Probabilistic Reliability
28 May 2021
As the world has become more and more connected it’s allowed more people to become well known through podcasts, books and articles. This is great. A greater diversity of ideas is what it takes to make progress in the world.
However it can be challenging at times to try and identify whether or not someone knows what they’re talking about.
I’m aware that many people trend towards a black and white approach for how to treat intellectuals and whether what the intellectuals are saying is true for everything or false for everything. I’ve started using a thought process I’m calling probabilistic reliability.
Instead of evaluating an individual as either reliable or not, trustworthy or not, you simply apply a probability to someone.
80% of what person A says turns out to be true.
30% of what person B says turns out to be true.
This allows you to navigate the difficult situations of when you encounter someone who at one point says something informative and useful, and at other times says something false or misleading. You have a system that allows you to accept partial belief sets from someone, rather than just accepting everything or nothing.
Another aspect to consider is the idea of knowledge scope.
I place a very high reliability score on my doctor for my health, but I would set a much lower score for how much I would trust him in regards to rewiring my apartment, or the right software for handling the version management of my website.
The most famous example of this is Economics Nobel Prize winner Paul Krugman who is quoted “By 2005 or so, it will become clear that the Internet’s impact on the economy has been no greater than the fax machine’s.”
You can absolutely trust someone in their area of expertise without being beholden to their beliefs and opinions about everything.
Now of course you might not like this system, but then again, I’m totally happy if you only take on board some of the ideas I present. That is, after all, the point.
If you enjoyed this post, consider subscribing!