FurBall Maps ≠ Results or Strategy
Big Risks in Seeking Pretty Maps.
Most social networking analysis (SNA) methodology is based on surveys, data mining or a combination of both (e.g., Valdis Krebs, FAS research, and others). This type of social network analysis works well when people need to participate in research as a requirement of their job or when the data on their connections is easily available. Today, access to most topline information is readily available through email headers and footers, phone records, bankruptcy records, or public documents. For example, the Enron emails, paradise papers, amazon’s entire catalogue of book sales, Peabody energy or citation and co-publication data are all just a few Google searches away.
However, in a distributed advocacy network or political context, social networking analysis doesn’t always work well. This is for two main reasons:
- Accurate data is very hard to get and too easy to “game.”
- The outputs of a social networking analysis do not always translate to an advocacy context.
Advocacy Data is Very Hard to Get and Easy to “Game”
Even in a simple advocacy context, you may be focusing in on a network of 300 to 500 people–I have seen bad attempts to go larger. For example, the network map our team at Netcentric Campaigns created in 2008 displayed a network of people working on US foreign policy and multilateral engagement efforts. Building the map was expensive in terms of staff investment and huge tax in terms of volunteer hours to complete. It required significant costs and hours from our team, consultants and the use of SNA survey software. We developed a strategy based on these maps.
We know the most connected political people are typically central to these structures. (see above left) They are also the ones with the busiest schedules and the least motivation to fill out a survey accurately. As a result, the central hubs don’t appear on the final map, and the network appears weak or decentralized. Additionally, smaller players will be given too much weight because the central hubs aren’t on the map. Notice how ideas around the breaks in the network or the distance between nodes are wrong. This also leads to the wrong interventions (like trying to build more centrality when it is not needed). User surveys that don’t work for decentralized networks result in bad data.
In the absence of user surveys, and without stolen Facebook data on 50 million users, we are forced to look at data scraping and data crunching, which can create mixed results. Who sends email to whom? Who calls whom? Who meets with whom? Who is on each other’s Board, advisory group, etc.? Which members attended the same conference? In a decentralized system (unlike Fortune 100 companies) we will not be able to construct the density of peer to peer interactions given that these separate groups are unlikely to share all their call logs, email traffic, all the Facebook data, etc. The closest you might get is who worked on the same campaign (but that is not enough to establish SNA). We do not work for the same company or use a universal system to manage our peer to peer chatter so it’s not an effective approach to use data scraping in an advocacy context.
Additionally, in a political/advocacy context, people have an incentive to distort the data in many ways. Personal, social and political connections are all their own form of capital. Every organizer has an incentive to both inflate connections with others and also deny ties that reach into “enemy camps.”
It is not recommended to use of survey based SNA tools or mapping in an advocacy or social change context.
Advocacy Activities Don’t Map Well to Social Network Analysis Outputs
The second factor that makes SNA less effective in an advocacy context is that the maps by design show generic social relationships. These social relationships aren’t always capable of transferring to advocacy. Social engagement and civic engagement are different beasts.
There are people who want to work on climate change that I will never like, nor would I ever want to sit down at the dinner table with them. Conversely, I have family members that I love and would donate blood for, but we do not see eye-to-eye on a single social issue.
If you invest in social networks you get social outcomes. Some social outcomes can be leveraged for advocacy. However, it is totally appropriate for people who have no social ties with each other or any relationship with each other to be able to work together on a social agenda. In the advocacy and social change context we work on flows in a network. How does traffic really move among the lines and create change?
Companies love SNA because it helps them target interventions. The CIA loves SNA for the same reason. By mapping who is speaking to whom you can target them more precisely. However, in an advocacy context, we have very blunt instruments to reshape networks (e.g., grants, conferences, training, social programs designed to reach groups of grantees, etc). These interventions may well only need a simple smoke alarm to sense where there’s trouble, not the detailed parts per million gas spectrometer testing of connections found in a social network analysis.
Unfortunately, the challenges with SNA in an advocacy context don’t end there. Other problems in voluntary association networks include the horrible first impressions that data request like this can make on users, security issues (putting your contacts into a database that is owned by others), exposing threads of connections among people to freedom of information requests (FOIA) which expose networks whenever government employees are involved, survey issues of inclusion, language issues, and the big issues that arise as maps reflect accurately or inaccurately, issues of bias and/or marginalization based on race, gender, sexual identity or religion.
Those administering and presenting a network to the participants in a social movement don’t always have the skills or context to deal with maps that are so personal, this is not an SNA being used as a tool for large HR departments.
Even in the best of circumstances, these SNAs are snapshots of organic living complex systems. Networks of people constantly change, weak links unreported in a moment can be activated and strong in the blink of an eye. Data that looks at the network in a moment, must have several sequences over time to observe any stability of changing trends which doubles and triples all the costs on volunteers and others alike.
Clearly, SNA and SNA mapping is an important and powerful tool within a firm, when you have 50 million stolen Facebook records, and any data-rich environment that results in effective single node interventions. However, for the reasons above –particularly the high costs of data collection, the huge impact of missing important nodes, and the nature of the limited means to intervene at the individual node– it is not recommended to use of survey-based SNA tools or mapping in an advocacy or social change context.