Robo strikezone: Not as simple as you think -- Baseball Prospectus

Plympton91

bubble burster
SoSH Member
Oct 19, 2008
12,408
“I only would go to the automated strike zone when I was sure it was absolutely the best it can be.”

I don’t understand why this should be the standard.

Shouldn’t it be, “I don’t want to make the change until I am sure it is better than live umpires.”?

And, is there anyone who doesn’t think that we are already at that point? As I said upthread, ask yourself two questions:

How many times have you thought, “ Whew, I’m glad there was a human back there because the system obviously got that wrong.”

Vs.

“Holy crap, how did the umpire miss that call when it was so clearly a ball/strike.” and the machine agrees with your view of it.
 

CarolinaBeerGuy

Don't know him from Adam
Gold Supporter
SoSH Member
Mar 14, 2006
7,468
Kernersville, NC
I’m all for it, but understand wanting it to be significantly better than human umps (if not perfect). That said, this can’t happen soon enough.
 

InstaFace

MDLzera
SoSH Member
Sep 27, 2016
15,334
Pittsburgh, PA
they've agreed to "cooperate and assist" with "development and testing", if and when Manfred decides to implement at the ML level.

"MLB has discussed installing the system at the Class A Florida State League for 2020. If that test goes well, the computer umps could be used at Triple-A in 2021 as bugs are dealt with prior to a big league callup."

The article also notes that MLB has yet to discuss with the players association.
 

Cumberland Blues

Dope
Dope
Sep 9, 2001
5,084
Bad for elite receivers who can't hit, but good for everyone else. I do wonder though - when the automated strike calling is installed everywhere in MLB and MiLB - will this create a divide between what catching skills are most desirable in the amateur and pro ranks? I can see top tier D1 colleges installing this (like ACC/SEC/PAC12), but smaller D1 schools, D2/D3 and HS are not going to have this - so in those places framing is still likely to be a valuable skill. But in the pros - you'll just need to block & throw. So catching may still be a defense first position in HS/college, but more offense oriented in the pros.
 

Murderer's Crow

Dragon Wangler 216
SoSH Member
Jul 15, 2005
20,728
Garden City
Bad for elite receivers who can't hit, but good for everyone else. I do wonder though - when the automated strike calling is installed everywhere in MLB and MiLB - will this create a divide between what catching skills are most desirable in the amateur and pro ranks? I can see top tier D1 colleges installing this (like ACC/SEC/PAC12), but smaller D1 schools, D2/D3 and HS are not going to have this - so in those places framing is still likely to be a valuable skill. But in the pros - you'll just need to block & throw. So catching may still be a defense first position in HS/college, but more offense oriented in the pros.
It's a good question. Framing will almost certainly have a diminished value but there is a good chance that without framing, other aspects of a catcher's defense improve. Gary Sanchez is a good example, when he focuses on framing he allows passed balls. When he relaxes his attempts to frame, his passed balls decrease.
 

InstaFace

MDLzera
SoSH Member
Sep 27, 2016
15,334
Pittsburgh, PA
How much of what today determines whether a catcher sticks at C vs moves to another position is framing, vs blocking and throwing, vs pitch-calling? My sense is that framing was elevated as a distinct skill only a few years ago (the Christian Vazquez era, let's say, or at most the Molina Era), when it could be quantified. I think that it's always been true that your offensive skills determine how much slack you get on catching defense, or if you get moved to another position. The catchers who hit well enough they'd have stuck in the lineup even at 1B or RF are few and far between.

I think players who have the catchers' defensive skillset and project out as ML-level hitters will be unaffected by a change in the pros / upper college ranks that would remove the value of "stealing calls from bad umps", i.e. framing. Far more important to learn to read hitters and call a good game, throw consistently and accurately to 2nd, have good mechanics on blocking, etc.
 

