February 24, 2005

  • Mercy Maud, do I treasure this idea! 

    Luzhkov Wants Weathermen to Pay

    Mayor Yury Luzhkov promised to punish the Moscow weather service for what he said were unreliable forecasts that frustrated the work of various municipal services and disrupted people's plans.

    Speaking at the weekly meeting of the city government on Tuesday, the mayor said the Moscow Weather Bureau would receive no more funding until it signed an agreement with the city that introduced "financial responsibility," allowing the city to impose penalties for inaccurate forecasts.

    Wouldn't it be a kick in the head if weather forecasters were docked for inaccurate forecasts, similar to a contractor being docked for not completing a project on time?

    To be fair, I realize there are extenuating circumstances with meteorology.  Sometimes fronts stall unexpectedly, etc.  Even then, however, it's not unheard of for one particular forecast to have gone against everyone else and hit the nail on the head.

    As an example, the sleet storm on Thanksgiving Day several years ago that took every weatherman by surprise save for precisely one . . . the man on Channel 4, I'm thinking it was.  It made the newspapers, how his was the sole forecast to accurately predict that sleet storm.

    Meaning, it was possible to predict. 

    Since we have multiple weather forecasts available here in the DFW area, I don't see why a penalty/bonus system couldn't be put in place, based upon the accuracy of the forecast, compared to the others.

Comments (18)

  • I think, though, what you have going on here is that there are multiple models (i.e. ways of analyzing the current weather data) for determining a given day's forecast.

    Some models work better in some conditions, some models work better in others. It might be that there's no way to predict which model works best today to forecast what is going to happen tomorrow.

    It's just an inexact science, based on a huge number of ever-shifting variables. People are free to determine how much stock to put in a weather forecast. Beyond that, holding people accountable for something that no one claims is a certainty seems pointless not to mention unjust.

  • The weatherman who accurately foresaw the sleet storm went against the various computer models, though.  He wasn't a slave to them...they were tools he could use or dispense with.

    He used his knowledge of meterology in general, and for this area in particular, to come up with the sleet scenario.  Those computers don't generally factor in the really persnickety bits peculiar to a specific location.

    If all they're gonna do is parrot what the computers say, then dump the weathercasters entirely.  Who needs 'em?

    How about this...slide it in incrementally, with rewards for accuracy put in place first, with penalties coming along a year or two later?

    Point is, right now there is NO consequence for being wrong.  None.  How many jobs are out there where this is true?  Can Ray mess up over and over and over without anything being said?  Don sure can't!

  • But my point is that it takes judgment to choose between, or choose to chuck, the models on a given day. And the guy who got it right that one day, that one storm -- was he more consistently right all the time, or did he just make the right choice *that day?* So I don't think the fact that they sometimes get it right means that they should always get it right, if the judgment of "how one gets it right" actually varies from day to day, which I suspect it does. To turn your argument around, if it was just a simple matter of figuring out the best way to do it that always worked, then there should be a clear divide between forecasters who get it nearly always right and those who don't, based on which ones use tyhe consistently best method. But ISTM that the errors and accuracy are pretty evenly distributed.

    BTW, when I said models, I wasn't referring exclusively to computer models, but as I said, to ways of analyzing at the data. I'm sure the guy who got the sleet storm right was using SOME kind of model -- otherwise he gets no credit, because the only other options are illogic that happened to give the right result, or dumb luck.

  • Oh, and as to the question of whether Ray and Don can mess up day after day and keep their jobs -- well, of course not. But that's the nature of what they do -- they do things where competent performance results in certain things. If you can show me somewhere that meteorological science has developed a system by which forecasters *can be expected to* get it right consistently, then you'll have a case. Eleanor?

    It's like I said originally -- if you don't want to trust the weather forecasters, don't. If you don't trust them, don't watch or read their reports. Life isn't risk-free, so if you get caught in a storm that wasn't predicted, just remind yourself that storms happen and sometimes you get caught in them. It really isn't anyone's job to prevent that from happening to you, in the grand scheme of things.

  • I don't see what the problem would be for an organization - whether a television station or whatever - tracking the accuracy of their forecaster's forecaster, especially compared to others in the same region, and making an annual or semiannual or quarterly bonus contingent upon being ahead of the pack.

    And the sleet storm was years ago...I don't recall whether whatshisname was more generally accurate than everyone else on a consistant basis.

    If he WAS, though, he should be rewarded for a Job Well Done.

