Artificial Intelligence and Wrestling

[ILLINIWrestlingBlog]

Let's say you have a quantum computer that approaches AI, and it has "watched" every wrestling video on the internet, has self-taught comprehensive knowledge and understanding of human anatomy, body mechanics, kinesiology and physics. Couldn't that algorithm develop new wrestling moves? Or, have we, like an infinite number of monkeys at an infinite number of typewriters discovered every tie and hold that there is to discover?

On another subject, I understand that there are services available that will provide wrestlers with scouting reports on future opponents. These will tell you what kind of shots your foe likes to take, his or her favorite ties, etc. How granular does it go? How granular will it go?

My guess is that someday it will be shockingly granular. 

Take, for example, a bar arm and half-nelson combo. With knowledge of the opponent's physical frame and apparent strength, and an understanding of universal human biometrics, as well as the special abilities of your adversary, the machine will be able to provide graphic illustrations on where to apply the pressure, how to apply it, and even when to apply it given your opponent's average stamina rating. 

The graphic would show with exact precision the angles of your elbow in the half, where exactly to sink the bar arm, and the perfect angle for you to get maximum rotational force with your legs, arms and torso given your size and body shape and the size and body shape of your opponent. 

I wondered how soon we might see this, so I conducted an experiment. 

As the biggest fanboy on this site (although still an impartial journalist), I asked various AI programs to provide renderings of the ILLINI coaching staff. As you can see from the images below and from what I'm about to tell you, the output was variously interesting, strange, awful, pretty cool and at other times silly. 

In the rendering of Coach Poeta below, the AI decided that he should be amputated below the right knee. In IMAR's rendition, the AI program decided that he should be depicted in a Cubist way, the ref in Modern Realism (though having an Infinity Stone on his forehead kind of ruins that effect), while the coaches in the background look to be done in Water Color. The ILLINI coach is wearing a scarf, which looks very gallant! Additionally, for some reason the program decided to put a Coke and a hot dog on the mat. 

As for the other ILLINI coaches, Jeremy Hunter and Austin O'Connor, every single render that I attempted turned one of them into a battle droid and the other into Taylor Swift. I gave up. 

In conclusion, I don't think wrestling coaches and assistants have anything to worry about in the foreseeable future. Their art and science will be needed. Additionally, my guess is that wrestling scouting reports will continue to be a thing and get incrementally better. All we have to worry about at the present is bad AI wrestling art. 

 

[ILLINIWrestlingBlog]

Oh, I forgot to mention headbands and wristbands. According to all of the free AI programs that I've used, every athlete in the world wears either wristbands or headbands or both. See the image of Coach Medlin above. It's like they're all Rafael Nadals or Bjorn Borgs. Also, each program will do its damndest to make people look like Taylor Swift or Anime or battle droids. 

Have they been programmed to provide honest output or what they "think" people want? 

[ionel]

Once Poeta goes full computer geek and starts using the power of Blue Waters for this, he will be the next Cael.  No one will be able to catch up.  Instead of The Professor they will simply call him The Geek.  

Edited October 18, 2023 by ionel

[jackwebster]

The monkey + typewriters + time = Hamlet bit is built on the assumption that the universe and time will keep going indefinitely. But, the universe won't keeping going. It will end at some point. More importantly for our Shakespeare monkeys, it will end before a single act is written.

[alex1fly]

13 hours ago, ILLINIWrestlingBlog said:

Let's say you have a quantum computer that approaches AI, and it has "watched" every wrestling video on the internet, has self-taught comprehensive knowledge and understanding of human anatomy, body mechanics, kinesiology and physics. Couldn't that algorithm develop new wrestling moves? Or, have we, like an infinite number of monkeys at an infinite number of typewriters discovered every tie and hold that there is to discover?

On another subject, I understand that there are services available that will provide wrestlers with scouting reports on future opponents. These will tell you what kind of shots your foe likes to take, his or her favorite ties, etc. How granular does it go? How granular will it go?

My guess is that someday it will be shockingly granular. 

Take, for example, a bar arm and half-nelson combo. With knowledge of the opponent's physical frame and apparent strength, and an understanding of universal human biometrics, as well as the special abilities of your adversary, the machine will be able to provide graphic illustrations on where to apply the pressure, how to apply it, and even when to apply it given your opponent's average stamina rating. 

The graphic would show with exact precision the angles of your elbow in the half, where exactly to sink the bar arm, and the perfect angle for you to get maximum rotational force with your legs, arms and torso given your size and body shape and the size and body shape of your opponent. 

I wondered how soon we might see this, so I conducted an experiment. 

As the biggest fanboy on this site (although still an impartial journalist), I asked various AI programs to provide renderings of the ILLINI coaching staff. As you can see from the images below and from what I'm about to tell you, the output was variously interesting, strange, awful, pretty cool and at other times silly. 

In the rendering of Coach Poeta below, the AI decided that he should be amputated below the right knee. In IMAR's rendition, the AI program decided that he should be depicted in a Cubist way, the ref in Modern Realism (though having an Infinity Stone on his forehead kind of ruins that effect), while the coaches in the background look to be done in Water Color. The ILLINI coach is wearing a scarf, which looks very gallant! Additionally, for some reason the program decided to put a Coke and a hot dog on the mat. 

