Saturday, 19 January 2008

The Video Gauntlet

The whole week at work was spent in a gauntlet with people perceptually rating video quality on something I'm working on. Perceptual qualification, whether statistical or empirical is a nightmare. Unfortunately marketing teams have to depend on human behavioral traits to sell, so doing perceptual quality analysis is supposedly good.

This is probably the fourth time within a quarter that people who originally admitted quality was much better than prior, reverted saying it wasn't so and needed improvement. That reminded me of the Akbar-Birbal story (sometimes attributed to Tenali Rama) of how a line drawn on the floor can be made shorter without changing it. Birbal in one version of the story just draws a longer line beside it demonstrating relativism. By induction that line too could be made shorter.

Thanks to some help from a Camera expert and a few more efforts, we did get a good percentage of techies to be happy with the results. I find that as human beings our qualification of everything in this world is relativistic and based on prior experience. If something is entirely new with nothing to compare with, we still find something in our mind and provide a relativistic score as to how "good" or "bad" something is. The whole technique is platonic.

This relativistic comparison is so deep-rooted that it becomes a major drive for social improvement and improvement of civilization as a whole. Human beings always comparing and trying to find the "better" are the drive for constant improvement. Kaizen is rooted biologically in our very selves as evolution itself would be.

Engineering attempts to freeze specifications are against this very biological instinct. We probably need a new technique in this world of rapid change to create products or solutions for people. The Gordian knot here is "when do we stop comparing?" This is food for thought.