Very much so. I would use it to time to the minute when to leave on my motorcycle to avoid rain, but it's less accurate a couple days out. But it seems it usually fails pessimistic, which is just what I want to be prepared. (And frankly, with a name like Dark Sky, it's just what I would expect!)
I thought it was well-known that consumer weather forecasting is almost always pessimistic.
Forecasting sun when it rains generates complaints; the other way around, not so much.
Personally, I prefer it when they give percentages (which I assume are still pessimistic). The Met Office appears to forecast "rain" on a "30%" chance of precipitation.
Consider me one of the 10k learning that today :) It certainly makes sense to do so, but I had doubted it simply because of the many times people would argue with me about how "it wouldn't be that bad/cold/rainy". I assumed one should plan for the worst, but many liked to be reminded of bright days ahead.
Can you qualify in what ways they're better? A brief look at them showed they have a nice rain animation but little indication of snow-line; multiple forecast graphs, which strikes me as unhelpful; and somewhat friendly phrasing of written forecast information, but nothing radically better than I've seen on, say, wunderground. They don't seem to particularly care about temperature - in particular they draw no particular attention to freezing temperatures. In terms of data presentation, there's nothing as original here as the 'storm distance' timelines on weatherspark, or the snow depth and wind speed probability density graphs you get from weather.gov.
The use case here is "Hey, I'm going to go outside in a few minutes, but it looks like rain out there: do I need to worry about rain? Should I rush?" (Alternatively, if it's already raining: will it let up any time soon?)
Longer-term forecasts and temperatures are incidental.
My user script https://greasyfork.org/en/scripts/4699-forecast-io-auto-show... expands the detailed forecast for today when you visit the website, without you having to click it. I use the detailed forecast often, to check if the weather would be significantly different if I delayed my outing.
A couple of immediate findings visiting forecast.io on my phone just now:
1. It defaults to my ISP's location 400 miles away in London, which is understandable but doesn't give me an obvious way to correct it. Can't select my city shown on the map or enter coordinates.
2. Despite guessing that I am in the UK it defaults its temperatures to Fahrenheit, which is utterly alien to Europeans.
There's a big blue bar at the top of the screen which contains an input field you can use to search for a different location. It starts out saying "Current Location", but you can adjust it. There's also a units switcher in the top right of the page.
Disclaimer: I visited forecast.io on my phone for the first time in years to check this. I may have missed subtleties of it not working.
"Continuing down that road would lead to unimaginable stress, burn-out, probable heart disease, and a slow and steady descent into functional alcoholism. That’s not where we want to go."
Dark Sky has been letting me down in some pretty grandiose ways over the last few months, to the point where my wife sees my unbroken faith in it as a kind of running joke.
The "Are they going to fuck things up?" section at the bottom of their post certainly sheds some light on why the reliability has dropped off of late.
For whatever reason where I live in Brooklyn is completely unreliable. If the problem is scaling, I guess it would makes sense that they're choking on the volume of users in the NYC region
Glad to see that instead of selling to the weather channel or something they're continuing to put engineering muscle behind it. Dark Sky & Forecast.io are just awesome and I really want to see these guys succeed.
Agreed. About two years ago I used to use Dark Sky Inwood amaze people with how freakishly accurate it was.
At this point I've given up. I use Forecast.io because it's a well-made app (except for no iPhone 6 update, very sad), but I completely ignore the forecast for the next hour. It's basically never correct, predicting rain far more than it ever happens and often telling me that it's raining when it's not.
I gave Dark Sky an honest shot for a few weeks, but their reliability was atrocious. Could you please explain what is responsible for your "unbroken faith"? Is there something about the idea, app, or people that I'm missing here?
Dark Sky when it worked was almost magic - it could predict precipitation to the minute. An alert would go off and you'd know you have 5 minutes to seek shelter before the rain started.. it was freakishly accurate.
The developers are an indie app store success story as well. Started the project on Kickstarter with a new way to present hyper-local weather data down to the minute, in an industry dominated by a few big players and conventional hourly forecasts based on region. Based somewhere in upstate NY if I recall too
This is fascinating from a startup funding point of view, especially the rather bracing bluntness in the last paragraph ("Are they going to fuck things up?").
Heh, the check's already in the bank. What can they do to us now? (Just kidding. I actually ran the post by them and they had no issues at all -- I'm honestly not sure what I would have done if they did have a problem with it).
