Subscribe here: Apple Podcasts | Spotify | YouTube | Overcast | Pocket Casts
How are we still getting caught in the rain? This week’s Galaxy Brain explores the world of weather forecasting—specifically the apps on our phones that we have come to rely on. As climate change intensifies storms and smartphones put hyperlocal forecasts in our pockets, we’ve never had more meteorological data. And yet plenty of people lament that their weather apps can’t get it right. Charlie digs into why we obsessively refresh our weather apps, why we blame them when they’re wrong, and what it really means to forecast an inherently chaotic atmosphere.
Charlie talks with the physicist Adam Grossman, a co-creator of the cult-favorite weather app Dark Sky that redefined minute-by-minute forecasting before being acquired by Apple. Grossman pulls back the curtain on how weather predictions are made—a process that includes government satellites, weather balloons, massive physics simulations, and machine-learning models—and explains why forecasts are improving even if it doesn’t always feel that way.
The following is a transcript of the episode:
Adam Grossman: It’s sort of the realization that all weather forecasts are going to be wrong, right? There’s nothing you can do about it. The key is: How do you convey that uncertainty?
[Music]
Charlie Warzel: I’m Charlie Warzel, and this is Galaxy Brain, a show where today we are going to get to the bottom of a question plaguing mankind since time immemorial: Do weather apps suck?
People have very strange relationships to weather apps. They check them obsessively, they love them, they talk about them, they pay money for them, and at the same time, they constantly complain about them. Weather apps often leave us high and dry or low and wet—whatever you want to call it. Weather apps are a feature of life, and yet the weather is super unpredictable. And so we get [these] tortured relationships with these devices, and they tend to be really, really important. As the climate gets more and more erratic, as there’s more instances of extreme weather, and as we become, increasingly, information junkies, we rely on these apps more and more. And frankly, a lot of times they don’t work the way we want to.
And so I wanted to demystify these weather apps. I wanted to talk to somebody who could tell me how they work, how they’ve gotten better, how they’ve gotten worse, whether we need all the information about the weather that we have. Remember, back in the day, we just used to look in the newspaper and get one forecast, or go to the local news and get a forecast in the morning and a forecast at night. Now we have all this information. What are we doing with it?
And so my guest today is Adam Grossman. Adam is a physicist who created the app Dark Sky back in the early 2010s, and that app quickly became an absolute cult favorite. It launched in 2012, and then Apple bought Dark Sky right around the pandemic and integrated it into their massive weather app. Adam then helped build WeatherKit at Apple for a long time, and he left to build a new app called Acme Weather, based all around this idea of trying to give people more access to more information and also communicate more uncertainty about the weather.
And so I thought Adam would be the perfect person to talk about this. He has this inside view of this platform, and he can help answer these questions: Why do we need all this information? Can we ever get a perfect, definitive forecast? Do weather apps suck, or do the users just simply expect too much from it?
Adam and I got to the bottom of all of this. Here’s our conversation.
[Music]
Warzel: Adam, welcome to Galaxy Brain.
Grossman: Thanks for having me. It’s exciting.
Warzel: So I want to start at the beginning here. How did you get into this job? This is an interesting gig, building weather apps, so have you always been a weather nerd? What is the background here? Walk me through that.
Grossman: So my background actually isn’t in weather. I have a physics degree. I ended up doing a lot of just software development, web development. But I think everyone is kind of a weather nerd, to a certain degree. Everyone gets kind of excited about the weather. I don’t know if it’s just built into humans in general.
I started doing weather probably 15 years ago. I guess it was in the summer of 2010. My now-wife, my girlfriend at the time, we were driving to Cleveland to go on vacation, because people go to Cleveland on vacation, evidently.
Warzel: I’m from Cleveland—I get it. I get it.
Grossman: Oh, are you?
Warzel: Yeah.
