UTK Special 1/31/23
Making The Robots Work
ChatGPT. Dall-E. ABS in AAA.
There’s a lot of alphabet soup regarding AI/ML, but what does that have to do with baseball? More importantly, what could it do for the game in positive ways? I have some ideas, but first we have to take a step back and take a look at something that’s definitely not artificially intelligent - pitch counts.
There’s been a lot of traction from a number of smart people around trying to fix pitching overuse in youth leagues. Josh Rudd recently got a lot of notice looking at MLK Weekend tournaments, which are tough because they’re so early in season, involve some elite teams, and heavy travel to warm weather. He’s developed some AI tools that look at the pitch counts, compare them to the regulations, and what do you know, he’s found a lot of “illegal” pitchers being used.
This isn’t a surprise. Pitchsmart regulations first came in about a decade ago, led by a lot of really smart people like Dr. Glenn Fleisig. I told Glenn at the time that I worried that there’d be a reaction against pitch count rules and rest regulations, and while I can’t say it’s the reason we’ve seen an explosion in travel ball over the last decade, I can say there’s a coincidental overlap.
Work from Deven Morgan at Driveline Academy, from Tracey Hayes at Mobility Chick, and from many others has pushed for a change to this, but the fact is that there’s two limiting factors: laziness and inertia. Simply put, it’s hard to count pitches and in most areas, there’s a mom with a clicker and a scorebook who might miss one here or there, or not know the regulations back and forth. We shouldn’t put yet another thing on a mom who’s not trained.
Inertia is the fact that there’s a lot of people, coaches, and teams having success under this system. They win. They get good players. Those players get drafted or get scholarships and the cycle continues. Travel coaches are making big money in many cases and travel organizations often have hundreds of players across age group teams, all paying into the thousands in fees, entries, uniforms, equipment, and travel.
While following pitching rules and guidelines is a nice first step, there’s always going to be those that try to get around them or just ignore them. If someplace like Perfect Game decides to be the good guys, then one of the other travel organizations is going to use that to their advantage - “Don’t be limited, come to a tournament where starting pitchers rule!”
I’ve been writing about pitcher development for a long time. Look at the date on this article and realize that while everything in that piece still holds true today, no one has ever tried something as simple as that.
Which brings me back to the robots. I think they might save us in this case.
Not exactly like that and these robots probably won’t have a good cigar to share either. The same cameras and machine vision technology that calls balls and strikes could also count pitches. Hawkeye does this at the major leagues and the increasing number of other places, but there are other systems in use. Almost all college programs have something for instant replay and these systems are getting better and better, with better cameras, better zoom, and better tech behind them to do the kind of things that would be needed to get not just pitch counts, but throw counts.
Most programs don’t have multiple locations where they practice. There’s a baseball field, lovingly tended, at the local high school or college. Even at most Division 1 programs, they’re “single site” and the cameras are there. If we could get computer vision to identify players — a transmitter like a Catapult or even a jersey with a readable number — it wouldn’t be hard to get throw counts. Once we get to throw counts, real world workload measures could be automated quickly and easily, saving coaches - and moms - a lot of data input.
Is it perfect? No. Snow and rain days inside the gym throw that off, but something is better than nothing and I am reminded of Aaron Schatz telling me years ago that “the best is the enemy of the better.” We could get better and better quick, especially as camera and vision systems get better and cheaper.
This isn’t that far off. I’ve worked in the sports vision field for a couple years and have seen rapid improvements. We’ve gone from spin rates to seam shifted wake. We’ve gone from markers to markerless to biomechanics from a phone. The concept that we could get smart enough computer vision to count throws and attach events to individuals isn’t a huge leap or a technical challenge. I wouldn’t be surprised if it could be done by current technology; certainly at the top end with Hawkeye and SMT’s FieldFX systems, but likely at the lower end as well, though that may require post-processing.
The issue then is the same as what we have now. Who does the analysis and the enforcement? This is a land where the moms rule, like it or not. If an organization or even a tournament director gets involved, it can get ugly. (You would not believe some of the dad fights I’ve seen.) Coaches don’t want to be seen as snitches and the kids just want to play. All the best intentions are right there, yet somehow get lost when the ump yells play ball, the last thing anyone will agree with him on.
The easy solution then is to let the robots save us. Hard and fast rules enforced by an unblinking, uncaring machine save us from ourselves and those best intentions. A red light, a programmed unwillingness to compromise, and a computer that won’t let the game continue until that ineligible pitcher is replaced isn’t just the best solution, it might be the only one.