Harnessing Social Media For Disasters

This article on social media and the Haitian earthquake points out one of the problems that I was forecasting with the Alert Retrieval Cache (ARC) project. Bear in mind, too, that I'm not just a geek. I'm a former Navy Corpsman who was BLS-I, ACLS, PHTLS, and administrated/supervised the NTC Orlando ED for 12 hour shifts. I also turned the other way when 2 juniors were racing wheelchairs in the parking lot.

The use of SMS text messaging as related to any disaster where the mobile phone infrastructure is intact afterward is one that has been apparent to many people on a global level. It's even starting to be used beforehand instead of a last minute implementation by civil society. This is a good thing. Even so, other social media factors in as well. Email, Twitter and Facebook have been some of the tools I've seen used.

Oddly enough, a closed email list (CIVIC) had to have a message forwarded to me so I could respond about SMS in Haiti. And the person who forwarded the message got my attention immediately with my... mobile phone.

Connecting: Receive

The system should receive all messages from all available systems. This, of course, means a lot of spam. On Twitter, as an example, #Haiti quickly got cluttered up. If we think of the '#' as a channel, it's a lot like CB radio when some moron starts whistling into their microphone.

So what to do? There are three main areas: Proximity, Trustworthiness and Relevance.

Proximity can be tricky - because a person who was at the center of the Capital may have great information to send out - but a doctor in a hospital 20 miles away is likely to be as 'close' to the actual disaster in terms of disaster rescue/recovery. A doctor who needs Normal Saline IVs immediately is just as important as one person texting their way through the dirt simply because the doctor can save more than one life with those IV bags.

And that's why proximity has context, and each proximity and its contexts has to have a separate channel of communication. You can work around it as long as you want and limp along by having algorithms scan messages... but even then, the right humans have to read the right messages. And even electronically, it makes sense to have the different channels simply to attenuate the right signals getting to the right people.

Having it available in raw form is invaluable because data can be extrapolated from it that can save even more lives and/or improving the quality of lives.

Trustworthiness is a matter of experience - a source is trusted if they have been proven trustworthy in the past and more so if other sources are reporting the same thing. But the real gems of information come from individuals reporting things on their own. Thus it makes sense that we raise the bar of everyone and re-assert the need for proper communication in education. Brevity now also has its upside. U cant wax poetic in a txt msg.

Relevance is going to shift in sources as different issues become more relevant. This will also make proximity shift as well. At the start of a relief effort, finding safe areas near to the affected area(s) to set up is most relevant and most important. After that, human life gains prominence. And after that, meta level human life gets into play: Saving and helping the most people. These are different levels of administration. It's a good start to explore disaster management (or more often, manglement) but would be a long write/read.

When a disaster first starts, more attention may go to one area where everyone has mobile infrastructure that permits them to get messages out. The more important issue will always be finding those that do not have that infrastructure - thus alternative communication must also be routed into the relevant channels.

Connecting: Transmit

The system should also send data to other systems.

So now that this part has come into being, we get into Phase II: Finding out which reports to trust and doing something useful with the information. When ARC first started, Sahana, the Free and Open Source Disaster Management System, was being written in the field! This seemed and still seems like a system such an emergency SMS system should be linked to. I haven't followed Sahana as much as I wish I did but I don't expect it would be too hard to integrate. Software is just a bunch of ideas made electron states in silicon.

Emergency broadcast channels are a must for the SMS system as well - that way alerts can be sent, such as 'Medical supplies on the way for Port Au Prince: 4 hours'. That way doctors and other volunteer staff can stay appraised of the latest developments at the meta level - and make more informed decisions at triage and administrative levels.

Media is extremely important as well, so there should be a press release channel - the media microreporting every little text message/tweet/whatever is and should be avoided, so the press needs a stream of data as well.

That covers the basics of the system. I suppose I could write stuff in more detail, but... that would be a very long post. Maybe in pieces on the project page sometime...