Although we’re constantly exposed to them, many (most?) people don’t understand the details of how to properly interpret weather forecasts. But even beyond the normal questions, there can be much more complex issues hiding beyond those (though most people will not care for those). This comic takes this to the ridiculous extreme of the weather reporters coming from some other profession where you look into those questions. It shows questions asked by three different people with different backgrounds: mathematics, linguistics, and (in the title text) software development. While some of those questions have actual answers (which you'd expect someone working in that job to know, such as the definition of "scattered showers" and how it's determined, what a "chance of rain" means, and so on), each professional finally ends up with questions that are almost disturbing in how they cannot be answered. (So management ends up calling security to remove those announcers).
Questions from the pure math meteorologist
The first meteorologist, Cueball, has a background in pure math. His forecast states that each of the next five hours has a 20% chance of rain. As a mathematician he sees how limited that information is. There is no information about whether or how those probabilities are correlated. This becomes obvious if you ask the question "How likely is it to rain this afternoon" (a question even some non-mathematicians might be interested in). Cueball states that he does not know (as no one only getting the information about 20% rain in each hour can know). And then lists some scenarios that all fit the the description, but have totally different results for "How likely is it to rain this afternoon?"
The first thing a mathematician would ask (and Cueball does here) is asking if those 5 events are independent. Events are independent if the outcome of one of them is unrelated to the outcome out of the others, i.e. knowing whether it rained at 3 pm has no effect on whether it rains at 4 pm, in which case the probability of any rain over the 5 hours is 1 − (1 − 0.2)5 = 67.2%. (Rain is very seldom independent, as usually having rain in one hour increases the chance to rain in another hour, as systems of rainy weather usually persist for many hours). Another common extreme in probability theory is a set of mutually exclusive events. In this example that would be the scenario that the chance of rain is 5 × 20% = 100%, but it will only rain in exactly one hour and not rain at all for the other four. (Also possible but quite unlikely).
In the second panel he continues to discuss what scattered showers means. Like most of the other weather terms in this comic, the term "scattered showers" is one whose technical definition is largely unknown but appears simple enough that most people would assume they understand what it means. "Scattered" refers to when the rain covers roughly 30–50% of the area at a given moment. To somebody who doesn't know this, like the first meteorologist, there's still the very valid question of how likely it is to rain in a specific spot (is it 30–50% of the total probability, or is it more than that because showers move and sweep out a larger area?), and how this is affected by the previous chance of rain. Not to mention, the percentage that defines "scattered showers" implicitly assumes a surface area that is accounted into the percent. Cueball rightly asks clarification on how large the location used to determine "scattered showers" is.
While the all but the last question of the first part of the second panel can be answered by looking up their definitions, the last one is "What if you have two locations you are worried about?" This is an extremely complex question. Because there is no chance at all to answer this question from the answers of the previous questions or even from most other data a forecast might usually produce. To answer this you'd most likely need to do all the weather modeling and super computer runs of the forecast again with a different algorithm that looks at those two locations. (And for any other two locations you'd need to do the same thing again). This is a common effect in mathematics: While for example a classification of one linear function between two vector spaces is a solved problem (which everyone will learn if they study mathematics), the classification of pairs of linear functions is something no one has yet any idea how to even start.
Finally in that panel Cueball begins to explain that he has asked the management about these things, but that they have stopped replying to his e-mails. At this point he spots the security guy coming over, and the screen goes black in to a technical difficulty screen that excuses this behavior to the viewers. It is implied that the security guy came over to force Cueball to leave the set, because he has been fired for confusing the viewers.
Questioning these things on air is likely confusing to the viewers, although they are all valid questions. But this may lose viewers and the news network is afraid of this. The technical difficulty panel further cements this, apologizing for hiring a person with a pure math background. Often seen as one that do not understand how to talk to regular people.
