« previous post | next post »

A reader asks the NYT Science Times (in C. Claiborne Ray's "Q+a" column on July 13):

Q. How does the weather service determine whether it was a tornado that caused the kind of destruction that recently occurred in Bridgeport, Conn.?

Before I go on, step back and ask yourself what sort of answer the reader was expecting.

If you're like me, you took the question to be about whether it was a tornado or some other kind of weather event that happened in Bridgeport. That, in turn, asks how meteorologists label storms as tornadoes or something else. So we'd expect an answer that distinguished tornadoes from, say, (severe) windstorms, thunderstorms causing wind damage, nor'easters, gales, hurricanes, or cyclones; a good answer would sketch the evidence meteorologists use to make these classifications.

Ray's answer starts out encouragingly enough (though there's some crucial stuff missing):

When a tornado is reported, the National Weather Service may send out a team of investigators to determine how closely the patterns of damage match the aftermaths of past tornadoes.

and goes on:

The levels of damage on the generally accepted Fujita scale are consistent with winds of different speeds. The scale runs from F-0 to F-5.

The Fujita scale, it immediately becomes clear (in case you didn't already know this) provides a list of six levels of tornadoes, each with characteristic damage and wind speed, all succinctly summarized in Ray's column.

Ray has skipped over the question asked and is on to a different one, which presupposes that we know we're dealing with a tornado.

We're used to such unresponsive answers in other question-asking contexts, notably in press conferences by and interviews of public figures, who regularly dodge an unwelcome question by answering some possibly related question they're more comfortable with (or just making a prepared statement somewhere within the subject area of the original question), even persisting through several iterations and reformulations by the questioner, so that the questioning comes to look like badgering nastiness and the questioner gives up (or is cut off).

That's not, I think, what Ray did. Somehow, he assumed that the prior question, of tornadohood, had been settled and went on to the natural follow-up, about degrees of tornado strength. No mention of a rotating column of air, with the (usually, but not necessarily, visible) funnel cloud in contact with both the ground and a cloud above (so that a rotating pattern can be expected in the damage at ground level), nor of the usual association with thunderstorms.

Ray's answer is unresponsive, but honestly so.



  1. John Cowan said,

    July 19, 2010 @ 8:00 am

    I read the question much more broadly: "How do we know it was a tornado, and not a hurricane, an atom bomb, or Godzilla?" The answer to that question is in the first sentence: if someone says a tornado was observed (and perhaps even if not), we check the pattern of damage from the incident against the known pattern of damage from tornadoes, and if it matches, we can reasonably exclude the other possible causes of damage. I take that to be a responsive answer: it would work even if no tornadoes had ever been observed directly.

    The rest is just Geek Answer Syndrome, which differs from its better-known variant Male Answer Syndrome in that the geek believes he (usually he) knows the right answer to a question you didn't ask, but which he would ask in your place, and wishes you to know it too. Per contra, MAS involves someone merely wishing to bullshit you into believing he has the right answer.

  2. Patricia said,

    July 19, 2010 @ 8:11 am

    As far as weather report is concerned, in a documentary about the Bikini Islands' nuclear testing in the 50's, archived interviews of the scientists involved at the time see them complaining that "The wind did not conform to the weather forecast", as an explanation for something that went wrong.
    So, it seems we already went a long way if today's complaints are just about whether we are checking if it was indeed a tornado that caused the damage and not some rebelliousness of the weather to abide by the forecast…

  3. Mark P said,

    July 19, 2010 @ 9:49 am

    I agree that the answer is not completely responsive. The real answer should have been something along the lines of 1) we look at radar returns to see if there was a pattern of strong rotating winds (tornadoes have a unique radar signature), or 2) we look for damage in the area that is consistent with the strong, rotating winds of a tornado and not simply strong, straight-line winds.

  4. Plegmund said,

    July 19, 2010 @ 10:11 am

    They knew it could not be a typhoon because it was in Bridgeport.

  5. Ellen K. said,

    July 19, 2010 @ 10:31 am

    Seems to me that first sentence could be meant as a partial answer to the question, or be ignoring the question. Depends if those investigators comparing the damage to past tornados are looking to see if it looks like tornado damage, or just rating how severe the damage is. But definitely not a full answer since, at best, it says that they determine whether it's a tornado by looking at the damage, which leaves a lot to be said.