Grossman: Well, there you go. I’m in Connecticut, and so we were just driving west. And we pulled off at a rest area, and it was just a torrential downpour. Just cars on the highway were going 10 miles an hour. It was just a mess. And I remember opening up whatever weather app I had in 2010—I don’t remember what it was—and I looked at it to see, Okay, when can we go back out to our car and continue driving? And the weather app said something like, Seventy percent chance of rain. It was like, This is not useful, right? It was a torrential downpour. And then I went to the radar, and you could see it on the radar. And I remember thinking the whole time is like, How can we do this better? Right? If there’s rain right there, your app shouldn’t just say, Seventy percent chance of rain. It shouldn’t just say, It’s raining, right? It should be, Rain is gonna stop in 12 minutes, or whatever.
I started just thinking about the weather then, started playing around with just radar data, trying to see: Can we do machine learning, can we do computer vision to try to figure out where these storms are headed, right? ’Cause if you look at a radar map, you hit the little “Play” button, you see the radar moving at time. You know your brain can parse this as moving through time.
Warzel: Right.
Grossman: A computer should be able to do this, right? So I built a little gizmo for just trying to predict just the next few minutes, right, up to an hour of what the rain’s gonna do, minute by minute, so that if it looks like it’s raining outside or you get stuck on the highway going to Cleveland, you can say, Okay, in 15 minutes, you can go out to your car. It ended up working, and then we decided, Hey, let’s do a Kickstarter to see if we can make an iOS app. And that was an app called Dark Sky.
Originally, Dark Sky wasn’t really a general-purpose weather app. It literally just told you what the rain was gonna do in the next hour. It didn’t even have temperature, nothing. And we always promised ourselves, We’re not gonna make a general-purpose weather app. There’s so many of those. That did not last long, because we realized people don’t want two weather apps.
And then in 2020, just as the pandemic was hitting, we ended up joining Apple to work on Apple Weather. And then four years later, a few of us left, and then shortly thereafter, we started Acme Weather, which is our new app and our new weather service.
Warzel: So I wanna get there, but I think it’s really important that you had kind of the canonical normal weather app experience, right, which is: You open the thing up; you see that it’s not reflecting your reality. Obviously, you had the means and the tools to do that—to change it, to make something different. But I think let’s just start very basic here.
I want to get into the nuts and bolts of how weather apps and forecasting works for the layman, ’cause I think it’s really important to foreground that for the rest of this discussion, which is gonna get into why these apps succeed sometimes, fail other times. But can you just walk me through—explain it like I’m 5—how do these weather apps work? Or how does weather—let’s start even there: How does weather forecasting work?
Grossman: Yeah, so it depends what kind of forecasting you’re talking about. So the one I just mentioned, which Dark Sky did originally, was very short term, very hyperlocal: This is exactly what’s happening at your location over the next few minutes. But that technique is not gonna work for what’s happening this weekend, right, or multiple days ahead. And also, when you talk about just climate forecasting, long-term climate trends, that’s a very different kind of forecast.
When people think about weather forecasting, they sort of think hour by hour, out 10 days, right? And that’s sort of the starting point, and then you could tack on other things. But the way that works, there’s sort of a pipeline. And the beginning of the pipeline is gathering a whole bunch of weather data. And by the way, the beginning of this pipeline is mostly done by government agencies, government weather services.
And so the first step is: If you wanna predict the weather, you gotta know what the weather is doing right now, right? You gotta know what the sort of initial state of the world is. And that comes from satellite data. It comes from weather balloons that they put up. So the National Weather Service puts up hundreds of weather balloons—I think a couple hundred every day. Weather balloons are nice ’cause it gives you sort of a 3-D slice through the atmosphere, so it’s temperature, pressure, humidity, things like that, but at different elevations, and that’s really useful for subsequently simulating the weather. There’s ground stations, weather stations, right? There’s buoys out in the ocean that measure things like water temperature and all that. And so you have all this data that gets collected.
