Computers undermine Occam's Razor

There is a principle in science called Occam’s Razor that states that when two models successfully predict the data, the simplest one is usually right.

I'm going to argue here that computers are not conducive to simplicity. They are, as Douglas Adams said, incredibly stupid and have to be told how to do things in great detail, but they are capable of being stupid millions of times faster than humans. Their ability to simulate incredibly complex systems like the climate system or spiral galaxies is potentially a huge benefit, but the disadvantage is that computers make it possible to get the right answer with incredibly complex and possibly wrong assumptions. Computers then are the opposite to Occam's razor: Occam's hair transplant.

For example, galaxies are observed to spin far too fast to be held in by their visible matter, according to standard theories of dynamics. This is a puzzle, but computers have enabled astrophysicists to calculate exactly what distribution of invisible (dark) matter would be needed to make general relativity and the observations agree. They then produce a beautiful fit and claim a success for general relativity and dark matter. One might as well attribute galaxy rotation to invisible swimming angels, or the spatula of God, since these are just as predictable and well observed as dark matter (ie: not!).

In my view, computers have enabled people to manipulate the "observations" in a complex way to support an esteemed theory, and that is the opposite of science.