  6. Mr Fnortner said,

    July 19, 2010 @ 11:21 am

    If P then Q does not imply If Q then P. ("Look at all the umbrellas; it must be raining.") Comparing the damage to tornado damage would only prove or disprove that the damage was like tornado damage. Thus, it seems that tornado is the default explanation, needing only confirmation that the damage is of the type produced by tornadoes. What other natural phenomena produce such damage?

  7. Dan T. said,

    July 19, 2010 @ 11:28 am

    The phenomenon of somebody rattling off a canned answer to a question vaguely related to the one you asked, but not actually answering your question, is quite common in e-mail responses from tech-support departments.

  8. fred lapides said,

    July 19, 2010 @ 11:52 am

    I knew it was a hurricane because my wife works in Bridgeport, and she was told it was a hurricane. My wife would not lie to me.

  9. Ellen said,

    July 19, 2010 @ 12:39 pm

    Dan T., not just tech-support departments. Customer Service in general. Drives me nuts. Apparently, I don't ask the questions they expect. Generally because I don't need to.

  10. Boris said,

    July 19, 2010 @ 1:04 pm

    Granted the Fujita scale does not answer the question (and don't they use the Enhanced Fujita scale now (EF0-EF5)?), but I think the first sentence gives the best possible answer that a non-meteorologist would understand. Perhaps the mention of rotational damage and other possible causes of damage (e.g. downbursts) could have been discussed, but how much is too much info in a non-technical discussion?

    On the other hand, things like radar signatures are irrelevant here because, I think, the question is about after-the-fact determination, especially since the Connecticut event was being discussed.

  11. Mark F. said,

    July 19, 2010 @ 1:54 pm

    I am generally a fan of the NYT science coverage, but their Q&A column has always seemed to have the weaknesses that Arnold observed. Sure, the first sentence or so is a kind of answer to the question, but it's pretty sketchy, and the space that could have been used to flesh it out is instead spent on something that's gotten to be a pretty familiar concept. I always want it to be a slightly stodgier version of The Straight Dope, and it never quite makes it.

  12. Stephen Nicholson said,

    July 19, 2010 @ 2:36 pm

    As a student, I see something like this this when I ask my professors a questions. I'll ask a question, and the professor will often rephrase my question as something else. The resulting answer may or may not be helpful.

    For example, a few months ago I was asking my professor some specific questions about a paper assignment that was due. He interrupted and said "the question from you is 'how can I get a better grade on my paper in the week we have left.'" The resulting answer wasn't as helpful to me.

  13. Mr Punch said,

    July 19, 2010 @ 3:53 pm

    This question came up last month in the Boston area, where the Weather Service ruled that trees had been felled by a macroburst, not a tornado. This too was considered an unsatisfactory response, because no one had ever heard of a macroburst.

    The wind-related event that forced cancellation of the Wizard of Oz Festival in New York state was universally deemed a cyclone.

  14. Doug said,

    July 19, 2010 @ 7:06 pm

    John Cowan wrote:

    "The answer to that question is in the first sentence: if someone says a tornado was observed (and perhaps even if not), we check the pattern of damage from the incident against the known pattern of damage from tornadoes, and if it matches, we can reasonably exclude the other possible causes of damage."

    But the question essentially was what IS "the known pattern of damage from tornadoes"?

    The phrasing of the question, which referred to "the kind of destruction that recently occurred in Bridgeport, Conn." strongly suggests that the questioner has already guessed that we look at the damage. (S)he apparently wants to know WHAT we look for.

    So it's like this:

    Q: I saw a bear. How can I tell if it's a grizzly bear (as opposed to another kind)?

    A: First, determine if it looks like a grizzly. Second, if it's alive, observe whether it acts like a grizzly. If it growls, determine if it sounds like a grizzly. If it's dead, it would be safe to touch it and observe whether it feels like a grizzly.

    I'd say that's not responsive to the question.