Heating up in the bullpen

Member
SoSH Member
Nov 24, 2007
700
Pittsboro NC
The New Yorker has a piece on the automated strike zone experiment in the Atlantic League (and beyond).
Invasion of the Robot Umpires
The minor leagues have been testing the Automated Ball-Strike System. But isn’t yelling and screaming about bad calls half the fun of baseball?
https://www.newyorker.com/magazine/2021/08/30/invasion-of-the-robot-umpires?utm_source=nl&utm_brand=tny&utm_mailing=TNY_Magazine_Daily_Subs_082321&utm_campaign=aud-dev&utm_medium=email&bxid=5df8f8742a077c3fe35b8f36&cndid=59409573&esrc=&utm_term=TNY_Magazine_Daily_Subs

As you may pick up from the subtitle, the author isn't a fan of the robo-ump. Despite that, the article is interesting and does offer a wide range of perspectives.
He quotes/refers to Bill James a couple times, first in the middle: "Of course, compassion toward the pitcher is cruelty toward the hitter. "I don’t know of any other sport in which the umpires would even talk about making up their own rulebook,” Bill James, a writer and a former Red Sox executive, widely considered the godfather of advanced statistics, told me."
Then toward the end: "Bill James suggested that enough significant mistakes from umpires could, from a fan’s perspective, make a game seem almost arbitrary. He offered a middle ground: A.B.S. could rule on obvious balls or strikes. But a couple of inches around the border would be a “zone of discretion”—up to a living, breathing umpire to decide."
The latter idea -- "A.B.S. could rule on obvious balls or strikes. But a couple of inches around the border would be a “zone of discretion”—up to a living, breathing umpire to decide" -- I find preposterous. To me the whole point of the system is to get those calls around the border right. With a few exceptions, umps can manage to call the obvious balls and strikes. It's the close calls, the pitches on the black, or two inches off the plate, where the human umpire so often fails the test.
 

shaggydog2000

Member
SoSH Member
Apr 5, 2007
9,285
The New Yorker has a piece on the automated strike zone experiment in the Atlantic League (and beyond).
Invasion of the Robot Umpires
The minor leagues have been testing the Automated Ball-Strike System. But isn’t yelling and screaming about bad calls half the fun of baseball?
https://www.newyorker.com/magazine/2021/08/30/invasion-of-the-robot-umpires?utm_source=nl&utm_brand=tny&utm_mailing=TNY_Magazine_Daily_Subs_082321&utm_campaign=aud-dev&utm_medium=email&bxid=5df8f8742a077c3fe35b8f36&cndid=59409573&esrc=&utm_term=TNY_Magazine_Daily_Subs

As you may pick up from the subtitle, the author isn't a fan of the robo-ump. Despite that, the article is interesting and does offer a wide range of perspectives.
He quotes/refers to Bill James a couple times, first in the middle: "Of course, compassion toward the pitcher is cruelty toward the hitter. "I don’t know of any other sport in which the umpires would even talk about making up their own rulebook,” Bill James, a writer and a former Red Sox executive, widely considered the godfather of advanced statistics, told me."
Then toward the end: "Bill James suggested that enough significant mistakes from umpires could, from a fan’s perspective, make a game seem almost arbitrary. He offered a middle ground: A.B.S. could rule on obvious balls or strikes. But a couple of inches around the border would be a “zone of discretion”—up to a living, breathing umpire to decide."
The latter idea -- "A.B.S. could rule on obvious balls or strikes. But a couple of inches around the border would be a “zone of discretion”—up to a living, breathing umpire to decide" -- I find preposterous. To me the whole point of the system is to get those calls around the border right. With a few exceptions, umps can manage to call the obvious balls and strikes. It's the close calls, the pitches on the black, or two inches off the plate, where the human umpire so often fails the test.
You'd want exactly the inverse. A human there to over rule the obvious errors that are going crop up now and then with the system over hundreds of pitches a game and a dozen games a day for months. And the close calls being left up to the machine that is always in the same location no matter whether the batter is lefty or righty, or how the catcher sets up.
 

OurF'ingCity

Member
SoSH Member
Apr 22, 2016
7,174
New York City
You'd want exactly the inverse. A human there to over rule the obvious errors that are going crop up now and then with the system over hundreds of pitches a game and a dozen games a day for months. And the close calls being left up to the machine that is always in the same location no matter whether the batter is lefty or righty, or how the catcher sets up.
Exactly. What they need is a the equivalent of the tennis challenge system, where the ump makes the majority of uncontroversial in/out calls but the players can challenge a close one. Maybe limit it to one challenge per at bat for the batter and pitcher each.
 

jon abbey

Shanghai Warrior
Dope
SoSH Member
Jul 15, 2005
62,236
Exactly. What they need is a the equivalent of the tennis challenge system, where the ump makes the majority of uncontroversial in/out calls but the players can challenge a close one. Maybe limit it to one challenge per at bat for the batter and pitcher each.
Actually during the US Open going on currently, they have totally removed the human element because of Covid. The initial calls are made by the electronic system and there is no challenging since they’re already using the electronic system, and it seems to be working great. Baseball will instantly be a far better sport if they can get this working successfully.
 

