In a bit of a digression from studying people abusing each other on Twitter, after I witnessed the progression of a particular hasthtag that was instead about how people are encouraged/outraged enough to report real-world abuse via social media platforms – I decided to employ my skills to find out how and why this happens, and the fascinating life of a particular social network meme.
I first came across the #ididnotreport twitter hashtag in early March 2012. The thought-provoking, emphatic and sometimes controversial reactions to this hashtag (which I discuss below) show just how powerful soical media platforms like Twitter can be in getting topical issues into the wider discursive forum. The story behind the #ididnotreport hashtag is quite a powerful one, and fits within the wikipedia classification of an internet/social network meme:
An Internet meme is an idea that is propagated through the World Wide Web. The idea may take the form of a hyperlink, video, picture, website, hashtag, or just a word or phrase, such as intentionally misspelling the word “more” as “moar” or “the” as “teh”. The meme may spread from person to person via social networks, blogs, direct email, news sources, or other web-based services. An Internet meme may stay the same or may evolve over time, by chance or through commentary, imitations, parody, or by incorporating news accounts about itself. Internet memes can evolve and spread extremely rapidly, sometimes reaching world-wide popularity within a few days. Internet memes usually are formed from some social interaction (Rage comic or reaction faces), pop culture reference (Xzibit in “Yo Dawg” or Bear Grylls in “Better drink my own piss”), or situation people often find themselves in (Socially Awkward Penguin or Futurama Fry / Not Sure If X).
Source: Wikipedia: Internet Meme
Bearing this in mind, I wanted to see if I could use visual network analysis tools to study how this meme transferred across social network platforms:
1) from its beginnings in the blogosphere,
2) to Twitter,
3) then to news and social commentary sites, and
4) perhaps a transfer into the audio/visual medium/sphere of the web too.
Finally, in the space of at least the month I had to study the #ididnotreport hashtag, the question of whether it would persist: would its spread across many different web platforms mean that it hung around longer than other trends that were based in singular platforms – indeed would an ‘issue network’ be created, and would this permeate outwards into more official organisational and governmental web-spheres?
Creation of the #ididnotreport Meme
During the first weeks of March 2012, the blogger Julian Norman blogged about the Mumsnet campaign “We Believe You,” and linked the under-reporting of rape to the non-reporting of other assaults. The previous week, a piece had been published in the Guardian about street harassment of women, and the number one theme among the commenters was: We don’t believe you.
Reading the comments on that article had annoyed Julian. She argued that
“everyday harassment is something that women in the UK experience routinely, to the point where we barely register it. We certainly don’t bother to report it – or at least, I don’t, and my friends don’t. Who were these people saying that they simply didn’t believe it happened?
When I shared the blog post on Twitter, another user tweeted back to say, “In my whole life I’ve never reported it – breast grabbing, hand up my skirt, etc.” In a spontaneous moment, I tweeted, “#ididnotreport the commuter who stroked my bottom on the central line” – an event from only the previous day. It was just one example that popped into my mind; I could have used a number from the last month or so. I invited others to share, using the hashtag. And so it was born – of nothing more than frustration at the levels of disbelief and a sadly large supply of material.
Where it’s easy for commenters to dismiss one woman as a liar, it’s less easy to dismiss thousands of accounts, and thousands there were – 500 in the first 24 hours, climbing to over three thousand. (Editor’s note: According to the social data companyPeopleBrowsr, the hashtag was used 5,468 times on March 12. Since then, it has been has 22,018 Twitter mentions.)
Below is a follow-up visual 30-day snapshot “History of #didnotreport hashtag” that I created on 19th April 2012, with the “Life of a Hashtag” infographic tool by Visua.ly:
Has the #ididnotreport persisted in the web sphere?
Defining an issue network:
So having read the press coverage of #ididnotreport, and seeing examples of the sobering tweets and posts across the blogosphere about rape, abuse and sexual assault, I wanted to see if weeks later, I could identify #ididnotreport as a visible issue in web sphere. More specifically, I wanted see if I could use network analysis techniques to see to see if this sombering meme was still a cross-platform issue, prevalent across Twitter, Tumblr, Pinterest and other websites.
