Do we have a name for this logical fallacy?

[Kalmageddon]

Person A makes a generalized statement.
Person B disagrees and offers a counter-example to the previous statement, only the counter-example is something that only applies to an exceedingly narrow spectrum of possibilities, thus resulting in something like:
"You (Person A) made a statement that applies 99% of the times, but since I cited the 1% where you would be wrong, this makes your entire statement invalid".

Appears more often as a way to dismiss an overwhelming amount of unfavourable examples by empahsizing the very few exceptions, often by citing personal examples like "actually, *I* a not like this/haven't experienced this". The two topics that off the top of my head tend to suffer from this fallacy more often are improper use of the Stormwind Fallacy and efforts to fight generalized statements regarding minorities percieved as vulnerable in today's society, thus often making the fallacy get a free pass because it's being used for a good cause.

A variation of this fallacy is presenting a counter-example that only works in a theoretical environment, while in practice it would almost never happen because of outside circumnstances or exceedingly unlikely conditions.

In short, this is the fallacy that says "if your statement isn't always right, then it must be always wrong, regardless of how often it would be applicable in the real world".

[daniel_ream]

We've always referred to that as the "Exceptions Don't Disprove Patterns" "fallacy".

It's not a classic logical fallacy, although I suppose it's a form of non sequitur.

If A is some form of universal statement, like "All 3.5/Pathfinder players are purple", then it suffices to only find one counter-example to disprove the statement. "I am not purple, and I am a Pathfinder player, therefore you are wrong".

If A is a particular statement, like "Most/Many/Some 3.5/Pathfinder players are purple" then a single counter-example does nothing to prove or disprove A, since the existence of some non-purple 3.5/Pathfinder players is allowed for in the original statement.

I suspect a big part of the problem is when A is in the form "3.5/Pathfinder players are purple". The rule of intentional logic is that if you don't explicitly say "all" or "every", then you mean "some". By convention, though, English tends to assume you meant "all" when the statement is ambiguous.

[Capt Spanner]

I've seen it referred to as the "counterexample fallacy".

[Trekkin]

It's also one example of the straw man fallacy called extreme manning, in that it involves interpreting an ambiguous statement in the most extreme (and therefore easily disprovable) possible way. If I say "X are Y", finding one non-Y X and using that to claim "not all X are Y" is valid, but going on to act as though you've proven "X are not Y" is the fallacy.

[Murk]

I always referred to it as "My Uncle Pete smoked his entire life and he had his 90th birthday last week!"

The problem is that (unless you are very careful with your wording), it isn't really a fallacy. Since uncle Pete did in fact become 90, smoking does not always kill.
The problem with that is that, even though it isn't really a fallacy, it doesn't contribute to a conversation either - because (as you said), the original statement was mainly just a generalisation to get the point across.

[Cozzer]

the fallacy get a free pass because it's being used for a good cause

Which is a meta-fallacy even by itself. :smalltongue: I've seen the "for a good cause" justification used to defend newspapers and other media spreading false data.

[Xyril]

The rule of intentional logic is that if you don't explicitly say "all" or "every", then you mean "some". By convention, though, English tends to assume you meant "all" when the statement is ambiguous.

Do you mean intensional logic, or something more obscure?

I'm reluctant to accuse people of making a logical fallacy if the accusation is predicated on the assumption that they parsed your statement according to the formal rules that are not largely known to the general populace, and misconstrued your statement anyway. It seems much more likely that the person providing the counterexample is parsing the original statement using generally accepted conventions of conversational English.

[daniel_ream]

Do you mean intensional logic

Yes. Bloody autocorrect.


It seems much more likely that the person providing the counterexample is parsing the original statement using generally accepted conventions of conversational English.

Well, yes. That's why it's important to be specific when you care enough about a point to be arguing about it. Claiming that by the rules of intensional logic "3.5/Pathfinder players are purple" only means "Some 3.5/Pathfinder players are purple" when being called purple is likely to annoy or offend many 3.5/Pathfinder players would be a *** move.

[A.A.King]

I don't really think we have a fallacy on our hand.

First of all. If person A's generalised statement contained the word "all" then it doesn't matter if it covered 99% of the people, he was wrong. Until he or she swallows its pride and corrects him/herself Person B will have won. Person A made a claim that was evidently wrong: He said all and a counterexample could be found. The 99% isn't everyone and Person A should have recognised that and not have tried to dismiss the 1% in the hope to make the statement sound stronger or more important.