And then that all gets fed into numerical weather prediction models. And again, these are generally models that are run by government agencies. And what they’re literally doing is just calculating the physics. It’s basically running a physics simulation of the atmosphere, given the initial conditions that you have. And they run these things on enormous supercomputers, and you get an output. They’re now starting to do things like using machine learning and AI to do the same thing, but dramatically faster.
Warzel: Let’s go to Dark Sky. You created this to solve, as you put it, a very real and kind of isolated problem, right? To me, as an outsider, I feel like, when you guys started to blow up, the push notification part of it was really important, right? Like, I have this app that is not only gonna tell me this thing, but it’s gonna reach out, make use of, then, what was sort of a relatively new thing—push notifications were somewhat novel at that time—and to say, Hey. Hi. This is gonna happen. Be aware of this. Grab the umbrella, or whatever. What did you guys feel like you solved that led to really blowing up there as an app?
Grossman: I think the big difference is Dark Sky was a weather app and weather service designed for your phone. I started it in 2010. We didn’t launch ’til, really, at the end of 2011, beginning of 2012. But these phones that everyone has with them at all times, these smartphones, they had not been around for that long, right? We have these always-connected internet supercomputers in our pockets, and they were pretty new then. And before that, something like Dark Sky just doesn’t make sense, ’cause you have no way to actually get the information, right? It’s very specific on where you are right now. And that’s not how people got their weather information, in their weather forecast before phones, right? They got ’em from your TV meteorologist, right, or the newspaper. And those necessarily have to be for broad areas, right, for your city, your part of the state. And so the type of forecasts that they would provide were for your region.
It wasn’t that I think we were doing anything super technically magical. It was the fact that we were tailoring it for your smartphone, and I think we sort of got ahead of the other weather services out there, who were still sort of thinking about it the old way, right? The forecast that you would give on the evening news, they just put that on your phone, and it was the same forecast, right? And I think it was the realization that you can do some fundamentally different things once you have an always-connected, always-on device.
Warzel: I wrote a couple of years ago about weather apps in general—and we’ll get into this a little more later, I think—the tortured relationship that a lot of people have with them. There was a stat that I pulled from this website called ForecastAdvisor that was just talking about Dark Sky in general during the time that it was up: It accurately predicted the high temperature in my zip code only 39 percent of the time. Do you feel like there were a lot of limitations to what you guys could do there?
Grossman: It was a lot of fumbling around, right? So, again, we just started with one specific kind of forecast. When we first started looking into doing longer-range forecasts, oh boy, we were very naive. We thought, Oh, you just go out and get the data and then plunk whatever the data says into the app, and you’re done, right? Yeah, it’s not like that, right? And so it takes a lot of work to try to figure out how to take this data and turn it into something that’s as accurate as you can be, right?
Warzel: Were there any big fumbles, where you guys did something and were like, Oh no, we didn’t mean to do that, or This does not work—abort. How does that work?
Grossman: Ah, man. Doing something like this, it’s pretty much all little fumbles, right?
Warzel: Okay.
Grossman: You open the app, forecasts are one thing, but also, what’s happening right now? What’s the temperature outside right now? Models will give you that—not necessarily great at that, again, ’cause of, like, microclimate effects.
I was just looking at station data now, and we’re here, and it was in the 30s, but sometimes there’s stations that’ll say it’s 78 degrees or –100, right? There’s conversion issues. We had so many times where our forecast would just be off by, like, a hundred degrees because of a faulty ground station that we were just trusting.
Most big data problems is 90 percent sanitizing the data, munging the data. And same with weather, right, is I think most of our issues come from just the data is weird in some way that we could have caught, but we didn’t catch it, because we were young and naive.
Warzel: How did you guys get better at that? Is that just simply the process of trial and error? Were there certain light-bulb moments down the road there?
