Our research group has had experimental papers rejected because they did not agree with first principles simulations. The issue is that at least in quantum chemistry and materials physics, the theory is simple and easy to write down but making quantitative predictions from the theory requires the use of (a) massive computational resources, (b) inspired approximations, or (c) all of the above. (unless it's an "emergent" phenomenon) It's debatable what the relationship between theory, experiment and computation is. I submitted an abstract to an APS conference yesterday and I noticed that you were asked to classify your submission as "theoretical", "experimental", or "computational". I think I agree with the APS in that all three styles are useful and complementary to each other.