Now, if this situation arises you simply have to concede the point, given Person B credits for having been pedantic but correct and the right moment and re-phrase your original statement: "Fair enough, I did not mean all, I meant most" and defend your most by listing examples of the 99%. If at this point Person B restates his original counter example and tries to claim that his one and only example is prove that it doesn't happen in most cases, if Person B now starts ignoring your many examples without trying to disprove them then you're just no longer having a discussion. Person B is now no longer engaging with Person A's argument. If Person A is listing many examples to try and prove 'most' and Person B only ever mentions one then you should just accept that Person B is too stubborn, she isn't willing to debate you. Just say: "I have 9 examples, you have 1. Out of 10, 9 covers most people" and if he/she can't accept that and simply repeats his/her one example expecting to 'win' the debate just move on.

However, how often does anyone really only have one argument? And how often do you really have the 99% on your side? Once person B brings more arguments to defend his/her case and once Person A's most stops being an obvious 99% but a probable 70% it will be harder for Person A to prove his claim and while Person B may not be able to disprove Person B's claim with his examples he can easily bring enough for Person A to not met his burden of proof (and Person A has the burden of proof seeing as he/she made the positive claim at the start)

So, to summarize:
1) If Person A's generalized claim contained the word All then Person B's singular counterexample disproves Person A's claim. He or she should learn to be more precise and not to overestimate. Never make an absolute claim you can't proof.
2) If Person A corercted himself/herself or made clear from the start he/she ment Most and not All then Person B is just too stubborn to discuss with if Person B genuinely believes a singular counter example proves that all of person A's examples can't be most
3) The situation probably isn't as absolute as the hypothetical which makes situation 2) too rare to create a named fallacy for it.

[danzibr]

I do this all the time.

When I hear words like "everybody" or phrases like "all the time," I take them literally.

[Murk]

An interesting one, that I sometimes hear, is if something does not go for everyone, but does for most people, and as such, probability is very high.
Eh, I can explain better with an example.
You say: "You shouldn't jump out of airplanes without a parachute. Most people who do die."
She says: "I know someone who jumped out of an airplane without a parachute and didn't die."
She is right, of course, not everyone dies. You, however, are also right that most people do, and as such, she probably shouldn't jump.

[veti]

However, how often does anyone really only have one argument? And how often do you really have the 99% on your side? Once person B brings more arguments to defend his/her case and once Person A's most stops being an obvious 99% but a probable 70% it will be harder for Person A to prove his claim and while Person B may not be able to disprove Person B's claim with his examples he can easily bring enough for Person A to not met his burden of proof (and Person A has the burden of proof seeing as he/she made the positive claim at the start)

I'm glad you said this bit. I've had a lot of arguments with people who say that my exception doesn't invalidate their rule that still applies "99%" or "90%" of the time, when in fact they've got absolutely no idea whether it's 90% or 9%. I call it the "broad brush" fallacy.

It may be that all of Person A's numerous examples have something in common that makes them conform to the rule, but may not be generally true if you looked at a wider population. For example, consider the classic "black swan" argument. An observer in Europe or America looks around and says "Swans are white". Shown a photo of a black swan or a black-necked swan, they might amend their position to "Wow! OK, but the great majority of swans are white". Which is still missing the point. What they could reasonably say is "(most) swans around here are white (but in other places this rule may not apply)".

[Traab]

I always referred to it as "My Uncle Pete smoked his entire life and he had his 90th birthday last week!"

The problem is that (unless you are very careful with your wording), it isn't really a fallacy. Since uncle Pete did in fact become 90, smoking does not always kill.
The problem with that is that, even though it isn't really a fallacy, it doesn't contribute to a conversation either - because (as you said), the original statement was mainly just a generalisation to get the point across.

"smoking is bad for your health"
"My uncle pete smoked his entire life and is 90 now!"

I like that, its a good example of the op's statement. Just because it isnt universally lethal at the age of 50 doesnt mean the original, statement is wrong.

[A.A.King]

"smoking is bad for your health"
"My uncle pete smoked his entire life and is 90 now!"

I like that, its a good example of the op's statement. Just because it isnt universally lethal at the age of 50 doesnt mean the original, statement is wrong.

