Leadership in online influence scoring is becoming embattled territory as services in this area include Klout, Kred, Tweet Grader (by Hubspot), and Tweet Level. But one should note, all are driven by black box algorithms that vye for the title of industry  “gold standard” for anointing online influencers. This isn’t pure vanity either; knowing the level of influence of any player is a PR and marketing professional’s dream solution but nobody to date has provided the silver bullet.

This seminal article on “weblogs” by Clay Shirky (February 2003) remains a classic in the social media cannon and is a favorite around the Nexalogy office. Power Law or ‘Pareto’ distribution, describes the observable social phenomena whereby the majority of content is created by the minority of producers. It’s an unavoidable response to the plethora of choice. Network theory shows how one’s individual choice to read a blog reverberates through their network based even though such choices are sometimes completely subjective – like preferring one style of writing over another or feeling politically aligned with like-minded sources.

The main thing to remember when considering influence graders is that they are at this point blunt instruments at best. They’ve grown from being impression measurement tools trying to measure context and subjectivity, as described above. I haven’t been astounded by the results of Kred so far but I have to say that I’m most optimistic about their methodology because they been quick to slice influence into domains. With Kred, at least social capital is seen as context dependent.

Tweet Grader (the free version, anyway) has long been outstripped especially in the insights department. Surely they can offer me something better than six words in a tweet cloud featuring “RT” as my most used word? Please.

One has to ask, is it a good thing that the most influence is wielded by a small group of people? Social media content creators that reach a certain momentum in twitter followers and blog links become the a part of that small bracket thanks to search engine ranking algorithms and twitter recommendation engines. The impossibility of knowing the blogosphere is exciting the way moving from a small town to a big city is exciting. Pseudonymity allows bloggers more wiggle room to express themselves, opinions are as diverse as they are plentiful and online social mobility becomes possible.

If there is one thing that the Occupy and Arab Spring movements have shown us in the last few weeks and months is that power imbalance has consequences and social media still has disruptive potential  Influence graders are bell weathers of power. I argue that a consequence of grading influence and doing it well will confer merit, power and its multiplication effects for those who actually deserve it.

Will these companies step up to the work for the interests of the 99%?

Special Thanks to Zach Devereaux (Chief Analyst Nexalogy Environics) for his valuable feedback on this post.

A shorter version of this blog post can be found on the Toronto Board of Trade website (link)

In one of the more important turns of the campaign David Suzuki endorsed Dalton McGuinty and the OLP. Green Energy is a top issue associated significantly with the Liberals; on Twitter over the Sept. 17th weekend their environmental record was significantly discussed. The Green Energy discussion also links to the major campaign theme to date of Jobs, as jobs are discussed in terms of environmentally friendly job creation. This shows that the previous weekend’s trending of green jobs really got traction in the week of September 12-16th.

For Sunday Sept. 18th, “Vote” was the top word, associated in particular with comparisons of the voting record of Andrea Horwath to Tim Hudak in the Ontario Provincial Parliament. Many Tweeters retweeted the message that Horwath and Hudak voted the same way more than 150 times despite their ideological differences. “Job” continued to be strong, with Tim Hudak’s announcement of a five-point job plan getting some traction on Twitter showing that the progressive conservatives know how to react to the biggest topic of the election campaign so far.

Health care began to trend for the first time on Sunday the 18th in the Twitterverse as a campaign issue with Dalton McGuinty visiting the sick kids hospital to emphasize the OLP’s record on health care garnering a number of tweets. Multiple tweets critiqued the PCPO and Tim Hudak for dismantling hospitals in the past and lacking plans to fund hospitals in the future. In a reflection of how the OLP is engaging well with the Twitterverse, a number of tweets associated the Party with an elevated level of trust compared to the other parties when it comes to guiding the Ontario health care system.

Overall Timeline

Timeline for the weekend of September 17th and 18th

Top Words reflect Green Jobs, and the David Suzuki endorsement of the OLP

The top publishers e.g. those Twitter accounts that produced the largest number of tweets are as follows:

However, the most important accounts based on interaction, our “Top Tweeters” (mentions and replies, but not retweets) are as follows and show that Horwath garnered the most discussion over the weekend, followed by Hudak. McGuinty was not discussed as much this weekend by name.

Top Shared Links

1. Similar to the previous week, an attack site against Horwath tops the shared links
2. The official Dalton McGuinty OLP Facebook page, with multiple videos
3. An OLP ‘rogue’ website comparing the voting record of Horwath and Hudak
4. The official Andrea Horwath NDP website for NDP videos
5. A video that has now been removed from YouTube by the original poster, which was the endorsement of Dalton McGuinty by David Suzuki

Implementing a new system can be costly in subscription fees alone; some enterprise SMA solutions will set you back over $100,000 per month! I argue they’re vastly more damaging to the organization in an intangible way: it’s a huge blow to morale to adopt and abandon if the tool is simply not right for your needs.

