Comments on Beyond Total Capture: A Constructive Critique of Lifelogging

Summary

This 2010 article, by Sellen and Whittaker, responds to some of the concerns raised by the concept of Total Capture Lifelogging particularly as practiced by Bell and Gemmell as seen in the MyLifeBits paper we discussed earlier in the week. They note that in many of the papers in the field the reasons or benefits to the user are not explained clearly and tend to be lumped together into five "R"s, Recollecting, Reminiscing, Retrieving, Reflecting and Remembering intentions. They propose some design principles to focus the creation of lifelogging systems onto something that is beneficial, including "targeting weaknesses of human memory" to focus the types of data collected towards actually needed data, "designing effective retrieval cues" focusing on how to make use of autobiographical memories, "support for the Five Rs" focusing on the use the system is actually intended for as opposed to be a general purpose external memory for the person and "Offloading and metamemory"focusing on how to balance efficiency compared to accuracy of the information as well as managing the integration with organic memory.

In general the authors suggest less aggressive systems for lifelogging and ensuring that the applications are actually useful for the users of the system.

Thoughs

As some of the in class discussion went, I found that the concept of total capture was discomforting and not useful. In particular the way in which these systems are designed, as technical exercises rather than as solutions to a given problem, is problematic. I think that the critiques and suggestions that Sellen and Whittaker provide are a good starting point for planning more useful lifelogging systems.

Further I think that these systems need to be based in need, or at least in finding answers to interesting questions. When looking at MyLifeBits, it is never clear what use the users get out of the system at the end of the day. Even if the questions are not of critical life importance (an overview of twitter statistics, for example) having any goal in the system will make the design of these systems more reasonable.

I also think that the total capture is somewhat misguided in terms of the amount of work that needs to be done. While there is something to be said for holding all of your data in a place you own and control, the fact remains that the majority of data we create/encounter each day continues to exist in whatever format and location and can be retreived at our leisure. Over long periods of time data does disappear and some things (like the number of cups of coffee I had to drink today) are more ephemeral, but on average it seems more important to allow a system to "travel back in time" to find what data can be useful rather than trying to gather everything ahead of time.