  15. lavya said,

    July 20, 2010 @ 12:01 am

    This definitely begs the question.

  16. J said,

    July 20, 2010 @ 2:10 am

    I heard a lot of those answers during my science degree, usually because I hadn't done the required reading and was asking a question so obvious that the professor couldn't believe it was serious. They would interpret it as a kind of conversation starter, like "tell me more about…". When I finally got them to answer the question I wanted, they could usually give me a great answer, along with a slightly funny look. Clearly Claiborne Ray has forgotten he's dealing with laymen and not his students.

    There's another issue here though. Many people get obsessed with labelling things correctly. To them, it really matters whether there was a tornado last night and not just high winds. It is vitally important that it is one or the other. To a meteorologist, that might not be an interesting question. This obsession with labels sparks e.g. massive internet arguments over details that wouldn't be worth publication in a journal. I hear that there are similar issues in linguistics, where the amateurs get bent out of shape over the correct labels for things where the profession answer is "well, we don't really care".

  17. Ben Hemmens said,

    July 20, 2010 @ 3:13 am

    that's nonsense!

    of course the second sentence is about identifying whether it was a tornado or not – what do you think the level F0 is for?!

    if someone reports a tornado sighting and you go out and find the damage is F0, then … was there a tornado? what's the guy supposed to say to the witness: you're making it up? no, he says well ma'am, we only found F0 damage, so there's no evidence of it having been a bad one.

    truth is, probably 3/4 of the calls they get are from nutcases, but if technical-expert-type people in some kind of public service tell members of the public they didn't see what they believe they saw, BOOM! you have an outburst of anti-bureaucrat rage. so in that sense the communication is very similar to that of a politician – the audience is never far away from being annoyed. the truth is: no tornado. the necessary message is: ma'am, we're taking your tornado seriously and we have it under control.

    i suggest to stop patronizing Ray and start appreciating his pragmatic strategy. which is not so dumb.

  18. Ken Brown said,

    July 20, 2010 @ 4:36 am

    Mr Punch said: "…no one had ever heard of a macroburst. The wind-related event that forced cancellation of the Wizard of Oz Festival in New York state was universally deemed a cyclone"

    The great storm of October 1987 was universally deemed a "hurricane" in England. Except for people who know what a hurricane is.

    Not helped by this famously embarrassing pice of television: http://www.youtube.com/watch?v=uqs1YXfdtGE

  19. Keith M Ellis said,

    July 20, 2010 @ 7:13 am

    Much of this discussion assumes that often unresponsive answers are intentional. J's contribution especially interested me because I've been thinking lately about question-asking in the pedagogical context.

    It seems to me that there's a bit of magic which goes on when one person asks another a question. The magic involves empathy—the parsing of the literal question, placing it within its context in the larger discursive environment, and (most interestingly) some guesswork about the questioner's true motivation and desire.

    My guess is that researchers in AI and other computerized question answering systems find that a great many questions which are easily parsed and answered by people are much more difficult to resolve computationally. The analogous AI problem I have in mind is computer vision.

    Anyway, in the pedagogical context, I think this matters because teachers with limited empathy likely make bad answerers because they have a great deal of difficulty doing two related things: A) assuming the cognitive context of someone with much less knowledge and experience than themselves; and B) utilizing available clues such as other recent questions, knowledge of that particular student's intellectual strengths and weaknesses, things which were recently mentioned which may have confused the student or piqued their interest, etc.

    Some people are very good answerers of questions, intuitively and immediately grasping the motivation of the question and the specific information desired…and often other information the questioner doesn't yet know he/she needs. Other people are very bad answerers. It's interesting that people who are very bad answerers generally place the fault with the questioner.

  20. Ellen K. said,

    July 20, 2010 @ 9:55 am

    @Ben Hemmens: Are you saying that, if there was damage from, say, a microburst, or lightning, or such, it would simple get a F0? Wouldn't then, instead of rating it a 0 on the tornado damage scale, say it was damage from something else, specifying if they know?

    In the initial question, it's a given that there was damage. The person who asked the question wants to know, how to they determine that it was from a tornado?