We thought it worthwhile to amalgamate the collected wisdom from these critical minds that have tried more solutions collectively than any SMA supplier could ever possibly do on their own. I’ve got a few notes on things to look for and things to avoid:

1. Make sure what you’re getting from your solution is actionable. Infographics are trendy but it’s strategy you want. Charts are useless if they don’t give you a clue as to what to do next.
2. Make sure your solution captures networks of information and not just reporting on social-networks-as-silos. The real insights are in network effects and amplifications, not in short, non-linear relationships.
3. Do for data what hi-def did for TV. This means Natural Language Processing and sentiment analysis that are better than “positive, negative and neutral” and pull the qualitative gems from the quagmire. This goes for reporting on metadata that can tell you the who, what, where and when to give you a sense of the diffusion and discourse.
4. Filter spam. Don’t just say you do it; actually do it. Paying through the nose for a SMA with poor spam filtering is like going to Le Cirque and being expected to bus your own table.
5. Creating linkages between networks is what you pay the big bucks for and you want something that uses “sleuthy techniques”, algortihmic-ninja fancy footwork to open up that pandora’s box of the non-obvious.

Before going into any of the above however, I will recall two beautiful latin words “caveat emptor”, which means “don’t be a sucker” in plain english. Writing on Microsoft Word, WordPerfect 5.1 or on parchment has no real bearing on the quality of the prose. Ergo, an SMA tool will never define your problem for you. Dedicate some quality time to your business objectives, how social media impacts them and how you want to measure that.

Resources:

Nexalogy White Paper on Social Media Intelligence vs. Monitoring

The other keynote address was given by John Kelly from Morningside Analytics, whose approach highlights the importance of segmentation in the blogosphere when it comes to telling the difference between ‘fake sharks’ and ‘real sharks’ in the waters of partisan blogosphere activity. John’s presentation was very interesting and he was nice enough to compliment Nexalogy’s approach as being a “very advanced semantic analysis capacity” that compliments link-driven analysis, which is Morningside’s strength.

The two presentations from the National Research Council were a highlight for me, with Dr. Mohammad Saif presenting the current state of NRC-developed sentiment mapping and Dr. Roland Kuhn presenting the potential for sentiment mapping and machine-translation to be applied to national security priorities. I am hoping to learn more from them about the subject in the future.

The day before I presented at the Public Security S&T Summer Symposium where the focus of discussion was on identifying social media signals from emergency situations; a challenge to be sure but one where a social media analysis solution can be a big benefit. The whole experience was positive, and the discussion definitely emphasized the importance of social media activity in terms of big data and public security.

A few selected areas of work for participants from government that were identified and that I am happy to have on my radar included:

• Knowledge governance
• Human capital utilization
• Empowering the public servant
• Government 2.0, 3.0 and beyond
• Forecasting, forewarning, foresight
• Participatory culture
• Multi-lingual information processing

If you want to download my presentation : click here to download (PDF)

Before I left on postdoc I ran across a job application that was looking for particle physicists or astrophysicists to help improve pedestrian traffic in Monaco.  It made sense.  People individually are very hard to predict, (duh) but thousands of people bunched together with strict constraints on where they can move , wouldn’t be that difficult to model.  Lots and lots of people =~ fluid dynamics.  A very viscous fluid but a fluid nonetheless. I had scored a postdoc fellowship and I was already commited to NYC or Paris.  But the Monaco job did peak my interest… It was a premonition of my future work here at Nexalogy.

One of the sources of “big data” is the never-ending stream of material emanating from the social web. People are taking many approaches to analyzing this kind of data, but at the end of the day, the goal of any approach is the same: to gain a better understanding of the whole stream.

To demonstrate how we approach this problem here at Nexalogy, while I was gone I had my team in Montreal gather every tweet that contained the main hashtags used at Strata (#strataconf & #stratconf). There were 4616 tweets between February 1 to February 7 from a total of 1455 individual Twitter users.

We have built an interactive lexical map of the entire dataset – go to http://strataconf.nexalogy.com to view and interact with the data.

There were some great moments during the day, which as a whole exceeded expectations. Some highlights included presentations by Pierre-Loïc Assayag (from Traackr), and a wonderful keynote by KD Paine. The panel led by Mark Evans (from Sysomos), featuring Darren Barefoot, Brian Cugelman (from AlterSpark), and Ilya Grigorik (from Postrank) was particularly excellent, and really moved the whole day’s program in a very productive direction.

Claude’s presentation during his panel was pretty well received – thanks to everyone for the warm reception. I think video may be available soon, but for now, you can download Claude’s slide deck here: Nexalogy-TTMM-Deck (2.1Mb PDF).

He mentions both cluster analysis and semantic analysis as avenues for future development and accurately describes some of the challenges that go along with these approaches. These techniques are at the heart of Nexalogy’s unique approach to Social Media Intelligence – and are a key element of the value we offer our clients to use social media to do much more than simple monitoring.

For me, this post is another indication that what we’re living through is not the birth of one industry but two: SM Monitoring and SM Intelligence. They’re different jobs with different goals, and the things required to do monitoring well (and that the key monitoring companies have done a good job deploying) are not necessarily the things required to produce valuable intelligence that can feed directly into brand strategy or corporate risk assessment scenarios.

Key quote: a quote from Chris’ interview with Claude:

“This is where physics comes into play,” [Théoret] explains, “Physicists know how to handle literally astronomical sizes of data sets that aren’t very well organized. With blogs for example, the data problem is the same. We need to find signal and reduce noise. Because the data is unstructured, analysis becomes very complicated – but that’s where Nexalogy comes in.”