The ad hoc fallacy needs a bit more explanation. It is generally encouraged for scientists (and other humans) to come up with explanations explaining specific cases, and to explain away abnormalities in empirical observations. Urbain Leverrier, for example, theorized an eighth planet in our solar system, in an attempt to explain variations in Uranus' expected flight path. It turns out that he was right; Neptune was affecting Uranus' orbit. Neptune was, in a very strong sense, an ad hoc hypothesis attempting to patch up an apparent hole in the theories of gravitation and Kepler’s laws.
But while ad hoc may be A Good Thing, it can be over-applied. I might claim that there are humans who have ESP abilities, and that I have shown it in controlled experiments. Why can't these experiments be duplicated? Because the hostile thoughts of sceptics disrupt this ability; because using a experimental setup the subjects are not used to throws them off-stride; because the planets were in a bad alignment that day... Any or all of these might be true, but there comes a point where it is no longer productive for mainstream science to research, or believe in, ESP. The fallacy ad hoc comes into play when the ad hoc explanations become too plentiful, too unbelievable, too untestable, and too unproductive. (You will note that Leverrier discovered a planet with his hypothesis... Now that's good ad hoc.)
Ad hoc is a rather subjective fallacy, but one people tend to put a lot of weight on. Watch out for those accusing others of this -- they are often right, but think it though for yourself.