  21. Jerry Friedman said,

    July 20, 2010 @ 3:17 pm

    @Doug: I agree completely. A responsive answer would have been, "They look for hook clouds in radar images, evidence of rotating winds, red wine, mushrooms, trimmings from the thick part of the tenderloin, and the like."

    @Keith Ellis and J: Some teachers are both good and bad answerers. In grad school I had a professor for quantum field theory who would answer questions with the buzzwords he was so fond of. Then without signaling the transition, he would segue into a clear and helpful explanation. Once I got used to it, it worked fine (at least as far as the question was concerned—I never really understood field theory).

    We teachers need to not only answer questions well, but also let students know that they can keep asking if the answer doesn't help them.

  22. Jesse Hochstadt said,

    July 20, 2010 @ 3:18 pm

    - The column is presumably meant to be pedagogic, not merely to answer specific questions. Ray is probably not responding only to the specific question, but also to the broader NY Times readership, who might want to know about how degrees of tornado damage are distinguished. So to call that part of the answer unresponsive doesn't seem correct to me, nor does it necessarily suggest Geek Answer Syndrome.

    - 'The phrasing of the question, which referred to "the kind of destruction that recently occurred in Bridgeport, Conn." strongly suggests that the questioner has already guessed that we look at the damage.' I don't agree with "strongly." In ordinary speech "the kind of destruction" is as or more likely to mean the degree, rather than the pattern or kind, of destruction. When a war reporter says, "I've never seen this kind of destruction before," s/he is likely to mean "I've never seen so much destruction before," not, say, "I've never seen glass fragmented with that particular pattern."

    - Ray's first sentence implies that only physical damage on the ground is relevant to determining if a tornado has occurred – that, for example, radar images, etc. are consistent with too many kinds of weather. I doubt that that's true: if you found tornado-type damage on the ground but all the reported weather was mild, you'd have to look for some other cause. But if Ray's implication is correct, then pragmatically speaking his first sentence answers the question.

    So IMO, the problem with Ray's answer (which, in case anyone didn't understand this, goes on for several more paragraphs after those first two sentences) is that the part that directly addresses the reader's question is too short and underdetailed. If he'd answered that question satisfactorily, the rest of it wouldn't count as unresponsive.

  23. Peter said,

    July 20, 2010 @ 4:43 pm

    @Jerry Friedman: I originally read your comment as "In grade school I had a professor for quantum field theory…" and thought you must have been quite precocious.

  24. John G said,

    July 20, 2010 @ 8:48 pm

    There is a good answer to the original question, though, even without contemporary radar. A couple of weeks ago a tornado was reported north of Toronto, and the environment folks went out to verify (the next day) if it had been a tornado or not. Presumably a list of what they looked for to answer that question was what the questioner in the NYT was looking for. That does not consist of the Fujita scale without more, unless an F1 is not a tornado but an F3, for example, is. But it sounds from this discussion as if all F# are signs of tornados (except maybe that F0 is the absence of a tornado).

    The empathy of the responder is definitely important. I recall the prof I considered the worst I had in nearly 10 years of post-secondary education – in the six weeks I attended his classes, three a week, before giving up on him, he did not answer a single question. He said something following the question but not once did he actually respond to what was asked. What I attributed at the time to stupidity was probably just indifference.

    I suspect that computer help desks generate responses by AI, as suggested above – and that is, in the non-technical sense, often fuzzy logic, or at least an imperfect connection to what the questioner wants to know.

  25. Rodger C said,

    July 20, 2010 @ 9:14 pm

    @Jerry Friedman: As a teacher who seems to have a reputation for roundabout answers, I'd say your physics prof may have been doing what I sometimes do: using the question as a prompt to talk a bit more about aspects of the topic that interested him and then zeroing in on the specific question.

  26. Jerry Friedman said,

    July 20, 2010 @ 11:36 pm

    @Rodger C.: My impression at the time was that my professor's initial utterances were meaningless, but I certainly could have been wrong. Possibly I was listening for an answer instead of what he wanted to talk about.

    I sometimes do the same thing you do, but I think I more often answer the question first and only then go on to something interesting that it reminded me of.

RSS feed for comments on this post · TrackBack URI

Leave a Comment