Democracy and algorithms

I was listening to an episode of the Talking Politics podcast where they revisit an interview to Harari from 2016¹. It triggered a question of what is the meaning of democracy when you have access to so much data on individuals.

Let's assume that with the available data, we can estimate what will each person vote with a reasonable error margin. If the error margin is lower than the gap in a binary process, such as the presidential election in the US then you can actually predict its outcome.

Of course this can trigger a lot of thoughts on different paths, but the more interesting one to me is what about policy shaping based on that information. If you could predict the outcome of elections, legislative bodies such as the parliament or the congress could continuously do popular consultations of the laws they are developing without actually asking anybody.

This is a very interesting approach, because you are not delegating the decision to an algorithm. You don't ask a computer what would generate the best outcome. You ask the computer what would people vote for. Therefore you are not disentangling the decision making process from human serendipity, moral, and its associated mistakes.

It would be a way of automatizing direct democracy, if I may call it that.