Van Everyman

Member
SoSH Member
Apr 30, 2009
22,843
Newton
The immediate tossing of Plawecki and then Cora was actually worse. The fact that these guys double down on their mistakes makes it so, so much worse.

I continue to think the solution is for balls and strikes to be taken away from umps but for them to still actually call them. It’s important they still manage the flow of the game, allow time to be called, etc. Let the umps see the balls and strikes through a Google Glass-type visual. That would take pressure off these guys to match what viewers at home with 4K DVRs at home have without completely draining the actual human element from the game – which despite all the chuckles I do think is important to the feel of the sport.
 

lexrageorge

Member
SoSH Member
Jul 31, 2007
15,035
View attachment 51551

We all know it was awful last night, but this is the worst run differential I think I’ve ever seen.
Here's the problem. Calls 2 and 3 are actually quite close, and getting 94-95% accuracy on those calls is about as good as it can get. #1 sucked as it cost the Sox at least one run, but it was in the ump's strike zone.

Hard to see how calls can get more accurate unless one goes to a robo system. But would it really be more accurate? Technology can be plagued with deployment problems that sometimes persist for longer than fans would care for.
 

OfTheCarmen

Cow Humper
SoSH Member
Jul 18, 2007
4,276
The problem is the consistency. You would need to see how many balls and strikes are called for the same location. If the Ump is "giving them the bottom" (or whatever arbitrary non-balls they decide to call that day) but not consistently, then the hitters and pitchers cannot make appropriate adjustments.
 

Bergs

Member
SoSH Member
Jul 22, 2005
18,233
Here's the problem. Calls 2 and 3 are actually quite close, and getting 94-95% accuracy on those calls is about as good as it can get. #1 sucked as it cost the Sox at least one run, but it was in the ump's strike zone.

Hard to see how calls can get more accurate unless one goes to a robo system. But would it really be more accurate? Technology can be plagued with deployment problems that sometimes persist for longer than fans would care for.
Yes.
 

jon abbey

Shanghai Warrior
Dope
SoSH Member
Jul 15, 2005
62,236
The tennis system seems to work perfectly, it's more two-dimensional than a three-dimensional strike zone, but I think they can perfect a baseball one in time. But they don't seem to be close yet...
 

Max Power

thai good. you like shirt?
SoSH Member
Jul 20, 2005
5,982
Boston, MA
The tennis system seems to work perfectly, it's more two-dimensional than a three-dimensional strike zone, but I think they can perfect a baseball one in time. But they don't seem to be close yet...
Former umpire Dale Scott has been making the podcast rounds lately and had a few issues with the automatic zone.

First is that it changes for each batter. The box we see on TV isn't really calibrated for each hitter and their stance. The umpire reports for each game come a day or so after they finish because someone goes through and manually sets the top and bottom of the zone for each hitter.

Second was the trickiness of a 3D zone and how it relates to what everyone would like to be a strike. A curveball that knicks the back top of the zone doesn't look like a strike to anyone. Neither team would want a pitch that's not close to the hitting plane called a strike. But a fastball hitting the same area would be an acceptable since it wouldn't have the same downward vertical movement.

The last problem he saw with it is simply the human ability to focus on the game when the machine is doing the work for you. When there's a technical glitch and the system doesn't work, there's no way an umpire is going to be locked into the zone and able to make a good call when he's used to just relaying the call the computer made. You could argue that one missed call during a glitch is better than a dozen missed calls under the current system, but it needs to work way more reliably than PitchCom for that one glitch not to turn into a whole bunch.

I'm still in favor of expanding the use in the minors to see how it all works out, but I don't think they're very close to ready to roll it out in the majors right now.
 

