It has been a month since Hurricane Sandy passed over the Caribbean and hit the US East Coast, and a lot of people have written excellent posts about the role social media played for either the affected population or the responders. Here are the articles that I found most interesting.
What the affected population expects
Kim Stephens wrote about what the affected population expects from social media in a natural disaster and what lessons responders should draw from this. Obviously, these lessons were learned in one of the technologically most advanced countries in the world, so it has to be taken with a grain of salt when looking at middle-income countries. However, two points stood out for me particularly:
- Citizens will use social media to ask for emergency assistance during large-scale disasters – and you need to have staff in place who can respond to these queries.
- People want hyper-local information. They already know there is a strom/flood/earthquake. What they want to know is whether the bridge in their town is closed or whether fuel will be available in their village tomorrow. Kim makes the point that this information is best provided by the community for the community, but disaster response operations should facilitate this process. (See also: the experience of the police in Queensland, Australia after severe flooding in 2010 and 2011)
Kim also shared the social media statistics from Fairfax County, Virgina, which make for some very interesting reading and show how you can serve your community via different channels.
How crowdsourcing helped with damage assessment
Based on the user-ratings, a grid was automatically created that showed the impact of the hurricane in different areas.
Patrick Meier shared details of a project where a team connected with Humanitarian OpenStreetMap got their hands on the aerial photos from the affected areas. They then put the photos up on the web, extracted the GPS coordinates and asked internet users to rate the damage visible on these photos on a three point scale of “light/none”, “moderate” and “heavy” damage. The result is a colour-coded map that shows which areas were most heavily impacted.
According to Patrick it looks like the data was indeed used by some government agencies, however I can’t help but notice that this is another case of someone setting up a technical system hoping that it will prove useful rather than finding out what would be most useful. Apparently, in this case it worked, but that doesn’t change the fact that I am no fan of the “build it and maybe someone can use it”-approach.
Having said that, I do think that the technology behind this project is interesting and could be easily duplicated in other disasters, including in low or medium-income countries as long as someone supplies the imagery.
Patrick also used Sandy to bring a study from earlier this summer into perspective that dealt with the question “What percentage of tweets generated during a crisis are relevant for responders?” The doctoral thesis asserts that as few as 8 % percent are relative, but Patrick makes the point that in the case of Sandy that still would have meant that 40,000 tweets generated in the first 72 hours contained useful information. Not bad.
A problem that became very apparent during Hurricane Sandy was that not only was there a lot of information, there was also a lot of misinformation some of which spread rather rapidly. The Atlantic shared a great collection of real and fake images which serve as a reminder that you cannot take anything for granted. If you want to know more about how to verify social media information, please look at the “Truth in the Age of Social Media”-report.
Do you have additional links related to Sandy that you would like to share? Please add them in the comments field.