Thanks coldtea, I think you're onto something important. I'll try to expand on it, tho it's hard to do this briefly.
People reason (and discuss) in several modes, which are ordinarily mixed freely. Deductive, a.k.a. logical (in the proper sense) reasoning is one, and the fallacy-or-not analysis is strictly applicable.
Another mode is that of empirical science, which is basically inductive. You can't prove logically that, for example, substance X causes condition C, but you can establish a good probability by isolating the case, ruling out other factors, showing that the relation is reliably reproducible, etc..
A third mode is humanistic or intuitive. It is an error to suppose that this is unreliable or categorically invalid. Say you have thefts at a workplace, and question employees, and most answer straightforwardly, but one is evasive, furtive, avoids eye contact, can't explain simple events - despite the inapplicability of the logical or scientific modes, this is a rational basis for directing suspicion.
Errors of a meta variety enter when one mode is evaluated by the standards for another. E.g., this writer depicts "slippery slope" as a fallacy of logic by inserting "inevitably" in his interpretation of claims - attributing to the argument-maker the claim that A inevitably leads to B. But in fact "slippery slope" belongs to the third mode, and as such it can be valid - censors do tend to expand their categories, and the actual claim may be, not that the progression is inevitable, but instead that it's unacceptable to go further in that direction - and on this kind of interpretataion there may be no fallacy at all.
It's a good list of logic fallacies, but we just have ot make sure we don't try to apply the refutations in the context of other modes of reasoning.