BBC R&D

Posted by Tristan Ferne on , last updated

This week in Prototyping...

The News linking team (Olivier, Chris L, Duncan and Joanne) are in data analysis mode. Earlier in the week Olivier was wading through screenfuls of SQL consoles and python scripts but managing to find a few interesting trends in the logs: some did not quite meet Chris' rigorous sense of statistical analysis, but others gave us a number of insights worth pursuing or validating - Olivier says "What's interesting here is that we are not only validating a few things we already suspected about how people navigate between news stories and news media, we are also refining a few metrics for links efficacy and popularity". Joanne unearthed a treasure trove of research papers on the topic. Any respectable paper MUST, apparently, include complicated-looking mathematical formulas which could more easily be explained in prose. And Chris and Duncan are now scraping pages for the things that aren't in the logs and looking at the correlations between visits, page length and click-throughs.

The news companion team (Vicky, Kat, Theo, Sean, Chris N) are in the research and idea development phase, moving on from our original idea and working closely with News to widen the scope and come up with a few possible directions. Nina from the News design team is spending a couple of days a week sitting with us. They've been diving into the relevant user research and are looking at personalisation, bubbles, dual screens, saving for later, serendipity and the completability of news.

We’re also wrapping up the previous phases of projects. We’ve been testing the Programme List (nee Watch Later) among the team - we need to decide if it’s good enough to go further or if there are key features which must be added. Theo, Joanne and Kat are working up some interaction design principles for RadioTAG, which is due to be trialled in the homes of a few research participants soon, and Sean put a few final touches on the spec. The prototype spans a physical device (a radio) and a website, which is pretty challenging. They’ve been going through all the possible interfaces and devices - some radio, some desktop computer sized - plotting every user journey they could think of, and noting down issues and potential errors. A couple more iterations and they should be there.

Kat and I returned to work on user data this week, after a month-long hiatus, to find it is “just as gnarly a subject as it was then”. Our plan and recommendations are getting into order after lots of healthy discussion (i.e. ridiculously long email threads) with colleagues from the wider BBC and I’m reading a lot about identity, federation and Google+. Chris and George have been by the seaside in Brittany revelling in an EU project meeting and we had more recruitment interviews, I think they’re all done now so it’s just decisions and offers remaining.

Olivier’s link of the week is a blog post in the Harvard Business Review by Eli Pariser (“he of ‘the filter bubble’, a term we hear floating around our office and meeting rooms a fair bit lately”) on Seven Things Human Editors Do that Algorithms Don’t (Yet)

Kat went to the Transmedia Meetup on Tuesday so her link is Varytale, a new way of publishing interactive fiction. “It allows readers to sync their reading across devices, and share what they’re reading with others, exploiting that opportunity suggested by the Kindle’s highlighting function. ebooks and ereaders don’t seem like devices that would be of much interest to the BBC, but it’s interesting how often Kindles, ebooks and interactive reading crop up in our workshops at the moment.”

Chris L enjoyed this paper on performing sentiment analysis on twitter data, particularly the phrase “A variegated mosaic of microblogging uses”.

I found this discussion of federated social networks very useful (though somewhat ironically but inevitably this was posted on Google+).

And this automatically turns Eastenders into a comic strip.