection of more data, asking of more questions, and continuing to be intellectually curious. This is as important for questions on the possibility of life elsewhere as it is for our own smaller problems in our corners of the universe.





33 views0 comments

Recent Posts

See All