The thing is though that "Smoking will Kill you" (which is probably closer to the statement that Uncle Pete was trying to disprove than the mere stipulation it is bad cor your healtg) is a needlessly absolute statement. It is exactly because (almost) everyone has an Uncle Pete that phrases like "Smoking will Kill you" have lost their impact. Yes, dear uncle Pete doesn't disprove the intended statement, namely that a smoking habit gives you a high chance of dying of a smoking disease before your time, but he does disprove what is being said and the anti-smoker arguments would have more strength if they didn't try to scaremonger with needless absolute.

It is not a fallacy to point out when someone is using an overgeneralisation and that his/her absolute is wrong.

Note: Pete doesn't disprove the simple statement "smoking is bad for your health" because you can have a bad health but still get old and if it is specifically that kind of statement the OP meant then the problem isn't bringing just one counterexample but the problem is bring the wrong counter example. However; if we are talking about what I think we are ("Smoking Kills" vs "Tell that to my uncle") then the problem remains using an absolute when you shouldn't have. It isn't that hard to add the words "most likely" or "will probably" to a generelization like this and it will have the added bonus of making your statement much more true.

[BannedInSchool]

But then one could also say that it's wrong that getting shot in the head will kill you, because some people have survived that. Therefore getting shot in the head is perfectly safe. :smalltongue:

[veti]

But then one could also say that it's wrong that getting shot in the head will kill you, because some people have survived that. Therefore getting shot in the head is perfectly safe. :smalltongue:

... riiiight...

So the Uncle Pete counterexample has been misapplied. The warning was modified to "Smoking is bad for your health" - but for all we know, Uncle Pete is still running weekly marathons and having wild nightly sex with his 22-year-old girlfriend.

It's not the magnitude of the effect that we should modify, but the likelihood. Rather than "smoking is bad for your health", we can say "smoking will probably kill you".

This also works for the shot-in-the-head situation.

[2D8HP]

I do this all the time.

When I hear words like "everybody" or phrases like "all the time," I take them literally.


I'm reminded of a couple of tricks I've used to pass "multiple choice", and "true/false" tests, and doimg so has gotten me my current job.

If the answer states "always", or :never", the answer is probably false.

If the answer states "mostly", or "usually", it is probably true.

Count the letters in the answers, the one with the most is probably true.



... riiiight...

So the Uncle Pete counterexample has been misapplied. The warning was modified to "Smoking is bad for your health" - but for all we know, Uncle Pete is still running weekly marathons and having wild nightly sex with his 22-year-old girlfriend.

Really?

I'm going to start smoking tomorrow!

[Traab]

Just remember, when phrasing your claims dont make them absolutes like, "Smoking will kill you" After all, only a sith deals in absolutes. Wait... &%$&^%!!!

[Cozzer]

Well, there's also the fact that in most case this fallacy happens because someone is interpreting your words in the most specific way possible, so that they can "prove" they're wrong and "win" the debate.

I think, in the context of a normal discussion, one can use rethoric absolutes such as "smoking kills you" and safely assume that, if someone interprets it as an actual absolute and replies with "ACTUALLY it just mostly kills you", that person isn't interested in talking with you and you can safely ignore them and assume nothing of value was lost.

[Telonius]

This sounds most like Ignoratio elenchi (https://en.wikipedia.org/wiki/Ignoratio_elenchi).

[2D8HP]

This sounds most like Ignoratio elenchi (https://en.wikipedia.org/wiki/Ignoratio_elenchi).


And that sounds like the name for my next PC!

[Murk]

This sounds most like Ignoratio elenchi (https://en.wikipedia.org/wiki/Ignoratio_elenchi).

Ooh, I like this one. It fits, I think: indeed, the person saying "not always" is right, but fails to address the implied point, and doesn't move the conversation forwards in any way, rather turning it into some kind of contest.
Ignoratio elenchi applies to other cases too (it's not very specific), but for now it'll do.

[Jay R]

Technically it is (or at least includes) the either/or fallacy, also called the fallacy of the excluded middle.

By refusing to consider anything other than "A is always true" or "A is always false", one triumphantly shows a single example where A is false to reject the conclusion that A is generally true.

[Algeh]

One of the other forums I used to hang out on used to use the phrase "the plural of anecdote is not data" as the most common counter-argument to the "uncle Pete" type statements on that forum. This is particularly useful with statistics/"usually" statements countered by the occasional outlier story.