ExtrACT!

Following on then PhD candidate Sara Wylie’s year of fieldwork on endocrine disruption in the western states of the USA, we identified several points of possible intervention, focusing on places where key community decisions might be assisted with better information. 

Why did these communities need better information? Rural communities typically have fewer local media sources, and few modes of infra-community communication. A large multinational corporation coming to Rifle, Parachute, or Leadville, Colorado, knows a great deal more about oil and gas extraction than anyone in those communities. Indeed, the business of extraction is an information intensive one, with a great deal of science and engineering going into the lateral drilling, hydraulic fracturing, and processing, storage, and transportation necessary to make use of methane or oil. In contrast, most communities know very little about the process before, during, or after their communities are impacted.

We built and tested three information interventions in the space of extraction: Landman Report Card (LRC), News Positioning System (NPS), and WellWatch.

Landman Report Card

First we found that communities where fracking had been going on for some time often said they had been misled about the costs and implications of drilling and extraction. By the time industry had really moved into an area, as in much of Colorado or North Texas, the costs were clear: environmental damage like explosive tap water or volatile organic compounds (VOCs) in citizens’ (and their livestocks’) blood; irreparable soil erosion; dangerous “orphaned” wells left unmaintained after an inevitable economic bust. Citizens who we spoke to talked about how they wished they could go back in time to when “landmen” — the used car salespeople of extraction who act as the folksy initial representatives of industry — had first approached them. These residents claimed they would have made different choices about how to engage with industry. Of course, barring time travel, we could not help these already affected communities; indeed, we did not know if the memories of these citizens were accurate. But could we build a system that could record their early interactions, and also share what they remembered about landmen and the negotiation process? Luckily we had a model: Yelp.

In Yelp (or similar review sites), consumers give each other tips about restaurants. If food or service are good or terrible, if atmosphere is declining or improving, if a restaurant has closed or changed names, Yelp is an unaffiliated way to get information that is far more current and detailed than Fodor's or an occasional newspaper review. Indeed, the existence of sites like Yelp provide a platform for generating information that otherwise would not be searchable or preserved. We wanted to extend the same functionality to landmen. Landmen are notoriously difficult to hold accountable. We were told that they change their names from community to community; they might work for multiple companies; they misled or gave incomplete information. Again, we didn't have much evidence of this, but by creating a platform for recording experiences, we hoped to generate enough data to allow individuals to make more informed decisions or take collective action. 

Moreover, this information would be able to jump across the silos of rural communities that do not usually have information flows between them. If a landman moved from New Mexico to Wyoming, for example, the platform would allow their reputation and relationships to move along with them. Of course, given the opinions of citizens who have worked with landmen, we knew we could face issues of libel. The same rules about platforms that protect Yelp and Facebook applied to this project as well, and we sought help from the Berkman-Klein Center's clinical law program to make sure that we did not intervene in the presentation of information from communities. 

All the reports we received are available in Wylie’s LRC appendix to chapter eight of her excellent book, Fractivism.

News Positioning System

Another phenomenon we witnessed in communities that had been struggling with oil or gas extraction was the emergence of activism. The extreme nature of some of the impacts of fracking (houses exploding, school-wide nosebleeds, outbreaks of rare forms of cancer, spills and off-gassing) pushed citizens to find ways to take action. Often an activist would express surprise to us at their own roll: for example an energy-independence supporting, dyed-in-the-wool, pro-business Republican would find herself suddenly working to defend her community against a BP or Encana, even as she was uncomfortable with the moniker "activist."

One technique that these activists seem to have invented independently in many communities was to fill a binder or scrapbook with articles about the impacts of extraction, culled from local newspapers. Road degradation from hundreds of loaded trucks per day, health impacts, explosions or leaks —  local papers are often filled with examples of how industry externalizes its costs onto the places where it works. An activist might have dozens of articles from their county and adjoining areas saved it the binder, ready to present to a journalist or congressperson, but possess no knowledge of the adjacent impacts another activist is archiving in their community down the highway. With engineer Dan Ring we created a system for bookmarking and mapping news stories, similar to Pocket or Delicious, but integrating the reports onto a map and using Thomson Reuters' OpenCalais AI-based tagging system.

The result was a rich map showing the incredible density of societal and environmental impacts of oil and gas extraction in the United States (and the world). Within months we had an aggregated, citizen-built database that had not existed before.

WellWatch

WellWatch came out of conversations we had with SkyTruth [https://skytruth.org/], an organization devoted to using satellite data (remote sensing) to understand and report upon industrial environmental damage. John Amos, its founder, speculated that people could "adopt" a well and monitor it as citizen scientists or reporters, to better understand and analyze the scale and impacts of industry on locations and their communities. We began to work with a technology called Semantic Media Wiki, which would allow us to quickly build forms and filters for contributing to and viewing information about wells, chemical pits, compressing and processing stations, and other parts of the massive industrial infrastructure spread across the nearly 40 states with oil and gas extraction in the USA. 

All the ExtrACT sites emulated what were called "web 2.0" style sites, like Yelp or Ebay. Web 2.0 refers to sites that primarily contain user-contributed information. Most of the pages on Amazon are generated by businesses, but most of the information on reddit is contributed by people who are mostly reading other people's posts. This leads to a chicken/egg issue: people generally read much more than they contribute. How does a site successfully get enough information to attract readers in the first place, who might eventually contribute something. Key to a web 2.0 site successfully growing is providing initial information that will draw a critical mass or readers, some of who will contribute and grow the site.

As large numbers of people are affected by oil and gas extraction, many people have information to contribute. But what makes them come to a site in the first place? And how do you make it easier for them to contribute? We speculated that we needed to make it easy for them to, for instance, enter their address and view the wells around them. Luckily, states maintain databases of every oil and gas well, often going back to the nineteenth century. Unluckily, the databases are often maintained in conjunction with industry, and they are notoriously unfriendly to users. Texas, for instance, serves hand-written scans of forms, making them impossible to search. Texas also charges tens of thousands of dollars for access to its database, though some of it is available in web forms.