One of the key tools to use for defining issues in the web sphere is the Digital Methods Initiative’s “Issue Crawler“, a web network location and visualization software that consists of crawlers, analysis engines and visualisation modules, that crawl specified sites and capture the outlinks from those specified sites.
I used Issue Crawler’s Co-link analysis module to crawl the seed URLs by page from the query term “#ididnotreport” through 3 iterations – this then retained the pages that received at least two links (at a crawl depth of 2) from the seeds. I then used a Cluster Map to plot my issuecrawl result as a spring map.
Click on the image below to view the full map:
The results were quite interesting, as they visualised a clear issue network of many different types of websites from .org organisations, .com newsites and blogs and official US .gov websites that made mention of or linked to websites that discussed the #ididnotreport hashtag. The issue map also reflected my later findings quite accurately, most specifically showing that more visually oriented social media platforms like Pinterest had fewer mentions of #ididnotreport in comparison to blogs and other mediums such as Facebook, Tumblr and Twitter, which are more built to allow expression in the form of words – these text-featured platforms seemed to cluster and have more mention of #ididnotreport.
You can see this most specifically in the cluster map of my issue crawl above, which shows Pinterest as an outlier (see top right), whereas Twitter, the main vehicle of the hashtag is the largest actor with the most in-degree edges. What was also interesting was that a lot of sites that campaign for and talk about womens’ rights issues were discussing #ididnotreport, such as ukfeminista.org.uk, jezebel.com, womenslaw.org, loveisrespct.org, southallblacksisters.org, unwomen.org and breakthecycle.org. There is also a cluster of US government websites (see the green .gov nodes) that mention the issue.
You can see this even more clearly in the list of Actor Rankings that Issue Crawler has outputted based on Google’s page rankings of the amount of times #ididnotreport was mentioned on the specific sites within the issue network (this was produced when I uploaded the .SVG output from my inital issue crawl into DMI’s Actor Profiler tool).
Click on the image below to view the full list:
Pinterest platform, an interesting outlier
I explored the Pinterest outlier result a little further by doing a simple search, and a Scraperwiki python using Pinterest’s search engine and the term “http://pinterest.com/search/?q=ididnotreport“. These both returned just 7 mentions:
A similar Google search with term: “ididnotreport site:pinterest.com” – returned only 1-4 results.
My thoughts behind this set of results are again, that Pinterest is more of a visual documentation platform, where people pin-up visual pictures that represent thier interests, as #ididnotreport is more of a descriptive meme, sharing and re-living accounts of abuse, harassment and assault – thus there would be a broader issue network on more text-based platforms like Twitter, Facebook, news and activists websites and latterly and possibly more prominently, in the blogosphere. I found myself want to prove this theory to be wrong, and go ahead to push this a little further in the Visual Timeline study below.
My initial Issue Crawl results got me thinking about how much the blogging platforms were talking about #ididnotreport. After looking at the APIs of various social networks, I decided to select the web sphere of the Tumblr blogging platform to discover how often #ididnotreport was mentioned, and to see if within this sphere, blogs that mentioned #ididnotreport, also discussed related issues based on key words such as “child abuse”, “bullying”, “victim”, “suicide”, “rape” and some of the other topics that seemed to appear quite frequently when browsing search results for #ididnotreport.
Google Scraper allows you to harvest the top 100 Google search returns for your chosen issue, and then input these into it’s processing tool, which then outputs a measured result in the form of “issue clouds” that you can use to analyse the prevalence of the perceived issue/key words you have scraped.
For this study of Tumblr, I scraped and harvested the top 100 Google search returns for “ididnotreport site:tumblr.com”.
The issue clouds that were outputted showed that although the term “ididnotreport” occurred a total of 206 times in the 100 tumblr blogs I scraped, these blogs spoke more about child abuse, rape, abuse, victims, suicide, harassment, and bullying than the term #ididnotreport. This may be explained by the mere fact that the term #ididnotreport has been around a lot less longer than these other terms – it might also be explained by the general content of these blogs, which may focus more on the discussion of these topics than other blogs that do not mention #ididnotreport.