Heating up in the bullpen

Member
SoSH Member
Nov 24, 2007
700
Pittsboro NC
In the meantime, MLB should make ball/strike calls challengeable. A blown called strike three can have as much impact on the game as any play out in the field. Give managers one or two challenges per game, and if they're right they get to keep the challenge (with some maximum to keep the requests to important moments). Some may argue more challenges would bog down the game. But I think ball/strike calls could be reviewed much more quickly than tag plays; we all know pretty much instantly when a call is wrong. And I doubt managers would use them frivolously. Try it and tweak it if need be. (And f*^k the umpires if they don't like it.)
 

Bergs

Member
SoSH Member
Jul 22, 2005
18,233
Former umpire Dale Scott has been making the podcast rounds lately and had a few issues with the automatic zone.

First is that it changes for each batter. The box we see on TV isn't really calibrated for each hitter and their stance. The umpire reports for each game come a day or so after they finish because someone goes through and manually sets the top and bottom of the zone for each hitter.

Second was the trickiness of a 3D zone and how it relates to what everyone would like to be a strike. A curveball that knicks the back top of the zone doesn't look like a strike to anyone. Neither team would want a pitch that's not close to the hitting plane called a strike. But a fastball hitting the same area would be an acceptable since it wouldn't have the same downward vertical movement.

The last problem he saw with it is simply the human ability to focus on the game when the machine is doing the work for you. When there's a technical glitch and the system doesn't work, there's no way an umpire is going to be locked into the zone and able to make a good call when he's used to just relaying the call the computer made. You could argue that one missed call during a glitch is better than a dozen missed calls under the current system, but it needs to work way more reliably than PitchCom for that one glitch not to turn into a whole bunch.

I'm still in favor of expanding the use in the minors to see how it all works out, but I don't think they're very close to ready to roll it out in the majors right now.
Ignoring Dale Scott's obvious biases for a moment: Everything he says can be true, and yet robot umps will still be better than what we have now. As for the bolded...I mean, what a complete and uttter abdication of responsibility. "I can't do my job if I'm not 100% locked in...waahhhhhhhh"

And who knows, sitting back there watching pitches get accurately called in real time might actually train the the umpires to be BETTER in the event there is a "glitch."
 

OfTheCarmen

Cow Humper
SoSH Member
Jul 18, 2007
4,276
That was actually my thought as well. There's nothing stopping them from being there and being "in the zone" and just making the calls the computer makes in case it doesnt. And you have to imagine there will be plenty of "i would have called that a strike, huh..." and vice versa.
 

Max Power

thai good. you like shirt?
SoSH Member
Jul 20, 2005
5,982
Boston, MA
In this case it would be the league telling them to abdicate their responsibility to the machine. Then asking them to pretend they're still making the call, even when they're not, just in case they have to. I can't think of anyone who would be as focused on a task when they're not supposed to complete it if everything goes right. It's just human nature.
 

trs

Well-Known Member
Silver Supporter
SoSH Member
Aug 19, 2010
201
Madrid
In the meantime, MLB should make ball/strike calls challengeable. A blown called strike three can have as much impact on the game as any play out in the field. Give managers one or two challenges per game, and if they're right they get to keep the challenge (with some maximum to keep the requests to important moments). Some may argue more challenges would bog down the game. But I think ball/strike calls could be reviewed much more quickly than tag plays; we all know pretty much instantly when a call is wrong. And I doubt managers would use them frivolously. Try it and tweak it if need be. (And f*^k the umpires if they don't like it.)
I oscillate between this and no changes whatsoever. Yes, it makes sense that when we see on TV that a call was blatantly missed by the width of a few baseballs and the call leads the inning ending or perhaps a fundamentally different outcome in the game, that coaches should be able to challenge just like they would an out call at first base or a foul ball.

That being said, the definition of the strike zone is inherently much more vague -- midpoints of shoulders and tops of pants along with bottoms of knees and a ball passing across a pentagramish-rectangle shape. The height of the zone changes based not only on the batter but the batter's stance -- all these things we know. That the "strike zone" is so obviously nebulous and hard to delineate at the corners, suggests to me that it was never designed to be measured with laser accuracy. I'm not sure that taking a ball 4 inches off the plate or 3 inches above the midpoint between your shoulders and your belt for a walk to win the game was part of the spirit of the strike zone. This all leads me to what every little league coach tells every hitter with two strikes -- expand your zone. Part of that was because Sully the Ump has lost half his water weight squatting behind a "catcher" getting pegged with passed balls and is doing everything humanly possible to get the game over, but part of it also was because the strike zone does get fuzzy around the edges, and that's part of the design.

