User-based Blending

The traditional approach to modeling is for modeling experts to observe, analyze, synthesize, etc. to produce a model of a real world decision space. This model is then used by a non-expert model user to explore that decision space.

Occasionally, a modeling tool emerges that allows non-experts to create fairly sophisticated models without expert intervention....the one I always think of is electronic spreadsheets, which were the initial killer app for the PC.

A potentially interesting development in user-based, non-expert, modeling tools is described in a recent TED presention...Siftables from David Merrill of the MIT Media Lab. The basic idea raises interesting possibilities and questions. For example, each Siftable could be instrumented to monitor how it is used, and log and report that data to a central repository which could then be mined by experts for insight into how to modify the existing Siftables' methods, display, etc. Or, users and experts could use Siftables as a collaborative design artifact to explore how to define a decision space. Or, Siftables could be used as a distributed workflow data collection tool by distributing them to individuals in a value chain/net with instructions on how to model decisions/workflows in that chain/net....sort of a physical wikipedia entry creation activity.

This sort of thing has been enabled in a virtual way with multi-touch capabilities like Microsoft's Surface. However, it may be that Surface is too configurable to efficiently address some modeling needs...the physical constraints imposed by Siftables may actually help catalyze more effective exploration in some situations.

Maybe this goes nowhere, but it seems like an innovative new way to explore a decision space by blending the physical and information domains...it's that distinctive blending of the two that I find most fascinating.