Discussion and Conclusion:
Visual Timeline: Video and Image Representation of #ididnotreport ?
But despite my initial findings across with the visual scrapbooking platform Pinterest, I still wanted to find out if people were representing #ididnotreport through visual mediums to a greater degree than my initial findings had shown.
I had noticed that Storify has been used by various news outlets and within the context of this study, by the London Feminist Network to create a timeline of tweets in relation to specific issues.
Click on the image below and scroll through the window to see a large Storify collection of #ididnotreport tweets collected by the London Feminist Network:
In addition to this, a quick Google Images search and a DMI Google Image Scrape showed that there were indeed more visual representations of this:
But I wanted to go a few steps further and represent these over a period of time. Surely I could do what the Visual.ly tool had done, and map the life of hashtag, but in this case a timeline of a hashtag meme across the visual web sphere?
While investigating various forms of presenting timelines for different forms of social network data, I was fortunate enough to come across Martin Hawksey’s originating post “Experiment to dynamically timeline media posted on Twitter using Topsy and Timeline (my contribution to @Arras95) #arras95“. In this, he discussed how one could utilise data pulled from the Topsy.com social search api, into Yahoo Pipes, through to a Google Spreadsheets template as an RSS feed, and then out again as web timeline with the help of Vérité Timeline App. I found this thought-provoking, and immediately began testing it with various data sources, including the data from a specific Tumblr blog covering the “Hunger Games Racist Tweets Controversy”.
I was about to employ Hawksey’s Timeline Method here, but was still troubled by the fact that it included pulling in text references as well as visual references. However, thankfully Hawksey was thinking along the same lines, and decided to revise his Timeline approach to cover visual representations only in his post “Revisiting: Experiment to dynamically timeline media posted on Twitter using Topsy and Timeline“. Perfect. So, using Hawksey’s revised method, I created my clone of his Yahoo Pipes, changed the search terms to #ididnotreport, pulled the RSS feed into a revised Google Template, and then tried to embed the Timeline into this post. I came across the same jQuery issues as Hawksey in his original post, so decided to utilise some ubiquitous Worpress plugins to make it stick. The result is the #ididnotreport Timeline below:
My initial Yahoo Pipes scrape outputted 31 lines of data, with a combination of images, Youtube and Vimeo video that utilised the #ididnotreport hashtag. Again, this goes some way to proving my theory that the audio/visual medium is not as easy a descriptive tool to use when expressing oneself within a descriptive meme (contrast the 31 lines of data with the 2.5k lines of data harvested via the Visua.ly “Life of a Hashtag” tool).
Also, not all of the videos in the timeline were directly relevant to the theme of the hashtag – the first 3 videos from March 13th 2012 being a good example, where these seem to have been uploaded using the #ididnotreport hashtag to opportunistically benefit from the strength of the meme by using #ididnotreport to push their video higher up the Youtube/Google page rank. This is an unfortunate outcome of using dynamic feeds to pull in data, where, try as one might, one cannot always outwit human opportunists, and in this case, delete unwanted lines of code when an RSS feed is auto-updating (unless one gets into the tricky area of machine corpa training and sentiment analysis algorithms of course).
To close on a more technical note: for those wanting to know how to embed their versions of the ubiquitous Vérité Timeline into their own posts, here is how I got around the jQuery issues preventing you from doing a straight embed into your WordPress post:
I downloaded the “iframe” plugin here (and thankfully the ‘how-to’ is pretty simple: http://wordpress.org/extend/plugins/iframe/
Once you’ve installed the plugin, you just put the link to the page you want to embed in this bit of code in ‘html view’, and change the width and height if you want:
[iframe src="http://twitterabused.org/didnotreport/timeline/index2.html" width="900" height="700"]
I used a WordPress plugin called “Easy fancy-box” to link from an image to a pop-up page containing an iframe of the code I wanted to show. You can get the plugin here: http://wordpress.org/extend/plugins/easy-fancybox/
Once you have the plugin installed, in your post, upload and link to an image/screengrab of the script you want to show, and then add the following class to the first part of the “a link” code:
Then make sure you change the url in ‘href=””‘ to the url of the page you want to link to (see example below).