When Numbers Work

Truth is...?

The question of the nature of truth is extremely confused. People from all walks of life claim to have a monopoly on truth; priests, gurus, artists and scientists. But what does truth mean and how can we approach it in our lives?

I think it's fair to say that religious truth comes in the form of edicts from a higher power. Artistic truth (and I'm no expert here) tends to be more concerned with an exploration of human nature. The problem is that these definitions don't really hold fast under scrutiny. Even within one religion the 'truths' are often contradictory, never mind if we simultaneously try to take all the worlds religions seriously. Truth through art is a subject that is being debated elsewhere on this blog but I think we'd all agree that it's certainly not rigourously defined. How about we attempt our own definition of truth with which we can later return to these troublesome situations?

Let's start with a toy example; we're given ten different maps of London. How can we tell which map is true? It's simple - we test them. If we get from where we start to where we want to go with only one of the maps, then we know that particular map is a truer representation of reality than the rest. Obviously we'd have to repeat this for as many journeys as possible to make sure that the first test journey wasn't a fluke.

And that's all truth is - reality exists in a certain way and you have a model of it in your brain. The accuracy of correspondence between the model in your brain and reality itself is a measure of the truth of your model. How to improve the degree of truth in your model is a complicated question since humans have a number of intrinsic biases that prevent us from seeing reality as it is (this is a fascinating page).

The general way to beat these biases is the to use scientific method. In a loose sense the scientific method simply dictates that you should make as strong a prediction as you can about a particular event and then perform an experiment to test whether your prediction was correct. If you correctly predicted the outcome (within the experiments degree of accuracy) then you can say that your theory has some amount of truth to it.

The map metaphor also clarifies why the scientific method works - the journeys that we take to test our maps are the perfect analogies of scientific experiments we make to test our theories.

Since we can never prove that our model exactly corresponds to reality, we can never say that we've reached the absolute truth. We can however, have a scale of truth; if I can more accurately predict the outcome of an event than you can then we can say that my model is truer. That's why it's a joke to compare science and religion in their quests for truth. Compare the accuracy of scientific predictions to those of any of the worlds religions and you'll soon see where the understanding lies.

It isn't surprising why this is the case either. Take a normal person and ask him to make a map of London by walking the streets and he'll probably do a fairly poor job, but you'd get somewhere near where you want to go when following it. Take even a complete genius and lock him in a room in Rome and ask him to make a map of London and it will be utterly useless. Religion locks thinkers into dark rooms and asks them to make maps; science demands that they explore the streets themselves.

This is why I don't really understand what people are referring to when they say to me "there's other ways to think that aren't purely scientific". I can see that it's true - there are other ways - but the scientific method only really requires that your beliefs correspond as closely as possible to reality. This seems to mean that the other ways of thinking must be defined as 'your beliefs do not have to correspond to reality'. Whilst that's an extremely easy criteria to fulfil I'm not sure that it's a very useful way to think.



These ideas are immeasurably influenced by both David Deutsch and Eliezer Yudkowsky.

[c/t to Andy for the discussions]

Posted by Marc at 02:07