  • I'm afraid Jane's right...not only are there too many variables to count, but there is just too much that's still unknown about the ways of weather overall. Yes, there are guys who seem to get it right a good deal of the time, but that's experience for you. Your sleet storms, too, depend on hour-to-hour conditions many times. Whether your area gets sleet or freezing rain, or only rain can vary from the southern end of town to the northern, as happened years ago right here: we had such an awful ice storm that from our area to about twenty miles north of us, power was knocked out for weeks.  Just three miles south of us, though, on the Purdue campus, not a thing was amiss. Nobody knows where those minor pockets of dry air or freezing air are going to hit.  Even "tornado signatures" in Indiana, it was discovered, are different for here than they are for areas farther south such as where you live.  We're just learning those things. 

    Bottom line: the closer one gets to the hour you need it, the more accurate the forecast is likely to be. And if your forecaster has been in the area for years and years, the probability goes up just from experience. You just can't penalise a young guy for inexperience. Young forecasters don't get paid all that much to begin with. Many come out of school and get into the Accu-Weather system which virtually enslaves them for pennies on a three-year contract.  They come out so bleary eyed and disillusioned that lots of times they quit the field altogether.

  • Then they ought to do away with those utterly bogus five day forecasts.  It's literally the only job I can think of wherein accuracy doesn't count for anything particular.  Do a lousy job and the boss doesn't care.

    Reminds me of a line in the movie "Ghostbusters" where Dan Ackroyd was taken aback at the thought of entering the private sector, after years in academia.

    "They expect results," he warned.

    BTW, the "bleary eyed" I can understand, but what causes the disillusionment?

  • Being paid peanuts (like, $18-20K) for working long, long, insane hours six days a week. BTW, alot of academians are also consultants, who know that results are expected.

  • Good heavens!  I had no idea meterologists regularly put in long hours like that, outside of Impending Severe Weather Events.  And for such miserly remuneration.  Poor things.

    I expect the NWS's continued consolidation of offices doesn't help anything, either, does it?  Fewer meterologists needed, plus they're not on the spot, becoming familiar with the odd quirks of a locale. 

    [apologetically]  BTW, I'm aware academia has its expected results, depending, of course, upon the field.  That particular film quote just seemed unusually apt for the topic under discussion, was all.

    It seems strange that as the weather honchos discover how much they don't know, longer range forecasts - guaranteed to be probably wrong based upon the factors you mentioned - have become the norm.  One would think they'd refuse to issue five and seven day forecasts, citing their inability to do so accurately.

    That's interesting, about the tornado signature varying from locale to locale. 

    Probably we should all just get up, stick our heads out the window and let it go at that. 

  • So here I sit at my wife's connection hearing all this talk about forecasts.  I probably should not dive in, but she said it was OK..SO

    Let me make some clarifications. Weather forecasts are based on info we have gained since 1940 when things like the jet stream were unknown.  Also it costs lots of money to know what is happening in the atmosphere- and the federal gov't does not have a deep pocket. So the forecast is made on less than ideal info. 

    Computer weather forecast models: They are based on measurments made at airports, at radar locations, from some satellite info and about another 80 places were measuremnts are made 2x daily through the atmosphere.  They do not take into account plants on the ground- only bare ground.  They do not 'see' any surface smaller than 32 km square.  They do not include 'everything known' about the weather because they already run on the fastest compiuters available with simplified science.  There are several models, each with some 'quirks' that thend to make bad forecasts during certain situations so the forecast ultimately is someones guess as to which model is right or which group of models are closest top being right.  Then of course the person can be wrong...

    Local forecasts: It does not take much for a forecast to be off big time- say expecting rain and getting freezing rain, snow.  I have my students track the accuracy of forecasted minimum and maximum temperatures and the occurence of rain or snow for 2 ten day periods each winter-spring in Lafayette, IN.  Generally the AVERAGE temperature error is only 1 to 3 degrees and the AVERAGE rain/snow forecast is right about 70% of the time.  That is not bad-- but of course the big single day temperature mistakes and the 30% rain/snow forecast boo boos are noted.  This is a next day forecast.  Actually forecats are prettry good to about 3 days, drop off in accuracy to 7 to 10 days (gives you an idea) and do not mean much further out.  Those long range forecasts used to be called 'outlooks' but now are accurte a liottle more than 55% of the time.

    Should weather forecasters be held accountable?:  Is the federal govt held accountable for the cost of living? the comumer price index? the prime interest rate?... You  can vote them out of office like you can change which weather forecast you listen to.  Are economsts held accountabvle for bad forecasts on stocks, bonds, interest rates, housing markets....Not that I am aware of.   So it is not unique.

    I heard a joke that says Adam Smith quipped that the field of economics was created to make weather forecasting look accurate.....

