Comparing AI's Failures with Ubicomp's Visions

There seems to be a growing trend to critique the calm and seamless vision of ubiquitous computing aiming to the “fantasy of the perfect” (referring to Lucy Suchman’s Human-Machine Reconfigurations). Matthew Chalmers pioneered with his notion of seamful design, slightly inspired by Weiser. Recently Genevieve Bell and Paul Dourish suggested that ubicomp is messy and seamlessness is a misleading vision. From my own experience, those points of view are a thin minority in a still extremely techno-utopian-driven research field. In contrary, it seems that designers acknowledge more the messiness of the everyday life. For example, in a post on reconfiguring the old future, Ben Kraal mentions that:

seamful design of ubicomp systems not only recognises the impossibility of a completely seamless, invisible, ubicomp infrastructure but embraces the messiness of everyday life.

He also points to the recent thoughts of Larry Irons who compares the failure of AI with the promises of ubicomp.

I was reminded of the many promises that artificial intelligence made for expert systems in the 1980s as he describes how the designers of context-aware, ubiquitous computing think they can make it work. [...] Having machines act in a sociable manner are credible after all the fiascos and shortfalls of the past 30 years. [...] The challenge is AI-hard. Yet, a fair reading would probably characterize it as AI-impossible.

I particularly enjoy Larry’s answer on a question about the politics of ubiquitous computing and how we should question first the hype around seamless interface as being the default objective:

Good question, but it won’t find a reasonable answer as long as designers are unwilling to ask the obvious questions about claims made by those who hype the need for a seamless interface to ubiquitous computing environments. You can only meaningfully address the question about the politics of ubiquitous computing ethics when a seamful interface is considered the default design objective. In my mind, this is why Greenfield is correct to insist that seamlessness must be the optional mode in such applications

Relation to my thesis: As mentioned earlier, I am really glad to see more thoughts on the perspective of ubicomp as being inherently messy. The comparison with the failures of AI is rather relevant for research aiming at a calm and seamless future. This is related to the “are we there yet” question around the definition of ubicomp and my e-minds paper Getting real with ubiquitous computing: the impact of discrepancies on collaboration which was my first (immature) attempt to highlight the limitation of the seamless vision around ubicomp. My “disturbed city” flickr set is also an abstract attempt to reveal the messiness of the urban life from which I do not perceive ubicomp as a solution. My possible talk at LIFT on Embracing the real world’s messiness, will be the opportunity to reflect on all this.

This also makes me think about the discussion between Brenda Laurel and Bruce Sterling about the design for pleasure and technological fairies (design for illusion, for magic) that happened at Ubicomp 2006.