Knowing where people are is one of the most important pieces of information in the immediate aftermath of a disaster. To help answer this question, Facebook has just announced a partnership with the International Federation of Red Cross and Red Crescent Societies (IFRC), UNICEF and the World Food Programme (WFP).
Through this partnership, Facebook will share aggregated and de-identified data about users’ locations and movements to inform situational analyses following natural disasters.
More specifically, Facebook will share data on:
- population density,
- population movement and
- the percentage of people who have marked themselves as “safe” in Facebook’s Safety Check.
As explained in a separate post by Facebook’s research team, Facebook gets that information from users who use the Facebook app and have location services enabled.
This approach combines a few ideas that have been developed and tried over the years:
- Mobile phone providers have shared data about user movements in a number of emergencies, including the Ebola response. See this post on the pros and cons of this approach. Facebook will now do something very similar with data gathered from its own users. One big advantage of the Facebook partnership is that the three partner organisations will not have to sign separate agreements with multiple mobile phone providers all over the world. Instead, they have a single point of contact. On the other hand, call data records can be mapped for all customers of a mobile phone provider, making their data more complete. Facebook can only provide data on the subset of users who use the app and have location services enabled.
- I know of at least one project that mapped a lack of responses to highlight areas that should be prioritised for physical assessments. Facebook is now doing something similar by highlighting the ratio between people who have marked themselves “safe” and those who haven’t. A lower ratio of check-ins could be an indicator that this area is more severely affected.
The part that I’m most excited about is the information about population movement. I remember how in Haiti, following the 2010 earthquake, aid organisations prioritised Port-au-Prince because that is where people had been prior to the disaster. However, the organisations did not fully realise that many had left the destroyed capital to stay with their families in the countryside. Many of them then returned to the capital comparatively quickly, because that was where the aid was. The result were overburdened camps with horrible living conditions.
Had aid organisations known how many people had fled the city, and where they had gone, aid could have been allocated differently. (Granted, this is painting a rather rosy picture and ignores the fact that some aid organisations did not want to work in the countryside. But that is a different matter.)
The image below shows how Facebooks data could be used to monitor populations movements on a daily basis.
Location data will be aggregated for 360 square meters, meaning that each data point shows the combined data of 360 square meters. I guess in urban areas that will be enough to ensure the anonymity of the data contributors, but I suppose in rural areas it might be possible to draw conclusions about the identity of people based on that data. The report “Ebola: A Big Data Disaster” contains an interesting discussion about anonymity and privacy in health crises, which in my opinion is just as relevant for Facebook’s disaster maps.
Personally, I think this is a really exciting development. I understand the privacy concerns, particularly in a health emergency, but given that the data is only shared with trusted partner organisations I think the potential use of this data outweighs the privacy risks. I would even advocate for including additional socio-demographic data in the datasets, such as gender and age. For example, I think it could be useful to know whether more young men migrate, while older women are left behind, because this could help shape the design of aid programmes. This type of information would also give the project partners a better idea of how representative the data is, particularly in countries where there is a significant gender difference in smartphone ownership.
On the other hand, I shudder to think how demographically enriched data could be abused in conflict situations. From Facebook’s announcement, it sounds like the partnership only applies to natural disasters at the moment and that is probably a very good idea. However, I can easily imagine that some governments will now put additional pressure on Facebook to share user data with their security apparatus. After all, if Facebook has a tool and an API to share data with WFP, then why not with the police? These are ethical dilemmas that will accompany us over the coming years.
Last but not least I find it worth mentioning that Facebook is not collecting additional data, but is merely making the data they already have accessible to aid organisations. Implying that Facebook engages in surveillance of disaster survivors misses the point. Facebook is constantly tracking its users, but at least now they are making these datasets available for humanitarian aid and not just to optimise ads.
What are your thoughts on Facebook’s Disaster Maps? Please share them in the comments!