Questions from the linguist meteorologist
When they get back on air gain a new meteorologist, Blondie, steps in. The management enquires (on air) to make sure she is not also a mathematician. She states no, but tells that she has a linguistics degree, which the management thinks is fine, and thus believes they have prevented the problem with Cueball. However, this proves to be in vain, as Blondie goes into a tangent once more but from a linguistics standpoint, rather than a mathematical one, detailing the true meaning of the word "it" as referring to the weather. After one panel of this the management calls for security again.
While, at the most basic level, human speech is broken into subject, object, and verb; for some reason we are capable of producing and comprehending speech without both objects or verbs, but in English there is a certain "resistance" to speech without a subject. Thus if you are in the passenger seat of a car going down the highway and happened to see some deer in the trees nearby, you could simply say "Deer.", rather than "there is a deer over there", deer being the subject of the sentence. However, if you noticed that it had begun to rain, you could not simply say "Raining." on it's own. Feel how that sentence just seems weird? Hence we have developed the tendency to use the filler word "it" despite the fact that when we say "It's raining." the "it" is not a reference to the clouds producing the rain, but the general state of the rainfall around us. (McWhorter, John. Understanding Linguistics: The Science of Language. https://www.thegreatcourses.com/courses/understanding-linguistics-the-science-of-language.html )
The first question is again quite harmless, and both possible answers ("it" being a dummy pronoun or referring to the weather) are valid answers, but the second question is much more disturbing. In "It's hot out, and getting bigger" the first part of the sentence might be a dummy pronoun or it might reference the weather. But the second part breaks it: With a dummy pronoun "getting bigger" would be the impersonal action, which is not what is meant. It is referencing something (the hotness, that is getting bigger). But if the it references this entity in the second part, by grammatical rules it would also have to reference that in the first part. But "The hotness is hot out" makes no sense at all. This is again a common occurrence with informal speech: From a grammatical point of view, it is pure non-sense. But it still has meaning people understand. So if you want a proper descriptive grammar, it needs to cope with those cases. But then most such informal sentences would be special cases. (Case of point: What is the grammatical function of the "out" in that sentence?)
Questions from the software developer meteorologist
In the title text, the news station has made the same error again, this time by hiring a software developer as the third meteorologist. This last person is stating concerns about the feasibility of the time system used to correlate to the weather patterns. Because it appears simple, many people would simply assume they understand what is being said when a meteorologist talks about "12pm" or "1pm". This is a common mistake because noon is neither post meridiem (pm) nor ante meridiem, and should be stated as "noon" or "12 noon" instead of "12 pm.". However, because software developers frequently have to deal with things such as specifying exactly what time-label means what, the new meteorologist begins to wonder what time period is actually meant on a per-hour forecast. On such an hour forecast does 12pm refer to the hour from 12 to 1pm, from 11:30 to 12:30 or is it actually only to the weather precisely at 12:00 that is referred to? The software developer also worries about an off-by-one error, which is a common error in software development occurring when boundary conditions include one element too few or too many: when counting by 24 once every set period (for example), it is common to forget whether the count should stop at 23 or at 24, especially if the number 0 (midnight) is included. In the 24-hour forecast, that means there's 25 hours represented every day, and the software developer worries that these 25 hours might add up and, every progressive day, the forecast is one more hour off. In theory these are valid concerns and notably less inane than his predecessors, but they are all things he should have asked before he went on the air.
Of course it should be pointed out that hiring someone without any meteorological training to read the weather does not make them an actual meteorologist, no more than say hiring a bricklayer as a doctor would actually make them a real doctor.
Answering the meteorologists’ questions
Management would certainly answer the mathematician's questions! The questions themselves have been asked of meteorologists before. The National Weather Service (NWS), a unit of the United States National Oceanic and Atmospheric Administration (NOAA), has published relevant answers for probability of precipitation, as well as timing and the meanings of particular forecast words. The naming is also addressed here.
Regarding probability of precipitation, NOAA forecasts give the probability that it will rain at all at any given point in an area. To rephrase it, it is the probability of rain occurring at all within a forecast area multiplied by the percentage of area affected by the rain. The "forecast area" is a clearly defined area of land and can be seen in the map of any official NWS forecast. Here is an example.
