Wednesday, December 17, 2008

Son of Small Worlds

Small world networks Last week I posted a brief note saying that the Small Worlds experiment had failed, which generated a surprising number of comments. In addition to a debate about the difference between an experiment and a hypothesis, numerous people have asked me to discuss in more detail what happened with Small Worlds and what I plan to do about it.

What was Small Worlds?

Small Worlds was my first attempt to leverage the power of social networks for research scientists. It was also a reaction against the construction of purposed social networks which inevitably turn into inward-looking ghettos. The "big idea" behind Small Worlds was to foster and expand on existing network memberships in order to create small world networks, which rely on multiple tightly connected groups which are loosely connected to each other, are the most efficient design to maximize the transmission of information across a minimal number of connections. In the sciences, these subnets correspond to existing natural groupings which have the most in common, e.g. individual laboratories and academic departments. By linking these to groups with related interests, e.g. biologists with chemists, geologists with physicists, we attempted maximize the online interactions and the potential value of the groupings. The idea was always that Small Worlds would have the smallest possible footprint - no membership, no walled garden, no API.

Why did Small Worlds fail?

The failure of Small Worlds was inherent in the design and timing of the project. The intention was not to create a new social network as this is a doomed path - even though may be a source of funding at the present time. I've seen so much money wasted by JISC in the last year through funding of doomed ghettos. In some cases, these expensive handcrafted sites didn't even work properly, let alone have a chance of attracting a sustainable user base. However, in order to facilitate the uptake of connectivity we needed some sort of hub so that people could find each other and set up connections. Thus the Small Worlds site was born, and this also became a place to host training materials for remote participants. We considered a range of options for the Small Worlds hub, including various wikis, Friendfeed, Ning and a number of other solutions, but in the end we settled on WetPaint.

From the outset, this spelt trouble, and in our early face to face training and promotion sessions, we could see how eager people were to use the social functions built into WetPaint, treating the hub as a destination rather than a starting point for exploration (No Zombie Vampire Superpokes Please, We're Scientists). I agonized over whether to move the hub to a new service but eventually decided to try to fix the problem in situ (How to build an antisocial network). This attempt to fix the plumbing on the sinking ship inevitably failed and traffic on the Small World site dwindled as people arrived at the site and stuck like flies to fly paper. The one bright spot has been an emerging Twitter network which has attracted scientists from the UK, USA and Canada and is a potentially useful resource for future development, but is not in itself a sufficiently structured tool to form a teaching conduit.

How do you plan to fix it?

Small Worlds is not repairable as the failure of the project was inherent in its design. I'll say it again: purposed social networks are a dead end (even if you can get funding to build them). It is possible to argue that Facebook was originally a purposed social network, but the large scale success of Facebook did not start until it outgrew its origins at Harvard and became a global system. Zuckerberg was an infinite monkey, but that was then and this is now. Going head to head with Facebook or Google is heading in the wrong direction. The difference between arXiv, which has been very successful at changing the culture in math and physics, and the Nature Network, which has had little impact in life sciences despite considerable funding from NPG, tells us that much of this is simply a turf battle. arXiv was the primogenitor of purposed social networks, a novel concept which flourished. Nature Network is a me-too Facebook wannabe. It's not about funding, it's about social spaces already being claimed. Erecting new social spaces in competition with established trusted brands is doomed to failure.

Freytag Pyramid A few weeks ago I stumbled across the Freytag Pyramid, buried in Web 2.0 Storytelling: Emergence of a New Genre, which I found that via a tweet from Alan Levine. Being a mere scientist - not too familiar with Aristotle's Poetics - this was a new one on me. As I explored the pyramid metaphor, the thought that came into my head was about the contrast between traditional, didactic courses and the more post-modern, deconstructed efforts I was struggling to pull together at the moment - in particular, the differences between our "loosely-coupled" undergraduate PLE course and Small Worlds. The PLE course has a timetable, an episodic arrangement of assessments, and the denouement of a final mark - Pass or fail. Small Worlds had none of these. The anticipation was there, but the climax was missing, and that feels ... unsatisfying. Small Worlds died because it didn't have a narrative.

How do/should we put the narrative in a Web 2.0 community? How do we make it sticky? The solution would be to run Small Worlds as a course, perhaps something like a mini-Connectivism and Connective Knowledge Online Course or a laboratory scientist-focused Social Media Classroom. So how do you build such a course? We've learned much in running our undergraduate "purposed Web 2.0" course in the last three months, which has emphasized the investment of time and effort required that participants need to invest in order to start to reap the rewards of network membership. To achieve this commitment, the course needs to be structured to encourage and reward investment:
  • A pre-determined timetable generates community cohesion and anticipation.
  • Video mini-"lectures" provide richness and a feeling of personalization.
  • Most importantly of all, assessment provides the narrative drive that propels participants to the spaces they should occupy online. However, managing assessment for this sort of enterprise is not a straightforward task, as has been much discussed in the wake of the recent Connectivism and Connective Knowledge Online Course. My solution is that after an initial handshake with the course organizers, assessment is peer-driven, for example community-building exercises on social sites most relevant to the participants. For life scientists this might well be on Nature Network or any similar site, in addition to the standard Web 2.0 tools such as Google Reader, Twitter and delicious. For other disciplines, alternative communities can be substituted.
  • To reduce dropout, participants will enter into an explicit contract agreeing that their course will be terminated if they miss the deadline for three successive assessment tasks.
  • Final "examination": the climax, a reflective post on a community site considering how the participant's professional online identity is represented and served. This will be followed by "graduation" with some sort of certification of successful completion.
The exact syllabus will need to be decided, but will be determined by the length of the course. We have a lot of feedback from our undergraduate course in this respect, which strongly suggests a six week duration. Our data shows that four weeks is too short to generate the recursive patterns of activity needed to embed the value of investment in online networks, and ten weeks is too repetitive and too big a commitment for most.

This is a long way from how I originally envisaged Small Worlds, but without such scaffolding, the concept seems to have little value. The narrative provided by the course structure and anticipation of the final outcome is the eduglu required to bring the concept to life. What are the nuts and bolts? To be decided. I favour concentrating on building and testing the syllabus which can then be delivered by multiple routes. An "Instructor's Manual" will accompany a Drupal module, Blackboard cartridge and Moodle equivalent to allow the course to be distributed and run locally where needed. Not Invented Here syndrome being as prominent in HE as it is, local instructors need to feel ownership of the process as much as the participants do. This could be flavoured for different academic disciplines, e.g. life sciences, humanities, social science, etc. This would requires a developer with the necessary technical knowledge, and that requires funding.

Which is my next task.