On January 18th I published The Indivisible Movement, which contained a visualization of the role accounts for the movement as mentioned by the four founders and the lead role account, @IndivisibleTeam.
On January 19th I published The Indivisible Movement’s Conversation, in which I explored the five core accounts and 173 additional role accounts that mentioned over 8,500 others.
Mentions are an excellent way to see who matters in the moment; keys accounts can have enormous numbers of friends and followers, but there is a limit to how much conversation happens in a given period. The Netwar System permits me to extract mentions in a given timeframe; fine grained control in which accounts are involved in a specific event.
I keep Maltego Classic, a penetration tester’s toolkit, which is useful for plumbing follower/following relationships, but it becomes clumsy when starting with more than a few dozen accounts. I have yet to lay hands on Maltego XL, which will handle more than the 10,000 node maximum of the version I have, but I am not particularly tempted. I have some skill with Gephi, a free data visualization package that will easily handle tens of thousands of nodes and which offers a wide variety of additional capabilities via its marketplace.
Today I undertook a type of analysis using Gephi which I’ve previously only done with Maltego – examining friendships in search of high centrality node that are otherwise innocuous. The numbers involved were an order of magnitude larger than anything I’d trust Maltego to handle without thrashing helplessly due to memory exhaustion.
My initial working set was only 1,738 accounts, 176 I am tracking as part of the Indivisible Movement, and 1,562 they were following. There were an amazing 25,230 links from the 176 accounts to the total set of 1,738. This is a very dense network.
This is the same network as above, only with the ‘ego network’ of @IndivisibleTeam highlighted so that we can more easily read the names of the nodes. The larger the text, the more links to a given node.
03-5-171-127-16575-min-degree-23 I used the filter methods until the only accounts that were visible had at least twenty three links. The 176 Indivisible Movement accounts were following 127 others, for a grand total of 16,575 links. A graph of 300 total nodes with sparse links is at the outer edge of my ability to lay them out in pleasing fashion. A graph this dense was basically impossible to simplify. I needed less data.
I manually removed a large number of the high degree nodes, Senators, news outlets, and the like. They are important, of course, but not in the context of attempting to understand an emerging social movement.
Growing weary of filtering and layout algorithms, I simply copied all of the high degree nodes to a new workspace in Gephi, and then deleted all of the 176 that I’ve identified as being founders and the role accounts for the moment. These remaining 58 accounts are, I believe, mostly humans that matter to the movement. Some are minor celebrities in their own right, some will prove to the the proprietors of various role accounts for the movement, and some are quiet, but pivotal players.
The first thing that jumps out at me is the keyword ‘resist’ and in particular @ResistanceParty. I suspect that group is another facet of the same social movement. I found about 4,000 accounts with the hashtag #TheResistance in their profiles using FollowerWonk.
A pool of accounts twenty times the size of those I’ve made out as being Indivisible Movement would require between two and a half and five days to collect in a serial fashion. I’ll have to rummage around in my sock drawer and arrange for parallel processing.
I see many parallels between the Indivisible Movement and other social movements in which I have engaged or studied in the eight years since I became interested in social network analysis. There are echoes of both Coffee Party and Occupy Wall Street, as well as other more subtle signs of other prior populist outbursts.