Again, I know this doesn't really fix anything, and it roils me just as much as anyone else when a 3-1 fastball six inches off the plate is called strike two or vice-versa an 0-2 curve is called ball one, but for me the issue is how the strike zone is defined. However, there doesn't seem to be any easy way to fix the height of the strike zone so that it's constantly visible and adjusted to the size and stance and of each hitter -- and of course not adjusting it, just setting it arbitrarily and constantly for every hitter, seems unfair.

So... do nothing?
 

Heating up in the bullpen

Member
SoSH Member
Nov 24, 2007
700
Pittsboro NC
I oscillate between this and no changes whatsoever. Yes, it makes sense that when we see on TV that a call was blatantly missed by the width of a few baseballs and the call leads the inning ending or perhaps a fundamentally different outcome in the game, that coaches should be able to challenge just like they would an out call at first base or a foul ball.

That being said, the definition of the strike zone is inherently much more vague -- midpoints of shoulders and tops of pants along with bottoms of knees and a ball passing across a pentagramish-rectangle shape. The height of the zone changes based not only on the batter but the batter's stance -- all these things we know. That the "strike zone" is so obviously nebulous and hard to delineate at the corners, suggests to me that it was never designed to be measured with laser accuracy. I'm not sure that taking a ball 4 inches off the plate or 3 inches above the midpoint between your shoulders and your belt for a walk to win the game was part of the spirit of the strike zone. This all leads me to what every little league coach tells every hitter with two strikes -- expand your zone. Part of that was because Sully the Ump has lost half his water weight squatting behind a "catcher" getting pegged with passed balls and is doing everything humanly possible to get the game over, but part of it also was because the strike zone does get fuzzy around the edges, and that's part of the design.

Again, I know this doesn't really fix anything, and it roils me just as much as anyone else when a 3-1 fastball six inches off the plate is called strike two or vice-versa an 0-2 curve is called ball one, but for me the issue is how the strike zone is defined. However, there doesn't seem to be any easy way to fix the height of the strike zone so that it's constantly visible and adjusted to the size and stance and of each hitter -- and of course not adjusting it, just setting it arbitrarily and constantly for every hitter, seems unfair.

So... do nothing?
Re: "strike zone ... was never designed to be measured with laser accuracy." If you follow that logic, how much of today's game is as was designed? The game wasn't designed to have a runner placed on second base to start the 10th inning. It wasn't designed to have parades of relief pitchers. It wasn't designed to have instant replay. The game changes with technology, with culture, for money, for safety. The people running and playing the game today get to decide how to do it. If they (we) want a more accurate strike zone, bowing to original design should be the last thing to stand in the way.
Another thing: if they had lasers to measure things back then, I'll bet they would have used laser accuracy as the standard for the strike zone. They did what they could with the technology available. Now our technology allows us to instantly know the speed of a pitch, it's spin rate -- it's goddamn spin rate, for chrissake! -- exit velocity, launch angle, and so on. Technology allows us to slo-mo and stop action and see (almost) for sure if a tag was made before the base was touched, if a ball was trapped or caught in the outfield, if a line drive down the line is foul or fair. We've got so much data about every aspect of the game and the players. Would it be so hard to have on file the upper and lower strike zone measurements for every MLB player, and for those numbers to pop up in the Robo strikezone system when that player comes to bat (or in the interim, applied to the strike zone box we all see watching on TV or on GameDay)? No, it wouldn't.
So... no, don't do nothing. Do what the game has always done. Do what's possible with the current technology. And do it now.
 

Red(s)HawksFan

Member
SoSH Member
Jan 23, 2009
17,030
Maine
The Pacific Coast League is testing a robo-strike zone system and last night was its debut. The umpire has an earpiece signaling ball or strike and he then makes the call as if he was doing it on his own. Here's a twitter thread from a radio announcer who called a game with it in use.

