by Brian on January 4, 2004

Here’s Wolf Blitzer’s current poll question

Do you think any of the Democratic candidates for president can beat George W. Bush?

I honestly don’t know what this means, so I figure I’d throw it over to the LazyWeb. It seems to me that if I answer ‘Yes’, I’m implying that I believe that any of the Democratic candidates for president can beat George W. Bush. And that’s false since I know Sharpton and Kucinich can’t. (At least if we ignore distant possible worlds they can’t.) But if I answer ‘No’ I’m implying that I don’t believe that any of the Democratic candidates for president can beat George W. Bush. And that’s false since I know Dean, Clark, Kerry etc can all handily whip Bush.

The problem is that ‘any’ behaves differently in positive and negative environments. Maybe this is just a presupposition failure, as in “Have you stopped voting Republican?” but I don’t remember seeing it discussed before.



Dedman 01.04.04 at 6:14 pm

A very lawyerly parsing of the question. Interesting.


Dick Thompson 01.04.04 at 6:26 pm

I didn’t have any trouble interpreting the question as meaning “Is there some one of the candidates whom you believe…”. In fact your other interpretation still seems ludicrously inappropriate to me. Am I just dumb?


digamma 01.04.04 at 6:32 pm

I think someone who works regularly with predicate logic would take the latter interpretation and vote yes on the basis of a single candidate’s (Dean’s, Clark’s, etc) ability to beat Bush. The proper way to state your former interpretation would be to replace “any” with “each”.

Nonetheless, the poll is poorly worded, since people read fast on the Internet and many will follow whatever interpretation pops into their mind first. Perhaps “Do you think there is a Democratic candidate for President who can beat Bush?” would be better, but even that sounds awkward.


Ophelia Benson 01.04.04 at 6:34 pm

Well the trouble with those bits of ordinary language that we have no trouble in interpreting the way we think they are meant – is that we may be wrong about the way they are meant. That’s one place that so much of the confusion in everyday discourse comes from, surely. So it’s always worth discussing it.

Just as it really is worth discussing what ‘is’ means. It’s not self-evident, and it’s not limited to only one possible meaning – the journalistic treatment of that whole issue was grossly unfair.


Matt Weiner 01.04.04 at 6:45 pm

I had the same question about the “should Ralph Nader run” website, which asked something like,

(1) Would you think that Nader should not run in 2004 if any of the Democratic candidates win the nomination?

I honestly did not know whether “Yes” was supposed to mean

(2) There is some Democratic candidate such that, if s/he won the nomination, I would think Nader should not run


(3) I think that Nader should not run if any of the Democrats wins the nomination.

I think, from the way the follow-up was phrased, they meant (2), which frustrated me because I wanted to answer “yes” to (3).

I also think Dick is almost right as to the intended interpretation here, because of the semantics: Few people think that Sharpton etc. could beat Bush.

Except: “Is there some one…” implies you have one in mind. It might be that I think that one of the candidates can beat Bush, but I don’t have an opinion about which one it is.


Matt Weiner 01.04.04 at 6:49 pm

Ah, I like digamma’s proposed rephrasing. That makes it clear that it’s existential, but you can cogently answer, “Yes, but I don’t know which”; you can’t answer that if you’re asked “Is there a candidate that you think can beat Bush?”


Brian Weatherson 01.04.04 at 7:06 pm

For all the people who think I’m just being pedantic here, note that there’s a really plausible principle that looks like it is being violated here. The principle is that when someone asks “Do you think p”, answering affirmatively is to say “I think p”. But that simply can’t be the right answer here, because “I think any candidate can beat Bush” unambiguously means they all can beat Bush. So that’s an interesting counterexample to the simplest transformative account of question meaning.


Matt Weiner 01.04.04 at 7:54 pm

Doesn’t that principle fall down on “ever”? If I ask, “Do you think that you will ever go to a baseball game again?” I can’t answer “I think I will ever go to a baseball game again.”

But the problem there could be that the answer is just ungrammatical. If the commenters’ consensus is right, we have a case where “I think P” is the wrong way to say “yes” even though “I think P” is grammatical.


Matt Weiner 01.04.04 at 7:56 pm

…and I see you made that exact point at your other blog…. Teach me to publish without a proper literature survey.


John Isbell 01.04.04 at 8:45 pm

I could happily beat George W. Bush. Give me a two by four and line him up. Or, I could work with a slipper if you bent him over the sofa.


