Jake Kolojejchick talked about his work bringing collaborative visualization to the military, through the DARPA project “Command Post of the Future”.
In the context of military, it makes sense to test the system under stress – he mentioned making the test sessions progressively harder on the system and users until it was obvious what was broken in the system. I wonder if establishing time pressure would work in other interface testing situations, even when typical usage is not time sensitive.
I also liked the default “everything is public” mode, but with default per-user views. Each user gets a view to their own workspace, a view to a shared workspace, and optionally a view to anyone else’s “private” workspace. This is really the lowest-energy configuration for quick collaboration: you don’t want all the clutter of everyone working on the same thing, but you also want to be able to quickly get to other’s analyses.Also, the military commanders did not want any access restriction in the software. That’s really impressive.
His point of making visualization the medium of discussion is really important, as is the point about collaboration literacy. How do we make visualization systems do that? I like Emanuele Santos’s Wikimedia support for VisTrails, for example (I’ll get a link to it here soon). It’s nowhere as sophisticated as the time-critical collaboration support in Kolojejchick’s work, but it is a medium that most everyone nowadays has used, thanks to Wikipedia. Wikis are nice, low-overhead collaboration platforms. We need better visualization there, and sense.us and ManyEyes are great examples in that direction.
Great observation by Ben Shneiderman: this great talk would have been a rejected paper at InfoVis and CHI, or a rejected NSF proposal… How do we fix this? Lively discussion going on right now about funding.