Dunno about the foul tip, and the infield fly rule is at least somewhat a judgment call IIRC. But I’m sure a machine can tell fair from foul balls a hell of a lot quicker than the fastest umpire.
And that’s one call that doesn’t have to be made instantly anyway. If you’re a baserunner, you act as if it’s fair until you’re told it’s foul, because if it’s foul you just walk back to the base you were at before the pitch was thrown. No harm, no…OK, maybe that’s not the best phrase here.
Changes to the uniform requires new rules. What does the umpire do about a player who is shifting his uniform to move those markers? That requires a new rule. What if the markers get covered in dirt during a slide? Baseball is all about rules, you can’t change the game at all without changing the rules.
I don’t believe an electronic strike zone system would require any new rules* or uniform changes–just some good programming and calibration. A player’s “natural” stance is the stance he actually takes in the box. The positions of a batter’s shoulders, belt, and knees are readily visible to cameras with the right angles. Live imaging of the batter in the box is certainly more objective than physical markers on each player’s uniform.
Well–apart from the rule handing ball-and-strike authority to such a system.
You build a model of a human body and then train the system. The knees are where the legs bend. Baseball players bend their knees when they hit, and the legs can’t really bend at any other point. The torso is the part of the body between the waist and the shoulders. The bodies and stances of ballplayers simply don’t vary enough to make this a difficult computer vision problem. You don’t need markers on the clothes. You don’t need anything special. A reasonably competent team of three developers with a few years of computer vision experience under their belt could hash this out in a few weeks.
I get that there are implementation details to be worked out, and it’s possible (although not at all likely in my opinion) that it would be difficult to make an automated system that’s better than a human. But there’s an incredibly easy way to test this.
Just set up a bunch of cameras to record some pitches in real games. Later, have a bunch of professional umpires go over the footage and categorize the pitches as balls and strikes, and also have your automated system do so. Then compare the actual on-the-spot calls with the after-the-fact consensus and the output of the algorithm and see which is better.
If the on-the-spot call is better than the algorithm, then you keep using them and work on the automated system until it’s better.
That’s if you’re serious about trying to solve the problem of bad calls. If you’re searching for stumbling blocks, I’m sure we can come up with some more.
By counting on the current level of technology to resolve the vertical portion of the strike zone you will be introducing tremendous controversy. You are introducing new definitions because they are not based on human judgment as the old definitions were. If this is accepted by the teams that’s fine, but I doubt that will happen. There will be constant arguments over the definition used by the machine, and if a video record is not produced that can be confirmed by human eyes we will end up with bad calls going to court instead of just some antics on the field.
Feel free to disagree with me, but I assure there will be no robot umpires in the foreseeable future of baseball.
But since you have to have the umpires in the game for everything else, you could design a computerized system that would buzz an earpiece in the homeplate ump’s ear for regular calls. Everything else he does as normal.
I still don’t like it, oppose it with all my heart and soul, and would hope to see such a system crash and burn.
I did provide a solution to bad calls. Video shot directly above or below the plate, and at right angles to the line between home plate and second base will reveal the ball as it moves through the strike zone. Officials can review it and resolve any questions about knees or torsos in the case of a bad call. I assume limited challenges will be allowed a la the NFL. The same technology can be used to review the performance of umpires after the games and get rid of the ones who are consistently making bad calls.
I think it’s quite likely you’re right. But it won’t be because we can’t make a technical solution that’s better than the current system by any reasonable metric you choose to throw at it. It’ll be because entrenched interests and people who don’t want to consider any changes on their merits piss and moan about it.
So it goes.
Like the old saying goes, if you can’t make it work perfectly on your first attempt, it’s probably not worth doing, right?
I think the idea of having a limited review system like the NFL is a bad one. Correct calls should not be a limited resource to be strategically conserved. I understand not wanting to bog the game down in constant reviews and challenges, but there are better solutions.
If we really think that the height of the strike zone is incomputable, then I think an automated system that only determined whether the ball passed through the column (and shows a quick red/green light to the ump to use in his call) would still be an improvement over the status quo.
I think it’s quite likely you’re right. But it won’t be because we can’t make a technical solution that’s better than the current system by any reasonable metric you choose to throw at it. It’ll be because entrenched interests and people who don’t want to consider any changes on their merits piss and moan about it.
[…]
That sounds like Karl Marx-type reasoning.
As far as I’m concerned, any kind of automation of umpires’ duties would go over with the fans like a lead balloon. I’m not concerned with a small group of purists.
I again have to point out that the technology exists, it’s called Pitchf/x, and it works near-perfectly. A human decides where to set the upper and lower limits for each batter, and the job is done. Honestly, there is no problem to solve here. It was solved years ago. It works.
Leave the home plate umpire there but just put a big monitor in center field that says STRIKE or BALL and voila, you’re improved the accuracy of ball/strike calls by 99% and slowed nothing down. The ump can still make calls on other stuff.
According to the link posted and the information provided here the position of the ball is calculated. If someone complains that a pitch was out of the strike zone there is no way to visually confirm it. That technology will never be used in MLB. It would be trivial for them to have a video record as I described, and then it could be accepted. Frankly I don’t know why anyone isn’t doing that, it’s trivially simple technology. I would have no objection to it’s use, but an umpire still has to be there, and there has to be a review process.
I have never been ejected from a bar for arguing balls and strikes.
It seems to who? And I don’t mean the guy on first. I know it’s more accurate than a human, but how are you going to prove that it’s always correct? How do you know the machine is operating correctly. Who (again not the guy on first) is going determine if the machine is operating properly or not when everyone on earth except for the team that was advantaged says some pitch didn’t cross the corner of the plate?
