Emma Brown's Portfolio

A Collection of Open Source GIScience Work

View on GitHub

Ethics of Volunteered Geographic Information

What types of uncertainty are most prevalent in the use of volunteered geographic information (VGI) for research on disasters and humanitarian crises?

Social media data has become a prevalent system of knowledge during crises and can contribute valuable insight towards disaster relief. However, the use of this data must be critically assessed, as issues of privacy and power relationships are involved in the use and dissemination of this knowledge. While disasters are commonly thought of issues that affect areas similarly in a physical sense, social contexts can alleviate or prolong the effects of disaster. Yet there are many uncertainties that arise, and further, the ethics of using this data is important to consider and discuss. Crawford and Finn (2014) discuss the ontological, epistemological, and ethical concerns of VGI that arise when social media data is used to analyze and understand disasters, and the uncertainty that arises because of these concerns.

First and foremost, there are issues of uncertainty with VGI given that disasters often lead to power outages that limit the ability to share information. Crawford and Kim refer to this as the “signal problem,” where a dataset may be assumed to accurately affect the social world, but “there are problematic gaps with little or no signal coming from particular communities.” Further, the overwhelming and traumatic nature of disasters can lead to a partial representation of the event that reveal only certain aspects of the situation. VGI is also representative of only certain demographics — for example, Twitter users are more likely to be young adults thus knowledge is often from this perspective. Older, less wealthy, and those living outside of urban centers are less represented in social media data, resulting in data that Cruther and Zook (2009, as cited in Crawford & Finn 2014) observe is “colored by the fundamental divides pre-existing inn society and in some cases can amplify them.” Further, only a small percentage of VGI such as Twitter data is geotagged, which significantly limits data availability and increases uncertainty by virtue of the smaller sample size. Thus, this wide array of uncertainty from VGI must be acknowledged and begs the questions of who is being represented, where the data is coming from, and what are problems that may arise when using this data.

Do you think there is any ethical obligation in relation to uncertainty? Or, do you think that there are ethical concerns in using VGI that complicate uncertainty?

There are numerous ethical issues that arise when using VGI regarding privacy, transparency, and power dynamics. Oftentimes, fully informed consent is not given, as in the case of the Mission 4636 and Ushahidi Volunteer projects that occurred in Haiti after the 2010 earthquake. These projects allowed Haitians to send SMS messages about their conditions for relief purposes, however, there was not much transparency or information on how these messages would be used by the organizations. Most people were unaware that their messages were being published or even the name of the organization. This one-sided dynamic created a power imbalance in which local Haitians were largely left out of the picture. This is incredibly problematic considering that often reports written about VGI analysis are inaccessible to the local community whose data was used, whether it be a language, educational, or financial barrier. Consent is also unclear regarding VGI like Twitter data, where users are often unaware that their tweets are getting scraped and used. While people are able to manage their own settings and, privacy is often unachieved. It is also important to consider the ethics of using tweets sent during a traumatic moment. In this way, the meaning of “privacy” changes depending on the context and situation.

Sources: Crawford, K., and M. Finn. 2014. The limits of crisis data: analytical and ethical challenges of using social and mobile data to understand disasters. GeoJournal 80 (4):491–502. DOI:10.1007/s10708-014-9597-z

Main Page