Grossman: I think the biggest thing is—so we didn’t just have a weather app, right? We had a weather app and a weather service to make forecasts, right? And so most indie weather apps, they work on the UI, and they call to a third-party weather service. So Apple has WeatherKit, which is their weather service that developers who make weather apps can tap into. I feel strongly that to make the best weather app, you should have your own weather service—for many reasons, and we can get into ’em, but a big one is because you’re going to get a lot of complaints and emails from users who paid you for this weather app and your forecast was wrong, and by far, user complaints are the No. 1 way that we learn about problems and then go and learn how to fix the problems.
But there’s no light-bulb moment. There’s a million problems, and it’s always something different. And so we wait for really angry customers to email us and say we ruined their wedding because it rained when we said it wasn’t going to. And then we go back and we try to figure out what the commonality between these complaints are and see what we could do to fix it.
Warzel: Do you have a funny or good example of a reader thing where you guys were just like, Oh no, oh geez, something that stands out to you?
Grossman: It’s things like ruining people’s day, and that makes us really sad. In Dark Sky days—well, according to the user—we ruined their wedding ’cause we botched a forecast, and it makes you feel bad. People tend to not email you when you get things right, right? No one’s just like, Good job. Go you. You kind of need thick skin, but it is super useful, right?
So in Acme right now, we have a “Community Reports” section of the app, where people can submit what the weather actually is outside. You make a report. You can see it on the map. You can see everyone else in your area’s reports on the map. That’s useful for a couple things, but one of the things is it gives us real-world data of what people are actually saying so that we can then look at that and say, Does this actually match our forecast? If it doesn’t, why doesn’t it? Right? There’s always gonna be noise, right? There’s always gonna be error, but are there systematic things we can catch?
The other nice thing about it is the weather forecast is always gonna be wrong, and so it’s kind of nice to have that ground truth from other users in your area that are like a sanity check, right? You can turn on notifications for that, so if multiple people say, Hey, it’s raining, you’ll get a notification, if you turn it on, that, Hey, other people in your area are saying it’s raining, right? And so I think that helps us get around just the inherent lack of certainty in a forecast.
Warzel: Let’s talk a little bit about going to Apple. You guys go in there, and the Apple Weather app is … So many people have [smartphones], and so many people default to the one that’s on there. That—
Grossman: I don’t think I’m allowed to say how many users. It is a crap-ton of users.
Warzel: (Laughs.)
Grossman: It’s amazing how many users use Apple. It is scary. Working at Apple on Apple, it is very scary ’cause it’s a ton of users.
Warzel: Was that just an unbelievable amount of pressure to be in there? What were you Dark Sky guys doing in there, specifically? And then, secondarily, was it just like, Oh crap, the stakes are so high right now.
Grossman: Yeah, so Apple always had a weather app, right, and they just used third-party data. Apple decided—this is an important app for people. Apple’s really big on sort of owning the technology that powers their ecosystem. And so they decided, We need to have a weather service; we need to have that capability in-house, so that they can do all the things that they wanna do. They don’t have to be reliant on a third party. And so that’s why they brought us in, was to work on that. And that turned into WeatherKit, which is, again, it’s the behind the scenes [application programming interface] API for developers to deliver weather forecasts. And that’s what the Apple Weather app uses, so Apple Weather uses the same WeatherKit that, if you’re an iOS developer and you wanna make your own weather app, you would use WeatherKit for that. And so that was what we did, was come in there and work on WeatherKit.
Again, it’s kind of scary going from sort of a very niche, small, tiny company with what we thought were a lot of users, but not compared to Apple, and then going to this giant company. Yeah, it was a little stressful.
Warzel: What was the reason to leave and start Acme? What prompted that?
