Ushahidi is often deployed for crisis and disaster response, where information needs to be timely. The teams and individuals using Ushahidi in places such as New Zealand, Haiti, Japan, Australia (and many more) can tell you just how important it is to get information quickly so decisions can be made.
The value of crowdsourced information isn’t always found in its accuracy and it certainly isn’t always easy to make sense of, as people are writing quickly and are in difficult personal situations. The quality of crowd-based information sometimes isn’t as high as if it was from experts, but there is value in both it’s mass (size) and it’s immediacy.
I was a little surprised to see a post by Jim Fruchterman, CEO of Benetech an organization that we greatly respect and admire, talking about how crowdsourced data is not a substitute for real statistics. Those blog posts, and the Fast Company article that they spawned, might be good for making waves and getting attention, but it’s of little value if you want to really dig into the problem and find answers.
This academic debate started after the European Commission’s Joint Research Center did a study on Haiti’s crowdsourced information and found a pattern of aggregated text messages predicting where the damaged buildings were concentrated. The crux of the argument is that this “positive correlation” either is, or isn’t, true. What I find more troubling is that this was a study specifically focused on Haiti, and I’m not sure why Benetech felt it could be extrapolated out to every other earthquake scenario?
Sitting back and thinking about this blogging dialogue, I can say that I too am very interested in knowing if there is an aggregation and prediction pattern that can be used in realtime. If so, responding organizations will be more effective with that knowlege. If not, then it won’t be and that is also good to know.
I’m concerned with how Benetech is positioning their answers on the same findings as being clear cut and right. My colleague Patrick Meier has written a piece digging deeper into the aggregation and patterns studies. He makes a clear case on just how unclear Benetech’s findings are compared to the European Commission’s.
The problem with academic arguments is that they disregard the reality of the mess that is a crisis response. If it were a clinical environment where we all got to sit back, sift data and take our time to make a decision, then Benetech’s expertise in cutting-edge quantitative research on patterns of human rights abuses is undeniably useful. In this case, it isn’t.
The problem is this; we don’t have the days/weeks/months needed to sit back and gather all of the statistical data before a decision is made. In a disaster situation, such as the above referenced Haiti, text messages are important and have a great deal of use at that moment. It doesn’t matter if aggregation patterns might/might not prove to be valuable, because we still haven’t seen anyone use that information in the real world.
Again, I too want to know if there is a true pattern on buildings and text messages. Since Haiti, there have been a couple more large earthquakes and the Ushahidi platform has been used to aggregate data around those as well. I’m sure we could convince the good people who did the New Zealand and Japanese response to share information, and that might give us enough data to make a more educated deduction. A sample of three is more valuable than a sample of one.
Either way, It’s worth pointing out that we’re further ahead than we were before Ushahidi started crowdsourcing crisis information due to the fact that we have realtime information, from the ground, that allows decisions to be made.