The Watchmaker: my fav argument

[Mister Agenda]

It would be more accurate to say that in science we go for the simplest (without any elements not required) explanation that accounts for all the observations. Saying God did it is functionally equivalent to saying 'magic did it' and doesn't actually explain anything.

I don't know how true that is of science. Yes, ideally you want a particular model to be as economic as possible. But you never choose between models based on simplicity. (At least that hasn't been the case in the cognitive sciences; our theories seem to get more complex over time.) Theories are tools for scientists, so perhaps in that sense they might opt for the lightest hammer. But simplicity isn't a replacement for experimentation. And I'd be interested to see an example where simplicity actually did what you say it does.

Edit: And if I may add: Given that you have no access to reality, except by your theories, you have no contrast by which to measure simplicity. In other words, you are unjustifiably deciding that a given level of simplicity is correct. But the more complex theory could be the simplest, and the one you've chosen an oversimplification. Simplicity is an unjustified preference, that reflects the limits of our brains, rather than the nature of reality.

They get more complex only to account for all the necessary elements. Sometimes new elements are discovered and a theory has to become more complex to account for them. I know of no pair of theories in which the simplest that accounts for all of the required elements is not considered the most robust. I never remotely suggested that simplicity is a replacement for experimentation, the idea doesn't make sense. The Copernican vs. the Ptolemaic view of the Solar System is an obvious example of a theory that became dominant primarily because it was more parsimonious; although we have additional evidence of heliocentrism since (though not of the sun being the center of the universe).

We always accept the more complex theory if it better accounts for reality, though we may use an earlier theory if it 'works well enough' while understanding that it's not exactly correct. Lacking elements that the more complex theory accounts for that the simpler theory doesn't, the more complex theory will only be correct by coincidence; as the additional complexity isn't necessitated by any observations.