Year 2 Summit Meeting

The Center for Spatial and Textual Analysis at Stanford hosted our Year 2 Summit meeting at their lab on October 15, 2013.  During the all day meeting, we had in attendance:

• Jon Voss, Nick Stanhope, Max Baumgarten and Kerri Young from the Historypin team
• Zephyr Frank, Gabriel Wolfenstein, Matt Bryant, Ryan Heuser, Franco Moretti, Erik Steiner, Sophia Paliza-Carre, and Raina Sun from the Stanford team
• Justin Quimby, Abigail Phillips, and Ethan Watrall from the advisory panel

The morning was dedicated to an overview of our Year 1 reporting and looking at some of the major themes and lessons that have come from the first year of research and experimentation.

From the academic side, Zephyr reported on the importance of the community engagement efforts, which we'll touch on below, as well as the challenges of creating the opportunities for students to participate and how their ability to engage in more depth happens in the summer when they can be employed for more hours.  This underlines the importance of planning for the waves of an academic calendar.

Nick focused on the importance of looking at various types of crowdsourcing and noted that the Year of the Bay and Railroads projects rely on knowledge vs. capacity.  Because we are tapping communities for contributions and information that requires experience and domain-specific knowledge, and is very much not a yes/no or many-hands-make-light-work kind of scenario, we have particular challenges.

One of the major themes that has continually been underlined in year 1 was the critical importance of community engagement and the long lead-time that goes into getting community support and buy-in, and getting to the point where people want to contribute to a project.

Ethan brought a fresh perspective on the project coming in from Michigan State where he's the Associate Director of the Matrix Center for Digital Humanities and Social Sciences.  He was interested in digging deep into the kinds of process and organizational relationships that are necessary to make this kind of collaborative work successful. While these sorts of academic/non-profit or even commercial partnerships are more common in the sciences, they see much less of it in the humanities and is worth documenting best-practices and guidelines for at this stage.  He said:

If crowdsourcing is going to be successful, we as academics need to learn now to work in collaboration and be flexible to find the overlap of the social engagement and outcomes.

 Zephyr mirrored this and added the importance of identifying where crowdsourcing is truly useful and when it is not.  He asked how many academics could tweak their methodology enough to really gain something useful from this kind of interaction.

comparative daily visits for projects

Before breaking for lunch, we zoomed in on community engagement metrics and analysis, looking at some of the approaches that have been employed on Year of the Bay, which Kerri presented.  This included use of social media, targeting particular audiences, using the a "History Mysteries" approach, conducting live events and workshops, one-on-one pinning sessions, traditional media outlets, the role of the Curating the Bay exhibit and the pinning station, putting out questions and calls on existing social media outlets (namely Facebook and Twitter), and harnessing that info for use and recording on the Historypin platform.

Much of this input informed the creation of the prototype version of metadata crowdsourcing tools ("MDCS"), externally known as History Mysteries, or just Mysteries, currently live on the site.  The first iteration of the tool launched in September and it allowed admins to create Mysteries, which were specific questions about metadata, either a title, date, keyword, location, or Street View.  For our first iteration we have been testing with Street View Mysteries and recruiting the public to a set of about 500 images from the San Francisco Public Library collections that had been geotagged, but needed Street View locations.  Traction has not been as high as we expected, and we're working on a couple of possible hypotheses to test.  There is certainly user interaction research to do on the prototype interface, but some of the things we are testing for is narrowing the scope of the Mysteries to one neighborhood, rather than the entire city.

Prototype version of Street View live on Year of the Bay.

Related to this, some of the conversations in the afternoon yielded further input from the team as to ways to engage particular communities, attract a diverse demographic from around the Bay Area, and look at alternatives for content acquisition.

In the afternoon we heard from Ethan about the Quilt Index, which has some similarities to the Year of the Bay and Railroads projects.  The project has been running since 1999, and is now self sustaining (I believe, need to confirm with Ethan).  Some of the key points here were around working with a specific subject matter and passionate community over time, and different ways of engaging those audiences.  This fed in later to the importance of our approach with the Railroads project to share ownership with the community from the beginning, and different ways to insure that connection with the academic community so that the project could be useful or sustained after the official project duration.

Turning our attention to 500 Novels, and shifting gear from Year of the Bay and Railroads, we heard a reporting of the work that Ryan has been leading with Franco using Mechanical Turk to derive emotional ratings of phrases specific to locations from historical novels featuring London.  The team had parsed out phrases and locations from TEI markup of over 250 novels that had location-specific information about different places in London.  We then had budget to run these through Mechanical Turk to have readers rate the emotions of the phrases.  The idea here is that you might be able to say that most descriptions of say, Covent Garden in a ten year period, expressed negative emotions.  This work has enabled the team to iterate and simplify the user interface, which are feeding into the user interface design work that the Historypin team is working on for the public version.  The focus here will be on visualizing the existing data while refining and adding to it.

Conceptual mockup of the 500 Novels interface.