Computer models are not complete junk....
.....but modelling one single hurricane over 48 hours and an entire planet's climactic system over centuries or millenia are not quite the same. Surely you agree with that? Climate models must include a HUGE number of variables, as Larry has already pointed out. "Burning asteroids", for example, wouldn't affect the development of a storm cell and need to be factored into a hurricane simulation, but would affect long-term climate modelling and need to be properly quantified. Hence, a little sarcasm is in order when someone says rather blithely "[computer models] really are used in the real world".