Wikistrat Investigation Summary
Having had some success in domestic policy decision making, with Progressive Congress News being the final result, I thought I would see if there was anything that needed doing in the realm of foreign policy. Wikistrat, [allegedly] the world’s first Massively Multiplayer Online Consultancy (MMOC), was something that was immediately visible once I graphed my personal contacts. I wrote six posts about them as I mapped their network.
Foreign Policy Process – I graphed my contacts in the foreign policy field, I found a bunch of the top organizations and subscribed to their news feeds, and then I noticed the Wikistrat group.
Foreign Policy Organizations & Individuals – two of Wikistrat’s 156 experts were LinkedIn contacts for me. I explored the subset of members who had Twitter accounts and speculated as to what additional information could be learned about them with just social media as a starting point.
Exploring Wikistrat With Maltego – Starting with the Twitter accounts of the roughly two dozen Wikistrat members, I extracted the information from current discussions of one of the busier members, hunting for signs of issue focused communities of which the Wikistrat analysts are members. I didn’t make any great discovery, this is just an exposition on the process I used.
Wikistrat’s Analysts & Friends – I extracted the list of well connected contacts for the identifiable analyst Twitter accounts. A small connected network was revealed, but it broke down as soon as I removed the organizational role accounts that were found. This fits my expectation – Wikistrat analysts have rich interactions, but they didn’t self-organize with Twitter as a base and it seems likely they don’t participate in public theater in support of their conclusions.
Wikistrat Full Network As Of 3/30/2013 – I finally had a full Wikistrat map – the names of every member and their associated profile on the company web site. Some had LinkedIn or Twitter accounts, with the professional network being found for 40% of members and Twitter accounts being located for 20%. Overlap of LinkedIn and Twitter accounts was rare – only 2% – 3% show this pattern. I applied Named Entity Recognition to the profiles on both Wikistrat and LinkedIn. I thought I might be able to identify geographic clusters, employment clusters, or education clusters. The Wikistrat profiles are very regular in their layout, but quite resistant to the efforts of the Alchemy and OpenCalais NER products. A hand coded script with a little regex could be applied to the Wikistrat profiles, but I have not continued down that path.
Hashtags & Humans – I retrieved the most recent dozen tweets for all of the Wikistrat analysts, then extracted the hashtags they were using. I found that there were some hashtags that were congregation points, but that it was more common for there to be clusters of related tags.
Maltego provides a slider that allows four different volumes of information to be returned from a transform, their term for a query. The settings are 12, 50, 255, or 10,000. Twitter related transforms often stop at 100 entities, a limit enforced by Maltego publisher Paterva’s servers. Named Entity Recognition services are tuned for actual language and don’t perform terribly well on bodies of text with specific formats, nor were they all that useful in terms of picking out entities from tweets. Once tweets were available, hashtag extraction produced useful information, but there are performance constraints here as well.
Technical performance considerations aside, this process did reveal useful information, and some old wisdom from noted social network analyst Yoga Berra are still quite applicable today:
You can see a lot just by observing.
Phi Beta Iota: [allegedly] added above. The above comments in English: small group, isolated, not leveraging the larger online world, and not particularly well-connected to the larger world brain either. In no way, shape, or form is Wikistats multi-anything. Twitter has enormous potential (as does Amazon and Crisis Mapping) but neither are rising to that potential. Twitter right now is useful in tracking broadcast channels, monitoring chatter & debate, identifying and recruiting human sources, and in targeting for other exploitation. It is not useful as a brain trust or as a source of aggregated alerts, insights, or validated conclusions. Amazon has the potential to organize authors, reviewers, and readers by topic, time, and place. Crisis Mapping is now ready to move beyond disasters and focus on everything everywhere, especially true cost and actual policy and budget outlays at the Congressional district level.
See Also:
Graphic: Twitter as an Intelligence Tool
Yoda: Real-Time Crowd-Sourcing + Twitter Meta-RECAP
and found in passing….
Howard Rheingold: News Filters for the Future – Technical Services or Human Networks?
Howard Rheingold: Open Source Intelligence Meets Real-Time News and Data Curation – SwiftRiver
Patrick Meier: Automatically Extracting Disaster-Relevant Information from Social Media
Patrick Meier: Crisis Mapping Syria – Automated Data Mining and Crowdsourced Human Intelligence
Patrick Meier: Innovation and the State of the Humanitarian System + RECAP
Patrick Meier: Mobile Technologies, Crisis Mapping, & Disaster Response
