I've noticed an insidious and increasingly widespread meme in the climate-change denial community. Having summarily failed to establish actual misconduct on the part of climate science research, or undermine the validity of their models and methods, climate change denialists have taken to denying the very idea of using a model to test a hypothesis (for an example, look at comment #29 here, or this).
I can hardly think of a more anti-scientific attitude, and the people making these statements have either never set foot in science or didn't know what they were doing when they got there.
Models are the bones of science. Forget any theory-of-everything mumbo-jumbo you might've heard from physicists, heady from decades of happy marriage to the gorgeous standard model of particle physics, bombshell thought it may be. We cannot simply take our best set of approximations and extrapolate up to whatever we want to test in the large-scale world. It doesn't work, for various reasons (not least of which is complexity, in lower-case and capital-C forms). Equally, we cannot test most things from a purely empirical standpoint, a benchtop experiment or scientific intervention with a simple yes-or-no outcome. Some behaviour doesn't appear at these scales, or presents itself quite paradoxically.
So we're left with empirical or theoretical models. Approximations. Good approximations. No more clearly can this be seen than in chemistry, halfway between the empirical fleshiness of biology and the theoretical purity of physics. Chemistry is a menagerie of models. Organic chemists operate at a level of approximation that would make any quantum physicist faint in horror; inorganic chemists engage molecular orbitals which are little more than convenient fictions of correct symmetry. These are grounded approximations, proven in experiment and traced to their roots in theory (electronegativity is a wonderful example), but approximations none the less. Powerful intellectual ideas, cutting to the heart of chemical behaviour, and allowing us to create and communicate ideas.
Models are deeply important in other fields of science. The recent discovery of a probably-diamond stellar remnant in orbit around another star was established by the proper modelling of various possible remants, to see what matched. Biochemistry uses carefully constructed patterns of reaction to represent living creatures. And climate science uses awesomely powerful computer modelling to figure out what's driving our atmosphere.
Deal with it.
One of my favourite chemists, R. T. Sanderson, has written on the use of real, physical models in teaching chemistry. I really like ball and stick models. Therefore you can imagine my delight at this video of augmented reality ligand binding:
