Thread by @rajhans

DRS hot take and a mini thread: a) Showing only the most likely trajectory and b) calling the uncertainty 'umpire's call' are doing a disservice to the acceptability of the DRS. A modeling and simulation + uncertainty analysis perspective: 1/ #AUSvIND

Analysing the dynamics and trajectories of a bouncing ball is literally the 'Hello, world' example of our community. You pick up any course material, and this is likely the first example, e.g., inst.eecs.berkeley.edu//~ee291e/sp20/… 2/

If everything is known *exactly*, there can only be one outcome, i.e., one possible trajectory. You can write mathematical equations about it and even plot a computer simulation. e.g., mathworks.com/help/simulink/…. If there is no variation, you will get the same result every time. 3/

The moment there is some uncertainty, one must take into consideration not one but a range of possible trajectories. For example, here are the trajectories of a drone trying an evasive maneuver to avoid a cliff, an example from my first paper. 4/

For the same initial condition, given the slight variation, you can have a range of possible outcomes. Blue trajectories definitely avoid the cliff, red trajectories definitely hit the cliff, and the yellow are the ones where the outcome is close enough that it is 'uncertain'. 5/

When you show only one trajectory on the screen, but call it 'umpire's call' to represent the *range of uncertain outcomes*, it confuses the viewers and the commentators alike. Everyone sees the trajectory shown as the *only possible outcome*. 6/

And so when they see "the ball was clearly clipping the bails" ... 7/

... you see takes such as these.

https://twitter.com/rohangava9/status/1343419299126394880?s=20

. It's not whether it is *just* clipping the bails or not, but that we don't know whether it is even clipping the bails. The graphic gives the former impression, whereas latter is the reality. 8/

Hurricane forecasts do a much better job and show a cone of uncertainty, e.g., this one from Sandy. See how there is now one nominal projected trajectory shown here. 9/

So a much better approach would be to call uncertain outcomes simply 'uncertain' or 'benefit of doubt' to co-opt the historical language, and award it consistently regardless of the on-field umpire's call. 10/

Currently the on-field call (the "umpire's call") is used to 'disambiguate' the uncertainty in the simulation, but it is confusing to everyone, especially DRS is supposedly to 'review' the on-field umpire's decision. 11/

Instead, a much better approach would be to award either the whole uncertain range of outcomes to the bastman (the historical 'benefit of doubt goes to the batsman' adage) 12/ ...

... or award the entire uncertain range to the bowler (modern day game is heavily in favor of batsmen anyway, so award the 'close enough' to the bowler). Regardless of the on-field umpire's call. 13/13

To clarify: I was talking about cricket in the large, including limited overs cricket which is what has become more batsmen-friendly. The thread is about not just DRS in Test cricket, but DRS in general. The most recent Test was simply the occasion that brought it in focus. 14/13

https://twitter.com/rajhans/status/1343754648776269830