PK 01.04.04 at 8:46 pm

The issue of “positive and negative environments” has been discussed pretty extensively… google “negative polarity items”.


Brian Weatherson 01.04.04 at 9:41 pm

Right, so the question could have been “Why on earth are questions NPI licencing?” But that would have probably been a little too geeky even for CT!

I used to know a little about NPIs, but I forgot most of it, so if anyone knows a snappy answer to that question, feel more than free to write it in :)


Aramis Martinez 01.04.04 at 11:19 pm

Geeks, gotta love em. I haven’t seen such an interesting and simultaneously pointless argument since my friend and I debated whether Newton’s definition of a derivative using limits was THE definition of a derivative (even though Newton’s work preceded Leibnitz’s work by years). I do wonder what motivates the issue since it seems clear to me that it means “Do you think that it is possible that any of the people running for president of the USA in 2004 as a Democrat can beat GWB in the November election?”. Is there a dialect of English where ‘any’ in this question is equivalent to ‘all’?


Brian Weatherson 01.05.04 at 12:24 am

Well no, because nobody thinks it is possible that they can _all_ beat him, because they can’t all run against him. But that’s presumably not what you have in mind.


Dell Adams 01.05.04 at 4:44 am

(Not that it matters in any way, but they could all run against him – just not all as the candidates of a single party.)


Fasteddie 01.05.04 at 5:38 pm

Yes. I have stopped voting Republican.


notapipe 01.05.04 at 6:06 pm

I think it’s fairly clear that the best, common-sense interpretion of the question (if we are forced to choose) is the one where “any” is “some” (the Yes answer). But suppose we make the seemingly innocent replacement of “can” with “could” :

“Do you think any of the Democratic candidates for president could beat George W. Bush?”

Now I suddenly read it the other way, where “any” is “all” (and answer “No!” because Vermin Supreme hasn’t a snowball’s chance in hell). And it doesn’t seem that it’s a difference in how near the possible worlds need to be or some kind of hidden subjunctive. I don’t know why though. Does anyone else read “could” the same way?


Keith 01.05.04 at 6:24 pm

It’s a standard Double-Bind question. The wording is rigged so that either answer produces the desired result (or more acurately, a result that, with parsing produces the desired answer). It’s a flaw in the Aristotlian logic of either.or questions. A third or fourth or fifth option would be more acurate but acuracy is not the desired effect, only unscientific data that back up pre-ordained ideologies.


Matt Weiner 01.05.04 at 9:29 pm

I do hear a difference in nearness/closeness of possible worlds, or something like that.
“Howard Dean can beat Bush” = something like “Dean will win if he tries hard enough, doesn’t screw up, catches a reasonable number of breaks.”

“Dennis Kucinich could beat Bush” = something like “Anything can happen, maybe the month before the election the whole Bush Admin. will be indicted on Plame-related felonies.” It makes a difference if you emphasize “could,” I think.


Josh Narins 01.05.04 at 11:12 pm

I over-analyzed on the math section of the SATs, and answered the more complex question I found.

Yah, the only one I got wrong (on that side).

But this does scan as an “excluded middle”


Tom 01.06.04 at 8:04 pm

Logically, it’s a poorly worded question, and you could certainly argue that a proper reading be that he is asking if all candidates can beat Bush. However, under this reading, if you answer no, you’d be answering that “not all candidates could beat Bush”, not that “no candidates can beat Bush”. The negation of “all” is “not all”, not “none”. So if you thought at least one candidate could not beat Bush, your answer would be “no”.

Of course, most of us choose to answer the question as we believe the questioner intended to ask, not the one they actually did.

When I was a child, and my wife asked me if I had cleaned my room, I suppose I could have answered yes, since I had at some point in my life cleaned my room. But of course I knew she meant “Have you cleaned your room since the time that I asked you to?” We tend to parse questions in context.


Mike 01.06.04 at 8:35 pm

Interesting that Dell says:
“Not that it matters but they could all run against him – just not all as the candidates of a single party.”

He means, “All at the same time, but not All of the same party”

Not that it matters but they could all run against him in the same party.

I mean, “All of the same party, not All at the same time”.

But could they all run against him at the same time and of the same party? I think so. But if they were to do so, there would be some interesting backtracking (e.g. it would have to have been the case that the party system was much different than it now is…)


Kate 01.08.04 at 6:11 pm

Tom – interesting childhood!

Comments on this entry are closed.