For the last few months I’ve been working on a new project in my spare time, which is called cumul.us. The (what I consider a) clever play on words revolves around two things: weather and people.
What is the idea? I could bore you with buzzwords such as aggregation, prediction markets, and the wisdom of the crowds, but the real point is to take advantage of these types of things in order to give you simple and accurate weather in a way that you can both use day to day, and also provide a way to make it a more interactive and interesting experience.
- Firstly, the site will combine as many possible sources of weather forecasts as possible. No one source is ever right all the time, so the idea is that if you aggregate them together, you don’t need to check several sources and you get a safer, more accurate forecast. If you also track all of these sources and check their accuracy over time, you’ll be able to actually see which ones are more accurate than the others.
- Secondly, you can predict the weather yourself. When you make prediction for a particular time and place, the site will go check all of its data sources and record what really happened, and give you a score based on how right you were. It could turn out that a random person is a better predictor of the weather than a professional meteorologist or organization. That person could even be you. Since the site will be tracking the accuracy of all of this, you’ll be able to see who is more right, and follow them.
- Thirdly (is that even a word?) the site will give you information on the real reason you check the weather: to find out what you should wear. As people submit what they are wearing, it goes into the aggregation of what everyone is wearing in order to suggest to other people what they should wear.
Will this all work? Who knows, but it only took me two months to make, and I wanted to find out. For now, I’ve been keeping track of like-minded posts on del.icio.us and thrown a few screenshots up on Flickr, but the site is slated to launch at the beginning of November. If you’d like a sneak peek, just send me an e-mail, because there won’t be any lame super secret beta site with invites to pass out on Techcrunch… it will just launch and that will be it.
sounds very interesting, would love to see it!! thanks.
“The Wisdom of Clouds”
Oh snap! Clever, clever…
Will you have an API so I can pull aggregate weather data based on ZIP code or geographic area? I’m doing this right now via Drupal’s weather.module which uses METAR (METeorological Aerodrome Report) data.
Very clever. Can’t wait to play with it.
Ben! This is a *great* idea. Can’t wait to see it in action. And, I’ll second the request for an open API. :)
Congrats, man!
Due to the overwhelming response, there may be greater delay and scrutiny of sneak peek requests… I was expecting to get jimray’d, not kottke’d.
An API is in the works, yes.
Oh, sounds yum yum. Once the traffic boils down I’m interested in taking it for the spin and giving feedback.
Nice idea – You can make the southern California page static with “Wear jeans and a t-shirt, it’s pretty nice out there.” That should save some development time!
Nice idea, and hopefully the implentation will be good too; I’m looking forward to seeing it.
One question; will this be UScentric, or do you have data feeds for the rest of the world? I believe Yahoo and METAR data is global, but I expect some of the forecasts not to be. (The three UK forecasts I tend to compare are the Met Office, the BBC (actually a re-presentation of the MO data) and Metcheck. None is a global source.)
this sounds like a fun project. might be fun to collaborate on some future idea together. i made an ambient weather widget that you might find amusing, http://stewdio.org/windmaker.
I love this idea, Ben. I’m a bit of a weather stats nut. A year and a half ago, I had designs on something similar to this, but it all got lost in a haze of other work. I started tracking data from local media outlets, the NWS and the online sites (that didn’t just cop the NWS data). I had hoped to find out who was the best at predicting. I stuck to temp because all other facets of the weather are so hit and miss block to block and site to site (“rain likely” versus “chance of rain” versus “50% chance of precipitation”).
Anyway, after two months of tedious gathering, I came to two conclusions, anything more than 5 days out was almost complete garbage and that the National Weather Service was the best predictor. I love the URL and I can’t wait to see it.
sounds super fun – let us know when the API is up so we can play :)