Monday, October 28, 2019

Fixing the Strike Zone - Pitch Challenge Proposal

Today we unveil a responsible proposal for fixing the strike zone through technology: not full robot-umpire pitch calling, but a challenge system instead. Tech is coming to baseball and MLB has the opportunity to fix obvious errors while preserving umpire autonomy regarding most aspects of officiating behind home plate. You might think this rather out of left field for us, but we did call ourselves The Left Field Corner all those years ago for a reason.

As we've known for several years, baseball's attempts at incorporating technology for balls and strikes has proven rather frustrating for fans and umpires alike: mostly because teams and fans tend to zero-in on perceived missed ball/strike calls shown on television broadcasts and internet graphics while umpires lament that fans are exposed to a flawed computer system that is subject to a multitude of errors.

Publicly available pitch data exposes teams and fans to statistics suggesting umpires are 91% accurate behind home plate while MLB's evaluations show an average 97% Zone Evaluation score...that creates a disconnect of problems.
Related PostPodcast - Truth About Baseball's Electronic Strike Zone (6/5/19).
Related PostCall for Umpire Accountability & the 97% Plate Score (4/19/19).

On the one extreme are fans demanding fully automated pitch calling, similar to the Automated Ball/Strike System (ABS) attempted in the Atlantic and Arizona Fall Leagues (which resulted in several coach and player ejections for arguing the computer's ball/strike calls...there were additional ejections to those we've highlighted, but unfortunately, they did not feature video).
Related PostComputer Strike Call Prompts Navas' AFL Ejection (10/16/19).
Related PostAtlantic League ABS Robo-Ump Ejection Encore (9/28/19).
Related PostHistory - Baseball's First Ejection Due to TrackMan (7/12/19).

Robots? Torre said no, but Manfred said yes.
On the other are traditionalists seeking to preserve the human element, and somewhere in the middle is Close Call Sports, which acknowledges that technology can assist in the ball/strike call mission, while also acknowledging that the very system baseball presently uses on MLB Gameday, Baseball Savant, Brooks Baseball, PitchCast / Fox Trax / ESPN K-Zone is prone to error.

Umpiring can be an art, while ball/strike analysis is generally a science. Through our proposal, we'll try and strike a balance: preserve the art while incorporating some of the science.

Both sides have to face some facts: For fans, no, a fully-automated system may not be realistic at this time. For umpires, the tech might have reached a breaking point similar to Replay Review before its 2014 expansion. The NBA just adopted a challenge system for some foul calls, the NHL expanded its existing challenge system, and the NFL red flag is famous. Maybe not now, maybe not next year or two, but soon.

MLB Chief Baseball Officer Joe Torre said in 2019, "I don't see the robotic strike zone happening," while MLB Commissioner Rob Manfred stated his motive: "I think an automated strike zone puts you in a position to manage that strike zone. Where should it be exactly to produce the amount of offense that you want?" Theoretically, a computer could be tweaked to only calls strikes between the thighs and belt, for instance, if Manfred wanted to "manage" the zone that way...assuming, of course, a most accurate computer.
Related PostCiting Atlantic Lg, Manfred Ready for Robo-Zone (8/19/19).
Related PostTorre Doesn't Want Robot Umpires in MLB (7/26/19).

Vertical zone error remains an unanswered ?
Proposal: To mediate the discussion, it would follow that a compromise might be at hand: to the robo-ump crowd, there are some elements of the game that computers are unable—proven time and time again—to adjudicate, while to the traditionalist faction, technology exists to help guide us to the concept of correcting "the obvious miss."

As such, we propose a challenge system, which we further break down into one of two options depending on how optimistic you feel about TrackMan (Hawk-eye, going forward)'s ability to adjudicate pitches.

For instance, the Automated Ball/Strike System (ABS) was subject to the following errors:
> Standardization and Radial Error;
> Calibration Error;
> Vertical Strike Zone Error;
> 2D vs 3D Zone Measurement Error;
> Final Approach Error (e.g., pfx is measured at 40 feet [NOT 60 feet]);
> Tracking, Capture, and Margin of Error (1-2 inches per statistics);
> Computer failing to see pitches (fails to register that a pitch was thrown).

