Question 5: Good luck or good management? Part Two.

In part one I argued that you cannot dismiss a model because it feels unlikely, and you cannot choose one model over another purely on the grounds that it appears less complicated. Something does not become a valid explanation simply because you like it and want it to be true. The observable behaviour of the universe is not subject to human whims and tastes.

(note: the observable behaviour of the universe is subject to human fallibility, see Observer Effect. Quite often the mere act of observing something has an effect on its behaviour. More on that when I get back to omniscience ... just in case Robin is about to pre-empt me again.)

To decide the value of a model you have to objectively test it. The model has to make useful predictions that can be tested against observation. This is an absolutely critical step in the scientific method: your ideas must be expressed in such a way that other people can test them and demonstrate that they do not work. One of the primary reasons for publishing a scientific model is to allow your peers to explain why it is rubbish and, hopefully, propose an alternative and better model in its place.


"The earth is the centre of the universe."


The Geocentric Model (the idea that the earth is a sphere at the centre of the universe and everything else revolves around it) was essentially the first scientific model for how the universe works. It matches quite well with casual observation: the earth does not feel like it is moving and the sun and the moon and the stars appear to revolve around it. Unfortunately, it does not match with closer examination: if you accurately map the position of the stars and planets in the sky, they do not move in the way that a geocentric model predicts they should.

The ancient Greeks and later astronomers dedicated a great deal of effort to modifying and elaborating on the geocentric idea to create a model that did accurately predict the movements of the heavens. Unfortunately, they were hampered by a single assumption: the earth is the centre of the universe. It was only when Copernicus dared to suggest that this assumption was false and that the earth was actually revolving around the sun that a simple model suddenly started to match observation. This led to a fairly dramatic conflict between the scientific method and doctrine: according to the scientific method this was a useful and accurate model of the universe, but according to common belief (and Christian dogma) this was heresy.

Science has since established that not only is the Earth not the immovable centre of the Universe, neither is our Sun, or even our galaxy. In fact, a conventional idea of an immovable centre of the universe may not even make sense.

"Space and Time are absolute"

Isaac Newton came up with a set of models for how objects interact (laws of motion and gravity) that formed the basis of classical mechanics. It is impossible to overstate just how clever Newton was and how important these theories are: very very very. Unfortunately, as in the geocentric model, Newton's model makes some assumptions that mean it is only useful in particular contexts (wiki link not necessarily for the faint hearted). As I claimed earlier, this does not make the model 'false', it just puts limits on when it can be used. It is still very very very useful.

Towards the end of the 19th Century, scientists began to make measurements that did not quite match with the predictions of Newton's model. In the early 20th Century, Einstein proposed alternative models (Special and subsequently General Relativity) that better matched the new observations. In this instance, Newton's model was not supported by quite such a firm religious doctrine as the geocentric model, but there was still some resistance. Newton's model is more intuitive and it had been around for a long time. People liked it. But (how hard can I bang this particular drum?) the value of a model is not measured by subjective popularity, it is measured by objective testing. At large scales and great speeds, the predictions made by Einstein's model are objectively more accurate than those made by Newton's model. The scientific method insists that you choose the better model for the job.

(note: there were, and still are, dogmatic objections to Relativity, but they are not anywhere near comparable in scale to the objections to Copernicus' suggestion that the earth revolves around the sun)

Final bang of the drum: the scientific method does not care what people like or what existing doctrine or tradition says. A model is only as good as the evidence that supports it. Badum tish.

Lemma 3: science does not know everything

This is less of a lemma and more of a clarification. An exercise in filtering the mud out of the water.

There is a common argument when one attempts to apply science to certain subjects. It runs something like this: "Science does not know everything. Science does not address X. Therefore X is somehow 'outside' science."

One step at a time then ...

Science does not know everything.

This is true enough. But not necessarily in the way that it is typically intended.

It is true in the sense that the universe contains a lot of stuff and it is highly unlikely that the human race will ever have the resources or technology to observe and catalogue and model every single bit of it.

It is true in the sense that existing scientific models explicitly exclude the possibility of knowing 'everything'. On the macro scale, general relativity tells us that there are parts of the universe that are unobservable. We can never 'see' them. They can never have any effect on us. (That's a slightly heavyweight wiki article ... if anybody wants to volunteer to paraphrase it in layman's terms they are more than welcome). At the other extreme of scale, Heisenberg suggests that we can only know a limited amount about the position and momentum of a particle.

And we've already mentioned incompleteness theory.

(Of course, it is more than possible that the current models of relativity and quantum mechanics are wrong and that we can know more than they suggest. As I've already said, science is always wrong. That's why it works.)

But this statement is true in a more abstract and arguably much more important sense. It is true almost by definition. Science does not strictly know anything at all. It does not even claim to. It claims to provide us with useful models that help us advance in the Real World. A claim I defy anybody to refute (unless they're living naked under a bush and communing with The Internet via herbal telepathy).

A better statement would be "Science can model anything."

If there is something that we can observe and attempt to understand then the scientific method can address it. If something can be defined, then we can attempt to model it. If it can be observed, then we can compare our models with the observation. Essentially, if something exists, then it can be approached with the scientific method.

Have we accidentally addressed the third part of the argument here? Apologies for jumping the gun. Back to the second part.

Science does not address X.
This can be interpreted in two ways. The first and simplest is "there is no current scientific model relating to X". This may well be true. This does not mean, however, that a scientific model of X cannot be formed.

The second interpretation is broader and more nebulous and, I think, is what is most typically intended: "X contains some peculiar quality that differentiates it from the usual subjects of science".

This is the meaning used by people who make claims of supernatural abilities such as telepathy. It assumes that 'science' is a finite set of things: theories and test tubes and computers and boffins in laboratories. But, if you'll excuse me hammering the nail in with yet another nail, science is not a finite and complete explanation of everything. Science is a method.

If X contains some quality, however peculiar, then that quality can be expressed and modelled and compared with observation. That quality can be added to the list of things that the scientific method has addressed.

So ... nothing is a priori 'outside' of science. You can, if you wish, apply the scientific method to absolutely anything at all. The response to the original assertion is:

"Science can be applied to anything. If X exists, then science can be applied to it. Therefore nothing that exists is 'outside' science."

Or, expressed in an even more controversial direction:

"If the scientific model is somehow inapplicable to X, then X does not usefully exist."