Yup, the storm going up the east coast of North America moved slightly to the east with the result that many areas received considerably less snow than the media warned might happen. See this, for example. People made plans and purchases, etc. that they wouldn't have made if they'd had a better idea what would happen.
The problem is twofold:Point estimates instead of interval estimates, and loss-minimization tactics.
Point estimates vs interval estimates:
We all know that weather models are imperfect. But the media don't want to take the time to say (as an example for the recent storm) "There's a 15% chance of 12" of snow, a 50% chance of 10", a 30% chance of 6", and a 5% chance of only 2" over the next 24 hours, depending on which way the major air masses drift." And I really wonder if many listeners/viewers would want that much detail. I have lots of friends who would respond, "Yah, yah, so what's gonna happen?"
However, the reports could present graphs of probability density functions showing the probabilities of expected precipitation, expected temperatures, etc. And given that different forecasting models spit out different probability density functions, it might even be useful to more than a few of us stat-type geeks to see a graphic showing the probability density functions from several different models.
Essentially this distinction was one of the errors made by forecasters in their submissions of information to the media and by the media in their presentations to the public. [see this, from WaPo]
When a forecast is so sensitive to small changes (eastern Long Island, not far away, received 30-plus inches), it is imperative to loudly convey the reality that small changes could have profound effects on what actually happens. ...
But the general lack of information provided about the forecast uncertainty is a major disappointment considering both the state of weather forecasting and the efforts some have made to improve how we communicate the forecast.
For many years, the need to express forecast confidence and communicate different scenarios during complex, high-stakes forecast events has been discussed and stressed in the weather community.
And that brings up the second problem,
Loss-minimization Tactics
Imagine if the weather services and the media had indeed presented interval estimates and probability density functions about the east coast snow storm, something like what I suggest above. Imagine further that New York City had received an unanticipated heavy snowfall of, say 18" [following the numbers used in the above example, this would have had probably only 2% probability attached to it.]. Imagine the outrage if the public and public officials hadn't been prepared. And especially if they hadn't been prepared for bad outcomes.
So what happens is that weather services shade their forecasts to allow for "what's the worst that might happen?" If they get it wrong on the extreme side, that imposes far lower costs and losses on the public (and hence on themselves) than if they don't place enough emphasis on the extreme outcomes. From the CBC link at the top of this post,
Kimbell said meteorologists at Environment Canada have the leeway to err on the side of caution, particularly when issuing warnings when public safety is at stake.
"It's better to say there is going to be a bad storm and save lives than to minimize it and be wrong on the other side and actually it's worse and the impacts are severe," he said....
"I would rather, if there is a lean one way or another, lean towards safety because I have seen the consequences the other way and it gets very frightening very quickly," said New York Gov. Andrew Cuomo.
[CBC meteorologist] Scotland said forecasters cannot always err on the side of caution, because if they do people may start to take warnings of dangerous conditions less seriously in the future.
Exactly. Most of us are familiar with "The Little Boy Who Cried Wolf".
At the same time, exacerbating the problem, the media love "Storm porn" [see here and here]. And because of this tendency within the media, a couple of years ago I really tore into The Weather Network (aka the Storm Porn Channel):
Open Challenge to The Weather Network
aka "the Storm-Porn Channel"
You were forecasting 30-40 cms of snow in this storm for the Southwestern Ontario corridor. And some of your announcers couldn't restrain themselves, saying (with drool running down their chins), ".... and it could be more in some areas." How much did those areas actually receive?
I understand that you never, never want to be accused of under-forecasting the seriousness of a storm, but over-forecasting the seriousness of storms consistently means that people develop an immunity to your warnings.
Can you at least start providing us with decent confidence intervals instead of only dire warnings? Please? Do you have any announcers who dare to say, "... but it might be quite a bit less, too..."?
Addendum: Keep in mind that the storm did, indeed, drop tonnes and meters of snow on some places on Long Island and in New England.That it did not leave so much snow in New York City and that that is what became the major news story reflects the NYC-bias of the major media.