That was very enlightening. I'm surprised that a there have been enough one time purchases through the app store to cover two developer's salaries, Apple's commission, and a hundred-ish Linode servers over the course of three years.
I would think with a one time purchase each user becomes increasingly unprofitable (until they eventually reach a negative LTV) and the operation becomes untenable. Does Forecast.io monetize in ways other than the Dark Sky app? (I admit only being vaguely aware of them as a non iPhone user)
Dark Sky is kind of an expensive app, as apps go: $3.99. Which means we have a small percentage of the market, giving it a very different growth curve from free apps. Our rate of revenue has been slowly increasing since we launched. And since our market share is so low, I don't see us running out of potential customers any time soon.
(And to answer your other question: Forecast.io makes some money, but not much. And we do have an increasingly large revenue stream in our data API)
Hi Adam, this is a fascinating post, thanks for writing it! I'm Jacob, of PressureNet, and in fact Dark Sky is the reason I'm still building weather software! You beat me to consumer-focused radar postprocessing (I was working on this using McGill's radar data in 2010, and stopped when you launched) and you did a really good job - such a good job I took a look around and started collecting barometric pressure data in a new attempt at making the best hyperlocal weather forecast! You and your team are inspiring. :)
I'd be thrilled to talk about your data API and forecast tech, especially if you're considering working with phone sensors to improve your forecast models. I'm jacob@cumulonimbus.ca, have a look at http://pressurenet.io if you're interested.
Thanks for building great products and writing great posts!
CEO of Riskpulse here. I've been through all of the monetization trenches for weather software and we are now capitalizing on real demand in B2B/enterprise. If either of you guys ever want to connect to talk shop or (in)formally partner or share ideas, let me know: matt@riskpulse.com
The service engine I've been building to power http://www.ifweather.com is powered by their API. I'm not sure I would have picked their API originally had I known the company was run by just two guys, but I am very pleased with the API so far. I'm happy to hear that the API is "going to be a big priority for us going forward". Many best wishes and best of luck to Adam and Jay as this new sunrise for your business begins.
You might use this instead of a simple IFTTT alert exactly because IFTTT is so simple. In other words, ifWeather.com is designed to be more of a weather alert management console for people that would like to setup and manage multiple alerts for multiple locations. I think that we in the technology industry forget that for so many other industries the weather really matters, and not just the weather where you or your phone are.
A couple examples:
- a surprising percentage of people that have signed up so far are photographers that want to be alerted to ideal forecast conditions for several locations.
- My wife is a hydro-geologist, so a lot of her sampling needs to be done after specific weather conditions. Her sample locations are spread all over a large region and it's a real pain for her to check weather conditions for so many locations daily.
I love the app but I've found the data feed can sometimes be spotty (maybe this is just in Chicago)? For example, it's snowing outside right now and Dark Sky shows no precipitation for the next hour. Yahoo Weather correctly showed it was snowing (and showed it was likely to continue for several hours, so I parked in the garage).
It would be useful to have some insight into the underlying data, it's freshness, and an easy way to report inaccuracy (I've sent support emails but they've gone unanswered -- I assume the support load is enormous for a small team).
I've found the near term forecast a bit flaky in New York as well, though I've always imagined the city to be particularly unpredictable. I still use it regularly, but I hope they'll devote some of their new resources to improvements to the core forecast algorithm.
I've been using their APIs in the past, they're definitely the most developer-friendly forecast service that I know of. It's good to see their hard work is paying off!
Congrats on the deal. I've been using Dark Sky for about a year and it compliments the longer term Weather iOS app fantastically. I can't tell you how many times I've been about to go for a walk only to check the app and see a heavy 10 min shower starting in a few minutes. It's great being able to delay a few minutes and not get soaked!
Why not for Android too? This looks like the kind of weather app I'd start recommending to other people... problem is, all my other people are Androids.
As a user of their great API I have asked myself (strangely very recently) what would I do if forecast.io ever disappeared... Hopefully with this change I'm less to have to find out.
Please keep the API up and running! And congratulations!
The API is going to be a big priority for us going forward. We have some new development planned on that front, and AI will be able to help us out tremendously.
Congrats on the news! I was one of the kickstarter contributors and love the service. Best wishes to your future growth. It's amazing to see a KS campaign blossom.
This is a shame. They are radically better weather experiences than any you've seen before. Go check them out now.