Loyalty discounts, the power of recommendations, serendipitous encounters with friends and colleagues, recognition badges, and stalkers. I think that’s a fair summary of most commentary about the growth of location-enabled services and tools.
Location is just one piece of information that can be generated by most smart phones, but is the most relevant for a marketer eager to deliver precise and context-specific messages to a consumer on the move. It is also a highly useful data point for a social scientist trying to measure the flow of human migration and socioeconomic progress, as in the case of Nathan Eagle’s research in the slums of Kibera, Nairobi, Kenya.
Between June 2008 and June 2009, Eagle and his co-researcher evaluated the calls recorded by mobile phones across Kenya (with all callers’ identification replaced with unique hashed IDs) to focus on calls originating or ending in Kibera. Their research tracked between 53,000 and 74,000 calls a month and a total of 18,000 individual callers during the year.
What did this data reveal about individual mobile phone users? “With each call, we can infer a number of individual characters such as
- spatial data (by the location of the cell tower that transmitted the call),
- economic data (the average length of each call, the amount of pre-paid minutes an individual has put on their phone, the type of phone),
- an individual’s regional or tribal affiliation, and
- a radius of migration for groups of individuals (by the distance between locations of cell towers calls have been made from).”
A first indication from this research is that Kenyans only live in the Kibera slum for a mean of 1.559 months. This high rate of movement and population turnover “supports the theory that slums act as a filter as opposed to a sink where there is a large amount of flux within the slum population.”
Amy Wesolowski, Nathan Eagle, Parameterizing the Dynamics of Slums
Eagle’s work in Kenya is an extension of a research project originally conducted at MIT, where 100 students were provided with mobile phones for 265 days. The mobile phones were equipped with custom survey software that recorded data and prompted the students with questions when certain conditions were met.
How much data?
“From the studies, we gathered 370 megabytes of raw data, including short recordings from 667 calls, 56,000 movements, 10,000 activations of the phone, 560,000 interaction events with our applications, 29,000 records of nearby devices, and 5,000 instant messages.”
Thankfully, from a privacy advocate’s point of view, the researchers also had to struggle with (a limited number of) weak points in their data sets – instances when the participants didn’t bring their phone with them, consciously turned the phone off, or simply ignored it. I would like to think that some of this reflected a conscious effort to mediate information collection, but it was probably just fatigue or forgetfulness.
There was one significant distinction between the two projects: the active involvement and acknowledgement of the participants. In Cambridge, the participating students were walked through the information collection process, provided with details about the information that would be collected, and required to complete a consent form(.pdf).
M. Raento, A. Oulasvirta, N. Eagle, “Smartphones: An Emerging Tool for Social Scientists“, Sociological Methods Research 37:3, 426-454.
This is an important point when it comes to the collection of location data, especially when it is associated with other personal information: individuals want to know what is happening with their information, and would like some element of control over its use.
A recent and exhaustive examination of the 89 then-available location-sharing services (really, who can keep track?) by researchers from Carnegie Mellon University noted that “the willingness to share one’s location and the level of detail shared depends highly on who is requesting this information (or knowing who is requesting this information), and the social context of the request.”
Supplemental interviews confirmed that potential users had particular scenarios in mind when evaluating the benefits and risks of these services: scenarios that would best be addressed with more detailed privacy controls, rules and conditions (explained in detail in the paper):
- Friends Only rules
- Granularity of controls
- Group-based rules
- Invisible status
- Location-based rules
- Network permissions
- Per request permission
- Time-based rules
- Time-expiring approval, or
- No restrictions
Janice Y. Tsai, Patrick Gage Kelley, Lorrie Faith Cranor, Norman Sadeh, Location-Sharing Technologies: Privacy Risks and Controls
Obviously, there are significant gaps in how personal privacy is protected when information is collected and analyzed in a large scale research project, a smaller experiment and within the context of online commercial services.