Delayed pitch calls remain an issue.
Trajectory Measured by Automated Caller (TMAC): The TMAC system is as black-and-white as you can get and assumes the computerized pitch tracker is 100% accurate at all times and never makes any mistakes. Margin of error is zero, for under TMAC, "error" does not exist. Another way to think of the TMAC system is a computer that after a brief delay calls the pitch on its own, similar to Hawk-Eye in tennis. NOTE: For TMAC to be incorporated at the present time means that baseball would BE REQUIRED to accept inaccuracy, which might be somewhat ironic since the whole idea stems from a fandom unhappy about perceived umpire inaccuracy.

Here are the basics:
> Each team receives 3 challenges per game. Like Replay, a won challenge is retained.
> > The system will declare each challenged pitch to be a Ball or Strike.
> Offense: Only a batter may challenge a call of "strike."
> Defense: Only the catcher or pitcher may challenge a call of "ball."
> For pace-of-play purposes, a challenge must be immediate upon the umpire's call.
> One additional challenge per team for each three extra innings (e.g., +1 in 10th, +1 in 13th, etc.).
> If Hawk-Eye malfunctions or fails to capture the pitch, no challenge is charged (call stands).
> There is no other method for Borderline/Call Stands because the system assumes that the computer is 100% accurate on all pitches that it sees (aka Zero-Error).

> The Atlantic League's ABS experiment turned up quite a number of malfunctions in that the system sometimes failed to capture as many as 50% (possibly even more) of pitches thrown during a given inning. On several occasions, ABS had to be shut off for the night due to poor performance.
Related PostABS Playoff Highlights - Delayed Calls & System Errors (10/1/19).
Related PostALPB - Inconsistent TrackMan Use Riles Coaches (7/22/19).

Robo-Umps: Bringing rivals together since '19.
Graphic Inaccuracy Logic (GIL): The GIL system acknowledges that the computer sometimes makes mistakes on close calls, are occasionally subject to errors, and affords the benefit-of-the-doubt to the umpire. Another way to think of the GIL system is one that operates under the framework of Replay Review in that all calls are Confirmed/Stands/Overturned, with Stands going to the original on-field ruling.

The basic structure of GIL is similar to TMAC, except that "Call Stands" is a realistic outcome if the pitch falls within the UEFL f/x's one-inch margin of error (could be expanded to ML Private/Zone Eval Equivalent's two-inch margin of error, depending on administrative preference). Like Replay Review, a challenge is considered lost if the call stands.

For instance, the GIL system would attempt to account for an error such as vertical zone error—see our work on postgame processing for example—while for the TMAC system to work effectively, no such error can exist.
Related PostPostgame Processing Changes Gibson's Strike EJ QOC (9/21/19).

The question is whether to nitpick or not.
Conclusion: That's our proposal to incorporate technology into the game while acknowledging its shortcomings and accounting for those close calls that the computer cannot decipher. The goal with the challenge proposal is to correct the "obvious miss," similar to what Tony La Russa once said was the original purpose of expanded Replay Review.

The idea with the GIL system is to avoid the level of pedanticism that plagues those out/safe reviews that overturn a safe call to an out call based on slow-motion replay that shows the runner—who clearly beat the throw—broke contact with the base for one-tenth of a second. I wrote about this level of technicality in 2018 (see related post)...but if you don't mind that, and assume the computer is always right, then choose the TMAC system.
Related PostReplay Rewind - Technically Correct or Spiritual Travesty? (6/9/18).

In either event, the goal should be to correct the "that strike was three inches outside" call. If the umpire's was truly an incorrect call by such an amount, as alleged, we'd like to help fix that.

Proposal video as follows:

Alternate Link: Fixing MLB's Strike Zone - A Challenge System Proposal (CCS)


Post a Comment