There's been an interesting confluence of commentary recently on why precision is not only unnecessary, but perhaps undesirable, in the formulation of communication services…
- Dopplr's Matt Jones reflects on Google's Latitude location-sharing service, noting that designing Dopplr to be 'nothing more granular than cities-as-place and days-as-time' is enough to fulfill the goal of surfacing serendipity.
- In Valleywag's Against Realtime, Owen Thomas argues that Facebook's recent makeover has emphasised recency and buried relevancy – in apeing Twitter, Facebook is assuming that 'the only news is breaking news' (Thomas' piece builda on comments from Om Malik's discussion of Facebook's identity crisis)
Dopplr it seems has been motivated by understanding context and what might be useful in a given situation, where Facebook's embrace of the realtime web has been driven by the faddish pursuit of a competitor.
Regardless, there are useful social models and design patterns that need to be abstracted from the Twitter, Facebook and Dopplr articulations of time, space, serendipity and relevancy, patterns that might enhance other services. There's an assumption that relevance and seredipity can emerge from simply aggregating together news items from social connections. Yet there's a growing anxiety that we're all drinking from a firehose of data.
Why can't Twitter, for example, learn to whom users grant their attention over time…or Facebook understand to whom I'm 'nearby' (at Matt says – 'hereish-and-soonish/thereish-and-thenish'), helping users make relevancy rather than recency based choices, that wire serendipity into the fabric of social software.