Forgetful Interfaces Workshop/Discussion @ LIFT
At last week’s most amazing LIFT conference I organized a workshop / discussion on forgetful interfaces. I’ve posted a report from the workshop on my very own LIFT blog, and it’s reposted here as well:
It’s a bit over a week since the workshop on Forgetful Interfaces. I proposed the workshop because I had a feeling that simply storing the data we collect about ourselves may not be such a good idea. To my joy and surprise, all proposed workshops happened, including mine! However, because my thoughts on Forgetful Interfaces were still uncrystallized, I decided to turn the workshop into a discussion. And what a discussion it was. Thanks to all of you who took part, it was truly amazing.
Now, originally I had approached the subject from the perspective of privacy, which turned out to be perhaps the least interesting aspect of forgetful interfaces. We’ve arrived at three different aspects, and there is a fourth one which was introduced to me, outside of the workshop, by Irwin Oostindie. These aspects are privacy, filtering, user experience and security (the latter proposed by Irwin). But, first of all, what is a forgetful interface? In my view, a forgetful interface hides artifacts from view. Here I’ve used the term artifacts rather than data, because forgetful interfaces are about data in a social context. I’m specifically not talking about deleting the artifacts. They’re still there, just harder to get at. Just like how you may forget things yourself, but given enough clues your brain can bring them back.
So, with that out of the way, let’s look at the aspects of forgetful interfaces.
We’ll start with privacy. The publicity of an artifact can be placed on a scale, from intimately private to extremely public. The closer to the public end of the scale an artifact is placed, the more a forgetful interface may be welcomed. Such an interface could, for example, “forget” certain artifacts after a month. Given that more private artifacts are only available to a smaller group of people, there is more social control over these artifacts which lessen the need for forgetfulness. An example was given of pictures from a party where somebody did not, under any circumstances, wished to have her photographs displayed publicly. Violating this agreement would most certainly place the violator outside of the social group of those attending the party.
Filtering is a type of forgetful interface which helps you find relevant stuff, without throwing away anything. Artifacts which never get touched may be moved out of the main interface, perhaps even be compressed and stored externally. The artifacts are still there, however it’s become pretty hard to use them. Robin Hunicke mentioned this is how the New York Times works: some articles can be found online, but you have to pay to retrieve others from storage. Jerome Ribot just published an awesome post about using forgetful interfaces for filtering. I suggest you read it :)
A third aspect of forgetful interfaces is user experience. Pretty much all social sites will send permission requests your way, to join a specific group, or add somebody as a contact. This bombardment of requests is an oft-heard complaint about Facebook. Last.fm and also the LIFT site will show you a list of people who have not yet added you back as a contact. Being confronted with this list time and time again may become a bit… painful. But, let’s face it, if the permission requests aren’t fulfilled within, say, a week or two, it’s probably not going to happen anyway! In order to improve the user experience, why not forget about them? (Yes, this may actually mean deleting the requests, so perhaps this could be called a deleting interface).
Security may also be a deleting interface. In short it’s about automatically hiding artifacts from third parties when they’re no longer of direct use to the primary party. In order to successfully hide artifacts, they’ll have to be destroyed or perhaps be obscured by large amounts of random data. Truth be told, I haven’t given this aspect much thought yet.
Determining which artifacts ought to be forgotten was given some attention during the discussions, but perhaps not in enough detail. It was argued that we’d need to keep track of use patterns of artifacts in high detail, in order to correctly predict which artifacts users remember and which they don’t. Personally this tracking of use patterns scares me, and I think some more simple heuristics may do quite well. Recency of the artifacts and their use, and some information about the social relationships formed around the artifacts, combined with a simple measure of attention spent on the artifacts should provide enough information to determine which artifacts are important and which are not. Artifacts which have been created or accessed recently have a higher chance of being accessed again, and should therefore be preserved. This is the principle on which most computer caches rely. However, this means that the pictures of your newly born child from five years ago run the risk of being forgotten – even though the social relationships around these artifacts are very strong. Here attention comes into play, carefully describing and organizing an artifact indicates it’s more relevant than artifacts which are never organized. Knowledge of social relationships helps those artifacts about your parents from being forgotten, even though you may rarely interact with these artifacts.
Finally, a word about how a forgetful interface should be presented to the user. We arrived at the metaphors of fading and decaying the artifacts. These metaphors are very much visual, and help convey that the artifacts are disappearing or falling apart. Triggering one of the heuristics may reverse the process so the artifact is preserved (for a little while longer).
The discussions I’ve had at LIFT about forgetful interfaces have been very, very interesting. I definitely plan to delve into this further, and I’d like to thank again all of you who shared your wonderful insights and experiences. Now, let’s discuss this some more ;-)
