How to beat confirmation bias in any given domain

Confirmation bias is a disease where the afflicted is looking for information that confirms his or her hypothesis rather than trying to contradict it. At present time, it is estimated that the whole of humanity is effected by it. Why is confirmation bias a problem? For one thing, looking for confirming evidence doesn't make you learn new things about the world. Since the world is pretty big, and humans are generally small, this is a problem.

Part one

Let's look at the opposite of confirmation bias: the scientific method. The scientific method is a very clever set of tools for figuring out things about the world, or at least parts of it. What it boils down to is basically putting up a hypotheses - the stronger the better - and a list of assumptions, and then try your best to prove the hypothesis false. If you have done that for a while, in an imaginative and honest enough fashion, you get some sense of whether the hypothesis is correct or not. If you are lucky, you might even make a theory out of it (something which seems to be likely to be true, given our current understanding of things), and if you are a bit less lucky, well, then you at least know something that you didn't before - that the hypothesis was false!

It took a while for humans to figure out the scientific method and how to learn things about the world (other than what is right in front of them). It wasn't only about finding the right method of course - a statement like that would be a hypothesis in itself - but about many other things, like possibility of trial and error, tools, the right type of culture, etc. Still: the scientific method is pretty darn useful. But where is it used? In science! And if you look carefully (let's not be too cynical here), not even in all parts of science.

It seems like the domain where the Method is applied matters a lot. In some domains, like art, it doesn't make any sense at all. But in others, like psychology or sociology, it is probably the right tool to use. What about other domains, like your mental model of another person? Or validating if a business model would work? Or debugging code?

For any given domain, it takes a while to learn how to use the Method and perform experiments in the right way. In parts of physics I imagine it being quite straight-forward, but in, for example, personal relationship I think most of us find it hard, not to mention unusual. We know the recipe though. By focusing on a single domain, and listing out your assumptions, hypothesis and ways to try to falsify your hypothesis, it seems likely that it can be done.

Part two

However, let's be realistic. We are claiming here that we can choose a domain, and then beat confirmation bias in that domain. This is our hypothesis. We are assuming that people can read instructions and then follow them, not just in the moment, but for an undetermined amount of time! This is a claim to be very suspicious of. Another assumption is that we are even operating in the same frame of mind when we read something and think about it, and when we are right in the middle of something.

One single instance of using confirmation bias in this domain after following this article would be enough to falsify the hypothesis. And you can probably guess what will happen: you will get excited a while because you can finally be the rational person you pride yourself on being, if only in a given domain. In a couple of days time you'll, if you even bother to notice, fall prey to confirmation bias once again.

Go ahead, test it. Take a fresh piece of paper, or emacs buffer if thats your thing, and write "How I beat confirmation bias in <whatever-domain-you-wish> and then try as hard as you can to disprove that you beat it. Pay someone to follow you around, waiting for you to slip up and show signs of looking for confirming evidence. One single instance is enough to convince yourself that your hypothesis was false.

So our title is probably false, and merely acts as a way to get eyeballs. There's no good take-away lesson here. And that's that.