Prototyping Weeknotes #58 (15/04/11)
A quote from Kat to start this week's weeknotes:
There's always a tricky line to tread with prototypes: what's worth building and what's not? How forgiving should your testers be? What's a sufficient proof of concept? How high fidelity should it be? Aim it too high and you risk over-determining the concept; aim it too low, and you risk obfuscating its purpose.
These questions have been asked by all of our different projects this week.
Sean and I started the week building a mobile application using the jquery mobile framework to allow tags to be made to our RadioTAG service in a simple way. We return metadata when a tag is made, and even added some pretty pictures!
On Wednesday, Joanne, Vicky, Kat and I sat down to discuss the research goals for the RadioTAG focus-group work and decided that the mobile application was too dissimilar to "real" tagging to be of much use. It's something we'll visit at a later date. A very useful discussion in the end though which resulted in a pretty clear plan of action for this portion of the RadioTAG work. We close out Friday afternoon with a deployment of our prototype RadioTAG service using a combination of Passenger and nginx.
Meanwhile, Tristan and the Watch Later team sat down with George and Chris G to review the Watch Later work so far. While broadly positive, there are some questions about the longer term aims of the project. The whole team now have accounts on Duncan's client and are starting to use the app in anger. Duncan made the app more powerful by adding Twitter authentication to the client using the aforementioned and fabulous OmniAuth gem, and switched the Solr code from solr-ruby to RSolr. As the API grows more complex, organisation of the code is becoming difficult. Duncan bit the bullet and ported the code over to Rails 3 whose conventions for database migrations and routes help a lot on larger projects.
Most of the team attended the final debrief on the Mythology Engine research on Tuesday afternoon. Tristan concludes that "it's a good piece of work with lots to consider, but was quite a difficult brief". On the one hand, we presented users with a polished prototype based around Doctor Who while on the other we also wanted to learn about drama and story-telling on the web in general.
A very informative presentation of the user research indicated that the data model we used in the Mythology Engine seems to match peoples' mental models of stories and that people are also quite good at sketching what happens in stories in terms of characters, chronologies and events. But our prototype exposed too many "things", and exposed them all at the same level without a clear indication of priority or importance to the story. In the end, the Doctor Who team seemed interested in the findings and we hope that the exercise has been a useful one.
Kat, Tristan and Olivier with help from colleagues across the BBC have continued thinking about user-owned data and identity this week. They are struggling with a complicated subject, while trying to keep in mind the implications to the BBC while not getting "bogged down in existentialism". Ev Williams, Twitter's founder, recently talked about the five pieces of online identity, an article that has been widely read in the office this week. We've been following reactions to the post with interest, while Tristan has got stuck in to the excellent W3C Social Web Incubator Group's final report, just one piece of a huge body of literature on the subject.
I'm getting the feeling that the light is starting to shine at the end of a long tunnel for Paul this week. Work on the charting capabilities of the Dashboard, and general cleansing of real-world data as it flows into the system culminated in a great demo of the project to the team on Thursday. There's still a seemingly endless list of edge cases to deal with though.