Twitter's Next Move - Disaster Detection

Twitter is fun. Twitter is informative. Now let Twitter be helpful.

The unique nature of a tweet makes it one of the best tools for detecting disasters. A tweet is:
  • delivered in real time
  • geo-tagged
  • keyword-tagged
  • notification enabled
  • human powered
Credit goes to various miners (the data kind) that have been sorting through the daily input of 140 million tweets a day. Cross reference these data miners with disaster detection initiatives and one quickly discovers the likes of TED. Twitter Earthquake Detection system is a premature project undertaken by the U.S. Geological Survey that parses Twitter data for keywords. Problem is latency. Since TED is solely dependent on data from Twitter, it cannot be relied upon for real time notifications.

I think Twitter can and should seize this opportunity and implement an in-house Disaster Detection algorithm. Why should Twitter do this?
  • Because its Twitter
    See reasons above that discuss a tweet. Add to that Twitter's userbase of ~200 million that pumps and digests 140 million tweets a day. The amount of incoming data and ability to reach so many people are reasons enough.
  • Twitter is the 'first receiver'
    TED (and the like) ping the Twitter Search API, download data, analyze data for false positives, cross-referene it to figure out an epicentre; only then do they publish it. Meanwhile Twitter can do all of this in much less time; its the first one to see the data and it has an infrastructure that has withstood 456 tweets/second
  • Twitter has real time notifications
    Disaster response requires notifying to all potentially victims. Today, this can be done through the internet; either tweet to them or send them a text message - capabilities that Twitter already offers its customers. This is a very obvious life saving tool.
  • Twitter can already do this
    Twitter already parses and indexes all tweets before pushing them to the universe. Leverage that to figure out epicentre of the affected area and notify residents.
  • Because onlyTwitter can
    Main source of data, the people, are on Twitter not for social service reasons but due to reasons of fun. If they can have fun AND contribute to saving lives, people will be glad to do it. Any new service that tries this cannot be nearly as successful - people would sign up because its the right thing to do but there is no value proposition to make them stay. On a different note, Facbook and Google+ simply cannot do this given their mix of privacy settings and value proposition.
  • Twitter needs to make money
    All services need funding to survive. Quick googling reveals interest of US Geological Survey. Charging third party services for real time notification is another source of revenue. I'm sure Twitter employees responsible for making money will discover other sources down this path; as long as they remember: do no evil
What I am proposing Twitter do:
  1. Parse and index tweets - Looking for keywords; "help, quake, fire, broke, save" come to mind
  2. Figure out disaster - If a certain number of tweets within a given geographic region contain markers
  3. Notify parties of interest - Notify users in affected area. Make an API that allows third party services to be notified - these service may include, for example, contacting local emergency personnel.

If twitter were to formalize such an implementation, not only does it stand to gain in number of users, business, and reputation, it would also be directly making the world a better place to live in.


Just do it, Twitter.