Here at the International Centre for Security Analysis we are interested in developments in social media and how we can understand social networks as valuable information sources. We produced a report on this topic: A Structural Analysis of Social Media Networks which is designed to be a reference guide for analysts and policy-makers. We also produced a podcast where we discussed the concepts in the report in more detail.
One of the most interesting features of social networks is their role in facilitating the emergence of communities. In the report we draw a comparison between Facebook and Reddit on the one hand and Twitter and Instagram on the other. Facebook with its group function and Reddit with its subreddits both have dedicated site structures for communities to form. In contrast, Twitter and Instagram have no formal group structures built into their sites and yet we see large, cohesive and resilient communities on those platforms. How do these communities form and how do they survive?
Case Study: #junkiesofiggg
The answer, in short, is hashtags. Hashtags enable communities to emerge on platforms where there is otherwise no formal group structure. To demonstrate how extensive these communities could be we looked at one specific Instagram community: #junkiesofiggg. and its predecessors #junkiesofig and #junkiesofigg. This is a community originally made up of heroin users – “junkies” – that has since broadened to include a wide range of drug users, dealers and rehab centres.
The original #junkiesofig hashtag was banned by Instagram and one would think the community would simply disintegrate as a result. However, users realised they could add an extra “g” to the hashtag and when that was banned they added a third. Ingeniously, the #junkiesofiggg users have appropriated Instagram’s functionality for their own ends. When a user searches for the original banned hashtags they are directed automatically to the functioning hashtag. As a result, the community could rapidly reform.
We carried out an Instagram search for #junkiesofiggg on Sunday 2nd May 2015. This yielded 9,769 posts of which a sample of 293 posts from 75 users was analysed. This was a convenience sample with the research designed to act as a proof of concept. The dataset was analysed by hand for three characteristics: post profile, drug typology and post location.
Here we wanted to assess how many of the posts referred to drug use, drug dealing or drug rehab, sorting the posts into “users”, “dealers” and “rehabbers”.
As chart 1 shows, 44% of the posts were drug users, sharing pictures of the drugs they are taking. A relatively small number, 12%, were identified as dealers, often encouraging other users to direct message (“dm”) them for more information. The majority of the dealers identified were also users. Finally, 44% were classified as “rehabbers” comprising users in rehab or dedicated rehab centres using the hashtag to encourage other individuals to seek help. A small number of users posted pictures that identified them as users, dealers and rehabbers.
Second, we analysed the dataset for drug typology to see if it was possible to identify which drugs were most prevalent within the community.
Chart 2 depicts the main drugs identified in the dataset. Unsurprisingly given that it was originally a platform for heroin users to share images, heroin featured prominently, accounting for 22% of posts. A range of drugs were grouped together as painkillers which accounted for 26% of posts. Users often identified the drug pictured using hashtags but a significant number of drugs (39%) were classified as “unknown” although some of these could possibly be identified through further research.
Finally, we analysed the 293 posts for the location they were uploaded from. This was done in two ways. First, a relatively small number of posts were geotagged enabling a precise location to be obtained. Second, many users included place names as hashtags, such as: California, Seattle and Florida. As chart 3 shows, 60% of posts had unknown locations. This is not unusual on social media and in fact we would generally expect to see even fewer known locations.
The US dominated this sample with certain states heavily represented. Florida had the highest number of posts based solely on one very active drug rehab centre. Ohio had one very active drug user posting while California had one rehab centre and one active drug user posting with the hashtag. Interestingly, the highly active drug user in California also tended to use Justin Bieber related hashtags to increase the visibility of their posts.
Even without a formal group structure on Instagram, the #junkiesofiggg community is surprisingly large and resilient. It is able to recover from efforts by Instagram to ban its primary means of social identity. This case study also shows how Instagram posts can be analysed for a range of factors including post location, content and identity. Other factors could also be analysed including: profiles of the most active users, post timings, user relationships and the number of likes and/or comments each post receives.
The possible applications of research of this type is also varied. Researchers can use it to analyse the formation and interactions of communities on social networks. Law enforcement could identify areas of drug use and patterns of supply and demand, new drugs entering the market and the relationship between drug posts online and drug use.