Grossman: So I have been a huge Apple fanboy ever since I was a tiny little kid. Getting to go to Apple and work on that was, for me, a dream come true. It was just absolutely amazing. Everyone there was great. The problem is, it’s a giant company, right? So you go from, like, the smallest company in the world, where you could just do whatever you want, and then you go to an enormous company, where there’s a ton of stakeholders, right—you can’t do whatever you want. Myself and the other Dark Sky people just found that we missed the small, scrappy start-up days at Dark Sky where you could come up with a crazy idea one day, work on it the next couple days, and then just ship it out. And if something breaks or people don’t like it, you can go and you could fix it and you can iterate. I think we just missed that, right? And so it’s just not something you can do at a big company, whether it’s Apple or anyone else.
Warzel: What do you, or did you—maybe it’s the same—see as the current hole in the market right now for weather apps?
Grossman: When I left Apple, and there are a few of us at Dark Sky who ended up leaving around the same time, I don’t think we thought we’d get back into the weather business again. But then it’s kind of hard, having done it for so many years and then having to use someone else’s weather app, right? It’s just like, Oh, but I want the weather app to do this, right? It’s like, Why are they doing it this way? I wanted to do it this way. And so we ended up just getting frustrated with the existing weather apps. And so our focus at Acme is—it’s sort of the realization that all weather forecasts are going to be wrong, right? There’s nothing you can do about it. The key is: How do you convey that uncertainty?
My favorite UI for weather, by far, is your TV meteorologist. You watch her; she says, Hey, there’s a storm coming in, but the European model has it being pushed up to the north, and so maybe instead of snow, we’ll get rain in the afternoon. They convey the uncertainty. They tell you what may or may not happen. And I think that makes a huge difference, especially for storms, right? Pretty much every weather app on the market just says, Hey, here’s what we think is going to happen. And this is our best guess. It’s, “How do you convey that uncertainty, and how do you deal with it?,” I think, is what was lacking in a lot of weather apps, and that’s sort of our focus with Acme.
Warzel: I wanna dig a little more on this with the current state of weather apps. And as someone who’s made them, how has the need for information changed, you think, over the last decade, decade and a half, as weather has gotten more extreme? Is it just that people are just more information-hungry, you think, now than they were, or do you think that there’s actually a genuine need, given the rise of more unpredictable or extreme weather?
Grossman: It’s probably both, right? And it’s not so much that it’s more unpredictable—actually, weather prediction has been improving faster than the weather has [become] more chaotic, so weather forecasts are getting better over time. Everyone listening to this is probably going to complain and say, My weather app sucks; it’s not getting better, but statistically, they are getting better.
But yet, to the extent that there are more just things that impact your day, people are just sort of more demanding now, right? Again, you used to watch the weather—you’d read it in the paper in the morning and then watch it at night on the news and then hope for the best, right? And so I think, now that everyone has weather apps on their phone, I think they’re more demanding for the information that they need right now or in the immediate future, right, definitely, because I think people are checking it way more often than they used to.
Warzel: So three years ago, I spoke to this weather-forecasting consultant and he told me, “The general public has access to more weather information than ever, and I’d posit [that] that’s a bad thing.”
Grossman: (Laughs.)
Warzel: Agree or disagree?
Grossman: No, I—well, I don’t know the context in which he said that. But no, more information’s always better than less information, I think, right?
Information overload is definitely a thing, right? And so weather apps used to be very simple. It was just what are the current conditions and then maybe, like, an icon and temperature for the next 10-day. And now people are demanding more than that, right? And it’s not that having that extra information is bad. It just makes it more challenging on what do you do with that information, right? How do you convey that in a way that isn’t information overload? That’s really on the people making the UIs and presenting that data, right? I think the demand for more data is, I think, totally legitimate. If that data exists, give it to me and give it to me in a way that I can understand it, I think, is the way to go.
Warzel: The context of that quote was this thing that you had just said a minute ago, right, where people are like, Just ask why they suck, right, why weather apps suck.
Grossman: (Laughs.)
