Last Friday was the first time I attended a talk at a university in a long time. After regular visiting of conferences in the past I had not had much time and opportunity to go, and so I did not attend talks or conferences for a few years. But with Mary Czerwinski from Microsoft Research being in the Netherlands and talking about worker attention I had to find time to attend. I know Mary from a joint term in the SIGCHI committee, but at that time we never really talked about work interest because the committee meetings were usually fully booked anyway. Besides, at that time I was not so much interested in information workers, at least not professionally.
I’m not going to summarize the talk here (which would be somewhat hard to do due to the fact that she was highlighting a number of different projects from her research group at Microsoft Research), but two things that stood out for me at the end were logging and context.
A number of the projects had user logging at its basis, and I completely understand the utility of the idea. There is so much information in what we are doing that can be useful later that it really is a waste to let all that information just disappear. On the other hand the amount of data can be vast, so some kind of filtering strategy is needed at the source to keep things manageable. Obviously this is not a new idea, I remember using a predictive unix shell in the early 90’s, for example, but today’s processing power and memory make it much more feasible to do this at a grander level. I will certainly raise the idea at work to see if we can add logging to our Wat vinden wij over Service or incorporate it in one of our future projects.
This would also be a nice challenge for the GNOME desktop I’m using. What can you log on my behalf, and how can you help me with that information later. Sounds like a big but interesting project, but due to the size not something I’ll be able to explore at the moment. I’ve mentioned this on the GNOME 3.0 wiki.
The other interesting concept was, as I mentioned, context. (Just putting you back into that context). A nice visual example of this is Scalable Fabric, where you pile up windows on the side of the screen when not in active use, and group them into clusters. A single window or pile can be restored easily. A more down-to-earth but immediately useful tool is Groupbar, which allows the taskbar to be manually grouped and clustered in meaningful sections, instead of clustering many open windows by application type as is the default in Windows and GNOME. I’ve opened a bug/feature request for that.
But the use of context goes beyond simply piling some windows together and hoping for the best. It also has to do with the documents, to do items, etc. that all make up the context of what you are working on right now, just before being interrupted by that phone call, and what was I doing again? Providing better support for maintaining and switching context can still use a lot of improvement, and will mean a major step forward in making computers truly usable.