  • You're right, naturally. ;^)

    Though admittedly it is annoying as all get out when we're told the chance of rain is 100% for THAT DAY yet not one drop falls.

    Apparently the DFW area is very difficult to forecast. We've had more storms, both rain and snow, head determinedly for us...sometimes the forecasts are so ominous the school systems go ahead and close down for the next day...only to have them literally split apart and go to the north and south of us. Not a drop or flake falls.

    Harold Taft, the late, longtime weathercaster for the local NBC station, once explained why this happened, but I can't recall it.

  • What Rich (as I assume) was saying about accountability, the government, and economists was what I was far less effectively trying to get at with my "if you don't like the forecast, don't listen to it anymore" comment. It's only skin off anyone's nose if the forecast is wrong if you choose to make it so by acting on it. Ultimately, the accountability is with you for depending on it, or not.

  • Hey, I wasn't talking about ME or YOU or ANY individual holding a forecaster culpable. 

    Remember, the article which sparked this thread was talking about the entity who pays the Moscow weather people in question (the city of Moscow, maybe?  I'm too lazy to open another browser window to reread the article and refresh my memory as to the details) be the ones who hold their forecasters' feet to the fire. 

    Just wanted to make it clear my intention is not to encourage maddened crowds of overheated or, conversely, damp people to converge on the nearest TV station, demanding restitution. 

  • Yes, I know, but my point was about the whole concept of accountability. There's no reason for accountability in an area where the effect of an action is entirely up to personal choice. IOW, since I don't HAVE to listen to the weather man, and it doesn't cost me or anyone else (unless they CHOOSE to act on his information) anything whether he's right or wrong, why should he be "held accountable" by any sector of society? You hold people publicly acountable for actions that necessarily produce positive or negative outcomes for people or institutions, not ones that people or institutions can freely choose to ignore if they wish.

    So, you might hold the government official accountable for his actions based on what the weather man said, because it was up to the government official whether he was a) going to listen to the weather man and b) if he did, what action he was then going to take. Same with an economic forecast -- the person in authority is still accountable for acting on or not acting on, or how he acts on, the forecast. It's not the forecaster's responsibility, ultimately, it's the person's who acts on it.

  • I think the idea is that there are entities, such as government agencies, airports (or do I mean airlines?), etc. who have their own weather forecasters on staff.  I mean, that's why the forecasters are there, on the payroll . . . to provide usable weather information for their employer.

    Now, fair's fair . . . if someone is paying for the forecast, oughtn't that someone have the right to set parameters as to what degree of accuracy is acceptable?

    If the forecaster considers it to be unreasonable, he needn't work there, after all.  True?  >:^>

  • Well....

    should stock brokers be fired if they don't accurately predict the market? I'm not saying "if they don't do a good job of giving advice to their clients," but if their long term predictions themselves, which are beyond their control and pretty much everyone agrees *can't* be done with any degree of accuracy, come out wrong?

    Okay, sure, he who pays the piper calls the tune. Granted. But what if he who pays the piper is stupid and unreasonable in trying to call a tune that no piper in the world can play? It's still his right to do it, and I can still call it a stupid and unfair idea. That's all I'm saying.

  • Actually, yes indeed, investment analysts such as my bro-in-law can - and will - be canned if their clients' portfolios lose money when most everyone else's is making money.

    Why the heck shouldn't they be?

    Just as Hal has been rewarded in the past for being the only one - the ONLY ONE - whose clients' portfolios gained in value, while everyone else lost.  (The muckety mucks in the NYC office hauled him in so they could hear firsthand how he did it.)

  • But that's not what I said! I said they shouldn't be penalized for mis-forecasting the stock market, but that I WASN'T referring to mishandling their clients! As I understand it, good stock brokers don't base their advice on large-scale market predictions anyway, (Tech's going to go up, up, up! Buy Tech! Any broker who used that sort of logic got HOSED at the beginning of the century) but on real-world analyses of particular companies, funds, sectors, etc.

    Again, I'm not talking about failure to do your job right (i.e. failure to gather the right information, failure to use an approved and appropriate model, failure to draw appropriate conclusions), but failure to predict the unpredictable.

    You said it yourself -- stock brokers get in trouble for *falling behind the market.* But they don't get in trouble when they make the SAME dumb prediction that everyone else makes, or if their predictions aggregate out to the same thing, because the market isn't subject to accurate prediction -- just like the weather. In fact, if they make a totally wrong prediction but by coincidence their wrong prediction results in their giving advice that MAKES their clients money at a higher rate than the market, they'll never hear a word about it.

Comments are closed.

Post a Comment