• March 2012
    M T W T F S S
    « Feb   Apr »
  • Pages

  • Marginalia

  • Accumulations

Crawl Space | Notes on VIsual Interface

Can automated exploration of networked entities be used to construct contextualized interactive information interfaces? It’s a question I spend a lot of time tinkering and pondering over.

If I were to use WIRE or Navicrawler to perform a focused crawl on, let us say, journalism education programs, or the recent industrywide PR campaign in favor of the private equity funds sector, can I automate the results to obtain useful metadata?

In many cases, just obtaining a list of sector players linked from a «newspaper kiosk» — a site with an extensive and systematic outdegree — is quite helpful. You can


the resulting file and then consult the subsequent visualization for information on its context.

A simple answer, then, is, yes, so long as the networked individuals — whose consent is demanded by Rebecca MacKinnon in a recent book — are engaging in deliberate search engine optimization, employing marketing techniques that are actually quite simple to understand and implement, and heuristically comprehensible, even if it is not year clear whether brokerage analysis can provide more precise analysis.

The fact is that our present-day blogging industrial complex is not that different in principle than the Victorian interlocking directorate and the Hearstian media zaibatsu — the internaut wears executive offices, board chairs, fellowships, grants and like a general wears campaign ribbons on his starched blouse.

For that reason we look to construct 2-mode networks showing the correlations between actors and multiple agencies, like the Scotland.net example described in the Pajek textbook:

Sometimes, the elective affinities of persons function as loci of informal institutional cooperation.

As for networked identities, I have to admit, my own blogs make use of the same type of strategy: running a WordPress «Blog This» on my English-language blogs — past and present — cross-references to my Portuguese language blog, and vice-versa, using the automatic ping feature.

A Robot For SourceWatch

In any case, admirable and necessary projects like SourceWatch’s attempt to keep up with these musical chairs, tend to suffer from «earth to wiki syndrome» — volunteers just do not have time to keep abreast of this garden of forking paths.

Here, I have produced an attempt to understand the institutional synergies of the Brazil Foundation.

The joint U.S.-Brazilian NGO shares k-cores, I think they call it, with official State Department content. Talkpoint is a communications company with DNA from CNBC and clientele that includes Intel, Morgan Stanley, Sanofi, Bayer and Cisco  …

The foundation’s Web presence can be analyzed into mainstream media outlets, at the core, and a blogosphere, on the margins. What seems like an official blog has not been updated for a year now.

Back to the project and its goal, meanwhile.

The difficulty I have is with producing interfaces useful to someone other than myself. CMapTools and yED provide drawing and rendering tools for network data that can be elaborated upon with other types of information, but are often not powerful enough to render, say, multilevel relationships.

It is easier in Pajek to filter out what I call the Social Code Effect, which sometimes gets us a better idea of a given interentity’s institutional backing and synergies, for example.

However, between the value of the analysis as an exercise in guided navigation and presentation in PowerPoint there is some progress to be made.  The example shown above is from my Portuguese-language blog —

I added real-world information to supplement what network analysis suggested might be relevant ties, in this case, the activism of the publisher of the RBS Group in the international «future of the newspaper» movement.

In the following diagram, I processed the k-neighbors of a World Association of Newspapers Web site and found what I expected to find from my own offiline knowledge — Rebecca MacKinnon is a prominent figure who from the early days of BloggerCon and has worn many hats for many masters, after abandoning a career at CNN to midwife the Harvard-Stanford blogging-industrial complex.

Synergies among what may appear to be unrelated projects at first glance link a world-class consultancy, the world newspaper movement, and the geniuses at Harvard Law.

Rebecca now adds to her CV a post at the New America Foundation and a promotional Web site for her freshly launched book on social media and democracy. Funders of the foundation include

  1. The Ford Foundation
  2. The New York Community Trust
  3. The Rockefeller Foundation
  4. Blue Shield of California Foundation
  5. Broadmap, LLC
  6. Carnegie Corporation of New York
  7. Foundation to Promote Open Society
  8. The Bill & Melinda Gates Foundation
  9. The William and Flora Hewlett Foundation
  10. The James Irvine Foundation
  11. Peter G. Peterson Foundation
  12. The Pew Charitable Trusts
  13. Wendy and Eric Schmidt
  14. Bernard and Irene Schwartz

Keep tracking funders and you start finding feedback loops. The Global Voices Online project is the locus of various intersections between Church and State — that is, between public sector «public diplomacy» and private-sector lobbying in emerging markets. Or so I would argue.

As to producing the usable visual interface with freely available tools, it is a shame that there is no way to import XML graph files into CMapTools, which offers better interactivity, such as local and Internet hyperlinks, mouseover contextual comments, and so on and forth.

CMaps offers space for sharing «ontologies» and related data structures, which I should take more advantage of … provided I could find the time.

Here we hear of a board member’s Harvard ties — a tie shared with other board members.

A project I keep an eye on for Brazilian colleagues is the Knight chair in Latin American journalism at UT Austin …

The network is extraordinarily complex, but can be reduced to overall patterns whose repetitive quality goes unappreciated.

WAN-IFRA k-cores  …