Warzel: And I’m like, Do they suck? I am a little bit frustrated on your behalf about this because it’s like—
Grossman: They do not suck. They are wrong sometimes, but I guess it depends on what you mean by “suck.” You can get into the statistics of it and be like, Okay, what’s the Brier score for your precipitation probabilities?, right, and you can measure things.
I think that, yes, always having your weather on you at all times does make it more obvious when it’s wrong. I think we notice way more when it’s wrong than when it’s right, right? When it’s right, it’s just like, Okay, of course it should be doing what it should be doing. When it’s wrong is when you get mad, right? And that’s what you remember.
So, yeah, I don’t know. They don’t suck. They’re getting better, slowly. Forecasting is getting better. But we contrast that with people are checking it way more often. If you’re just doing tick marks on how often it’s wrong, you’re gonna have a lot more tick marks now just because you’re checking it way more often.
Warzel: I sort of agree with this. I think that it’s that people want certainty; they want definitive. And I think this is just the way that things are right now, right? We are in a moment of low trust, right? Just broadly speaking, in the world—I work in news. It’s a moment of relatively low trust of institutions of all kinds, right? They want something definitive when things feel uncertain. And I think, at the core, nobody can offer a truly definitive thing. Do you agree with that?
Grossman: People would love certainty, but I think what they’re really after is, if it’s uncertain, they wanna know that, right? It’s sort of, What is your certainty around your certainty of your forecast? Right? And I think that’s what people really want. If there’s a storm incoming and it’s just different models are saying different things, it’s very different for a weather app to just make a guess and be like, Okay, I’m just gonna go with this and give this. It’s a very different thing to say, Okay, look, the forecast is uncertain now. Here’s what might happen. Here’s how you can prepare yourself. Same amount of certainty in both cases, but being able to actually convey and tell people, We are uncertain, is, I think, a form of certainty—I’m trying to figure out what the right word is, right? I think people want that information. If it’s uncertain, they wanna know that it’s uncertain.
Warzel: You said that these forecasts are getting better. You mentioned machine learning and artificial intelligence. What, as you see it, is the impact right now of AI? Is AI actually making these forecasts better? Is it giving you, as someone who’s running their own service, more opportunities to crunch the data better, organize it, present it? What is the generative AI stuff doing for you right now, as someone building this?
Grossman: Yeah, so there’s different places to insert AI and machine learning into the forecasting. The big one is using it to do these numerical simulations of the atmosphere. The benefit there isn’t outright getting better forecasts. The real benefit is, it’s computationally orders of magnitude more efficient and faster to run a forecast. Doing the physics is just ludicrously expensive, and AI can do it at a minuscule fraction of the cost. And what that gives you is (a) you can run these much more frequently. So something like [Global Forecast System] GFS, which [at] the National Weather Service, [is] their global model, that updates four times a day, right? If, with AI, you could do it once an hour or once every half hour, you could get much more rapid updates, which is important for things like extreme weather, right? If you have a storm coming through the Midwest and it could spawn tornadoes, you want the best, most up-to-date forecast you can, right? And so doing it faster is huge. Because it’s so much more efficient, you can do it at higher resolution—you can capture more of those microclimates and potentially get better forecasts just by doing it that way. And so I think that’s where AI is helping.
And I should note that when we say “AI” here, we don’t mean plugging in data to ChatGPT, right?
Warzel: ChatGPT, yeah, exactly.
Grossman: These are weather-specific machine-learning models. And so what we do is we take those models, the model outputs, and then we use machine learning to do things like microclimate adjustments so that we can take advantage of high-resolution terrain data to give you better forecasts. We do it for generating thunderstorm probabilities, precipitation probabilities, and so we train models to do that.
What’s, I think, really interesting—we haven’t done this yet, but I think generative AI, things like ChatGPT, might be able to help convey that information. Again, like I said, I think the best UI is your TV meteorologist, right? But maybe, with the new on-device models that are coming out, things like that, maybe it could figure out how best to convey that information, how to convey the uncertainty. If it knows who you are and what you care about and that you walk your dog every morning and every evening, maybe it can help you tailor the forecast for that. And I think that’s more speculative, but there’s different places where I think machine learning can slot in and it can help in each one of those steps to make it better.