Regarding the timing of the forecast, an hourly forecast gives the probability for each particular hour, stretching from the time listed to right before the next hour listed. So, the forecast for noon describes the time period from noon to 1pm. The forecasts for individual hours can be correlated; for this reason, the NOAA generates forecasts that stretch over longer time periods, giving a useful estimate for that time range. Thus, the chance of rain for "Today" specifically means: what is the chance of it raining at any given location during any time between 6am and 6pm?
Regarding phrases like "scattered showers", this specifically means a 25-54% probability of precipitation from convective cloud sources. Other phrases, and when they are used, are detailed in the chart at the end of this PDF.
So, to conclude:
- "How likely is it to rain this afternoon?" We don't know, you need to show the 12 noon to 4pm forecast, not the hourly.
- "Is each hour independent? Correlated?" Hourly values are given for that hour only. They can be correlated, hence why they can't be used to calculate the answer to "How likely is it to rain this afternoon?"
- "Is rain guaranteed and we're just unsure of the timing?" You cannot tell from the data given. It's possible (though unlikely), that this is the case.
- "It says 'scattered showers.' Is this the chance of rain somewhere in your area?" Yes, it is, and it means the the rain will come from convective cloud sources with a probability of precipitation somewhere between 25 and 54%.
- "How big is your area?" It's detailed in the forecast the mathematician would be reading from.
- "What if you have two locations you're worried about?" Then all chances are off. While the other open questions like "How likely is it to rain this afternoon?" might have an answer management could supply, for this they do not really have any chance at all.
- "Hey, when we say 12pm, does that mean the hour from 12pm to 1pm, or the hour centered on 12pm? Or is it a snapshot at 12:00 exactly?" It means the hour from noon to 12:59pm.
- We have been effectively nerd-sniped.
- [Cueball is presenting a weather forecast while seated with his folded hands resting on a table. A graphic to the left of Cueball shows the weather for five consecutive hours from 12pm to 4pm, each with a rainy cloud icon and the same percentage of 20% written below the icon. The TV channel's logo is shown on the bottom left, with the 4 in a white font inside a black circle.]
- Cueball: Our forecast says there's a 20% chance of rain for each of the next five hours.
- Cueball: How likely is it to rain this afternoon? It's a simple question, but I don't know the answer. Is each hour independent? Correlated? Or is rain guaranteed and we're just unsure of the timing?
- 12pm 1pm 2pm 3pm 4pm
- 20% 20% 20% 20% 20%
- [Cueball still sits at the table, but the weather graphic is gone and he looks to the right.]
- Cueball: It says "scattered showers." Is this the chance of rain somewhere in your area? How big is your area? What if you have two locations you're worried about?
- Cueball: I've asked management, but they've stopped answering my emails, so—Hang on, the security guy is coming over.
- [A black screen is shown with white text and two short white lines between each of the three segments of text. The TV logo is shown below the last text, with the white 4 inside a gray circle with a white border.]
- Technical Difficulties
- We apologize for hiring a meteorologist with a pure math background.
- We'll be back on the air shortly.
- [Blondie now sits at the desk, in the same position as Cueball, but without the graphic. She looks to the right towards a person who speaks to her from outside the panel. This voice is indicated with two square speech bubbles, connected with a double line and with a small arrow pointing to the right off-panel from the top bubble.]
- Blondie: Sorry about that. Hi, I'm your new meteorologist.
- Person off-panel: And you're not a mathematician, right?
- Blondie: No. I do have a linguistics degree.
- Person off-panel: That's fine.
- [Blondie continues in the same position but now looks into the camera at the viewers. The off-panel person only speaks one word, which again is inside a square speech bubble with a small arrow pointing to the right off-panel.]
- Blondie: It might rain this afternoon.
- Blondie: But what is "it" here? Is it a true dummy pronoun, as in the phrase "It's too bad?" Or is the weather an entity?
- Blondie: Also, what if I say, "It's hot out, and getting bigger?"
- Person off-panel: Security!
add a comment! ⋅ add a topic (use sparingly)! ⋅ refresh comments!