Although #NoDAPL is a hyperlocal issue, it has received attention across the country. Briefly looking at the locations stated by Twitter users (in their bios), they seem to be geographically diverse. Given that the Dakota Access Pipeline is a domestic issue, I expect to see tweets mostly from the northwest hemisphere, specifically the United States. For tweets actually using geolocation, I would expect to see many from Standing Rock, from users actually at the site. Recently, a Facebook campaign to stymie law enforcement officials had users checking in at Standing Rock remotely, which created a glut of checkins in the area, making it difficult to determine exactly how many people were actually present in North Dakota. Given that Twitter uses actual location data, I do not expect to see a similar phenomenon on the site.<\/p>\n
My hometown, in Connecticut, is roughly at 40\u00baN 90\u00baW, relatively far from the epicenter of the protest. Given that Glastonbury is a small, suburban-rural community without major political activism, I doubt there will be many tweets regarding #NoDAPL in the area.<\/p>\n
I found 22\u00a0mappable tweets out of the 27,556 I scraped, meaning 0.079% of the tweets collected had a geolocation. As far as I can tell, all of the tweets are in the northwest hemisphere, most likely in the United States. Given that this is primarily a US issue, this is unsurprising. All of the tweets are in English, which is also consistent with my expectations.<\/p>\n
Given that law enforcement is known for using location data to track protestors, this may be a disincentive for people to share their geolocation. Standing Rock, the epicenter of the protest, is also remote, meaning that access to the internet and penetration of access to the internet may be limited.<\/p>\n
The small sample size leaves a high degree of uncertainty in any conclusions made from the data. Yau states that any assumption based on a small or incomplete data set runs the risk of misrepresenting the case. Especially considering the potential negative consequences of using a geolocation for those already at risk, I doubt the sample I have obtained will fully represent the actors involved.<\/p>\n
I don’t believe that my dataset is ‘big data.’ Using boyd and Crawford’s categorizations of “fire hose,” “garden hose,” and “spritzer,” my mappable data falls well below even the lowest bar, 1% of the available data. My data was also collected over a short period of time\u201414 days out of the roughly 2.5 years for which the Dakota Access Pipeline has been publicly known. boyd and Crawford do not set a hard limit for the size of big data, saying instead it is about the “capacity to search, aggregate, and cross-reference large data sets.” While my data has been aggregated and is searchable, there are no complementary data sets that can be used in conjunction. Therefore, it is difficult to argue that my data is ‘big’ either in scale or interconnectedness.<\/p>\n
Jordan’s #BlackLivesMatter hashtag had roughly 85,000 tweets over the same period of time, just over triple the number. #BlackLivesMatter seems to directly relate to a much larger population than #NoDAPL, meaning there is a higher likelihood for it to be recognized and talked about. The digital presence of both movements is important to consider as well. While the #BlackLivesMatter movement has taken the form of physical protests, it lacks a physical location much of its activity is conducted online. Conversely, the #NoDAPL protest has a highly specific geographic location in the Standing Rock Reservation in North Dakota and a continued presence there. Perhaps the physical presence has reduced the need (or at least the appearance of the need) to continue the conversation online.<\/p>\n\n\n