Warzel: It sounds like the project of Acme Weather right now is, as we were talking about, to not just convey the uncertainty, in a way, but to build some of that trust, right to work through that. And something that this makes me think of, and, again, not to get overly political, but the government is what collects a lot of this data, right? There’s been a lot of change in the government, a lot of shake-ups around research, but also around funding, cuts to—
Grossman: Data collection, right? Satellite, earth science, yeah.
Warzel: Yeah, cuts to different government organizations that may or may not be collecting this information. Does that offer concerns for the quality of the forecast?
Grossman: Anytime projects and funding gets cut, there’s downsides to that, right? You don’t have as much data that you would otherwise have. Or maybe, with proper funding, you would have gotten new satellites that have new capabilities that can push forecasting further, and then you just end up not having that, right, and so the improvement in your forecasting isn’t where it needs to be.
That’s what I’m worried about, is things like that. I’m not worried about politicizing the data itself, right, ’cause I don’t think I see much of that, but I think the issue is, as funding gets cut, there’s less we can do, less data that we can collect.
Warzel: Do you feel like that makes your job more difficult, as someone who’s building one of these things, if there are those concerns just out there in the ether?
Grossman: It just adds uncertainty, right? I’m an optimist. think we’re gonna muddle through. I don’t envision NOAA just dropping all their weather forecasting, right? If so, we rely so much on their data collection and data from other organizations. I worry about it in the abstract, but I don’t think it affects our day-to-day yet, and again, fingers crossed.
Warzel: What would you say to the person who, again, this is similar to the Do weather apps suck? or whatever, and I’m not asking you to defend them, but in terms of the state of this particular slice of the weather industry, the weather apps, what’s your message to them right now? Is it “Trust us”? Is it “We’re getting better”? Is it “Tell us exactly what you need”?
Grossman: Yeah, well, that’s the thing about the weather space is, especially weather apps, is everyone has their platonic ideal of what they want their weather app to do and everyone’s idea is different. Our pitch is: If we’re wrong, we don’t wanna surprise you that we’re wrong, right? If we’re wrong and that’s surprising to you, then I think that’s a failure on our part, right? We wanna tell you if we think we’re going to be wrong so that if we are, you’re not like, Goddamn it, you ruined my wedding, right? I want to avoid that, right? And so I think that is what we’re striving for, is to not catch people off guard. But if you are and you think the weather sucks, please let us know because, again, that’s the best way to fix it, is for people to yell at us.
Warzel: Adam, this has been extremely eye-opening and informative, and I feel like I have a better handle than I did when we got into this conversation about what the heck’s going on when I pull to refresh on my phone. So thank you so much for this.
Grossman: Thank you. This was fun. Feel free to email me if the forecast is wrong.
[Music]
Warzel: That’s it for us here. Thank you again to my guest, Adam Grossman. You can email him, but please be nice if his weather forecast ruins your day. If you liked what you saw here, new episodes of Galaxy Brain drop every Friday, and you can subscribe to The Atlantic’s YouTube channel or on Apple or Spotify or wherever it is that you get your podcasts. And if you appreciated this work and you wanna support it and the work of all my other colleagues, you can subscribe to the publication at TheAtlantic.com/Listener. That’s TheAtlantic.com/Listener. Thanks so much, and I’ll see you on the internet.
This episode of Galaxy Brain was produced by Renee Klahr and engineered by Dave Grein. Our theme is by Rob Smierciak. Claudine Ebeid is the executive producer of Atlantic audio, and Andrea Valdez is our managing editor.
The post Why Is It So Hard to Make a Good Weather App? appeared first on The Atlantic.