View: https://twitter.com/Josh_Suchon/status/1526934642514747392


Last night was the debut of the Automated Ball Strike (ABS) system in the Pacific Coast League. First thought: if you didn't know the umpire had an earpiece telling him ball/strike, it's doubtful you would know. Here's a thread:
The home plate umpire, Brennan Miller, seemed to inform everyone of calls as quickly as he would if was making the decision. (Miller told me he previously used the ABS in the Arizona Fall League.) I was never awkwardly waiting to get the call and relay that on the broadcast.
Most importantly, no pitches outrageously out of the strike zone were called "strike" and no pitches down the middle were called a ball. No pitches bounced into the strike zone that were called a "strike." It was honestly uneventful in that regard.
Were players upset at borderline calls? Of course. That's normal. This is Alan Trejo's at-bat in the 9th inning. The second pitch clipped the strike zone by millimeters and was called a strike. I instinctively called it a ball on the air, and had to correct myself.

The system is designed to call a strike if any part of the baseball touches the strike zone. I'll be curious if this gets tweaked in the future. By the way, the plate is 17 inches wide. An extra inch was added to each side, so the "strike zone" was 19 inches last night.
Each team was provided a tablet in the dugout to see exactly what the computer is reading and be able to make adjustments. I heard the tablet in the Topes dugout was a few pitches behind. I don't know if that was by design, or the entire game, or a hiccup that needs correcting.
Catchers still instinctively move their gloves to try "framing" the pitch. This made me smirk. They've been doing that their entire life. It'll take awhile for those habits to change. It's still good practice if they get called up to the majors.
I'm reserving my judgment until I see a larger sample size, from multiple cities and umpire crews. For those who were at a PCL game last night -- media, fans, scouts, or watching on MiLB.TV -- what did you think?
This is the official ABS Reference Card that we were provided. This is posted in most dugouts and clubhouses.
With the caveat that all systems are going to be buggy here and there, this seems like it is workable.
 

Max Power

thai good. you like shirt?
SoSH Member
Jul 20, 2005
5,982
Boston, MA
Interesting that they went with a 2D zone instead of a 3D one. The visuals on Altuve and Judge make the top of the zone seem a little too low, but maybe that's correct when you're pushing it back to the center of the plate versus the front.

If you're going to give even a fraction of the ball crossing the zone as a strike, then it should be the width of the plate and not any larger. A hitter would effectively have to cover the width of the plate, plus 2 inches, plus a ball's width, which doesn't seem fair.
 

trs

Well-Known Member
Silver Supporter
SoSH Member
Aug 19, 2010
201
Madrid
Re: "strike zone ... was never designed to be measured with laser accuracy." If you follow that logic, how much of today's game is as was designed? The game wasn't designed to have a runner placed on second base to start the 10th inning. It wasn't designed to have parades of relief pitchers. It wasn't designed to have instant replay. The game changes with technology, with culture, for money, for safety. The people running and playing the game today get to decide how to do it. If they (we) want a more accurate strike zone, bowing to original design should be the last thing to stand in the way.
Another thing: if they had lasers to measure things back then, I'll bet they would have used laser accuracy as the standard for the strike zone. They did what they could with the technology available. Now our technology allows us to instantly know the speed of a pitch, it's spin rate -- it's goddamn spin rate, for chrissake! -- exit velocity, launch angle, and so on. Technology allows us to slo-mo and stop action and see (almost) for sure if a tag was made before the base was touched, if a ball was trapped or caught in the outfield, if a line drive down the line is foul or fair. We've got so much data about every aspect of the game and the players. Would it be so hard to have on file the upper and lower strike zone measurements for every MLB player, and for those numbers to pop up in the Robo strikezone system when that player comes to bat (or in the interim, applied to the strike zone box we all see watching on TV or on GameDay)? No, it wouldn't.
So... no, don't do nothing. Do what the game has always done. Do what's possible with the current technology. And do it now.
Oh I'm pretty sure I agree with everything you've written here. I certainly don't think that MLB should ignore technological advances or do absolutely nothing to make the game better. My somewhat belabored point was that given the actual language of the definition of the strike zone, referring to parts of the body that can't really be pinpointed with that language (midpoints between shoulders and tops of uniform pants (don't hike them up before each pitch), that unlike other rules that are much clearer in distinguishing one response or another, such as did the ball touch the player before or after the player touched the bag, the strike zone has some wiggle-room baked in.

Of course we can change the definition of the strike zone to make it easier to measure, lazers or no, because it's certainly been changed many times over the years.

Anyway, I'm glad alternatives are being tested out as in the PCL, and I'm curious to see what the outcomes are!