If something is probably the case, then the fact that it turns out not to be the case doesn't alter the probability of something being the case, so even if a isn't G, the conclusion remains true independent of the facts--so long as the connection between almost all warrants the probability. Right? My cat is probably outside. If it turns out that my cat is inside, was I incorrect? I would have been incorrect had I said my cat is outside...if instead it's inside, but if my conclusion is about probability, then I don't see that what is actually the case is relevant.
Picture a man sitting on the sofa in his living room. His cat is on his lap. He is petting his cat. And he explains this theory about the cat: "This is my cat here. Nice kitty. Usually, at this time of day, my cat is outside, not here on my lap as he is today. It stands to reason, then, that my cat is probably outside right now, rather than on my lap. In fact, the force of that logic is so great that I actually believe my cat is outside rather than here. Yes, nice kitty. Daddy loves you, but your purring isn't fooling me into thinking you're really here. Kitchy kitchy."
You can think that's good logic if you want. I'll continue to think it's nuts.
As for your example, when you thought your cat was outside when it really wasn't, I'll assume that that was a good example of when the proposed was reasonable.
- 1. Almost all F are G;
- 2. a is F;
- 4. Therefore, a is probably G.
That is, you didn't have any contradicting evidence.
- 3. We don't know anything else about a.
I'll agree with you that it is sometimes reasonable to conclude that if almost all F are G, and a is an F, that a is probably G. But it seems clear that this isn't always the case.
1. Almost all blacks are Democrats.
2. Colin Powell is a black.
3. Therefore, Colin Powell is probably a Democrat.
That's not good reasoning. If you think it's good reasoning, how do you cope with this:
1. Almost all military types are Republicans.
2. Colin Powell is a military type.
3. Therefore, Colin Powell is probably a Republican.
So now, if the argument is always good, we have proven that Colin Powell is probably both a Republican and a Democrat. But in fact we know that this is improbable. This form of argument could easily be made to produce a multitude of other contradictions and absurdities. Therefore, we know that arguments of that form are not always reasonable.