As for the other ILLINI coaches, Jeremy Hunter and Austin O'Connor, every single render that I attempted turned one of them into a battle droid and the other into Taylor Swift. I gave up. 

In conclusion, I don't think wrestling coaches and assistants have anything to worry about in the foreseeable future. Their art and science will be needed. Additionally, my guess is that wrestling scouting reports will continue to be a thing and get incrementally better. All we have to worry about at the present is bad AI wrestling art. 

 

I think you're on to something re: new moves and opponent analysis.

[Wrestleknownothing]

The Athletic has a story today about AI and football that is worth a read.

[Plasmodium]

T

17 hours ago, ILLINIWrestlingBlog said:

Let's say you have a quantum computer that approaches AI, and it has "watched" every wrestling video on the internet, has self-taught comprehensive knowledge and understanding of human anatomy, body mechanics, kinesiology and physics. Couldn't that algorithm develop new wrestling moves? Or, have we, like an infinite number of monkeys at an infinite number of typewriters discovered every tie and hold that there is to discover?

On another subject, I understand that there are services available that will provide wrestlers with scouting reports on future opponents. These will tell you what kind of shots your foe likes to take, his or her favorite ties, etc. How granular does it go? How granular will it go?

My guess is that someday it will be shockingly granular. 

Take, for example, a bar arm and half-nelson combo. With knowledge of the opponent's physical frame and apparent strength, and an understanding of universal human biometrics, as well as the special abilities of your adversary, the machine will be able to provide graphic illustrations on where to apply the pressure, how to apply it, and even when to apply it given your opponent's average stamina rating. 

The graphic would show with exact precision the angles of your elbow in the half, where exactly to sink the bar arm, and the perfect angle for you to get maximum rotational force with your legs, arms and torso given your size and body shape and the size and body shape of your opponent. 

I wondered how soon we might see this, so I conducted an experiment. 

As the biggest fanboy on this site (although still an impartial journalist), I asked various AI programs to provide renderings of the ILLINI coaching staff. As you can see from the images below and from what I'm about to tell you, the output was variously interesting, strange, awful, pretty cool and at other times silly. 

In the rendering of Coach Poeta below, the AI decided that he should be amputated below the right knee. In IMAR's rendition, the AI program decided that he should be depicted in a Cubist way, the ref in Modern Realism (though having an Infinity Stone on his forehead kind of ruins that effect), while the coaches in the background look to be done in Water Color. The ILLINI coach is wearing a scarf, which looks very gallant! Additionally, for some reason the program decided to put a Coke and a hot dog on the mat. 

As for the other ILLINI coaches, Jeremy Hunter and Austin O'Connor, every single render that I attempted turned one of them into a battle droid and the other into Taylor Swift. I gave up. 

In conclusion, I don't think wrestling coaches and assistants have anything to worry about in the foreseeable future. Their art and science will be needed. Additionally, my guess is that wrestling scouting reports will continue to be a thing and get incrementally better. All we have to worry about at the present is bad AI wrestling art. 

 

Sometimes while walking or biking I pencil out a simulation using 3d models to develop or improve wrestling techniques.  The simulation would have to be really good.  It is the minute details that separate success from failure.

[ILLINIWrestlingBlog]

45 minutes ago, Wrestleknownothing said:

The Athletic has a story today about AI and football that is worth a read.

Here's the article, which was free for me to read. Thanks! 

That is a very niche application: Predicting blitzes. But based on it, you can envision a Wrestling AI being able to predict a wrestler's takedown attempt after viewing and analyzing that wrestler's movements in dozens or hundreds of videos. 

In college, a Freshman wrestler with immediate success gets to hear, "Wait until he gets scouted." With the stockpile of video from high school at Flo and youtube and other services, nobody will have to wait once AI gets involved. 

 

[ILLINIWrestlingBlog]

9 hours ago, jackwebster said:

The monkey + typewriters + time = Hamlet bit is built on the assumption that the universe and time will keep going indefinitely. But, the universe won't keeping going. It will end at some point. More importantly for our Shakespeare monkeys, it will end before a single act is written.

"It was the best of times, it was the blurst of times. You stupid monkey!" 

 

[Wrestleknownothing]

4 minutes ago, ILLINIWrestlingBlog said:

Here's the article, which was free for me to read. Thanks! 

That is a very niche application: Predicting blitzes. But based on it, you can envision a Wrestling AI being able to predict a wrestler's takedown attempt after viewing and analyzing that wrestler's movements in dozens or hundreds of videos. 

In college, a Freshman wrestler with immediate success gets to hear, "Wait until he gets scouted." With the stockpile of video from high school at Flo and youtube and other services, nobody will have to wait once AI gets involved. 

 

Indeed. Very niche. But that is how it starts. Define a single problem to solve. Fail a whole bunch of times, before you finally succeed (if you succeed). Take what you learn there and apply to another problem. Do it with enough problems and you might have more generalized solutions, too.

[Wrestleknownothing]

9 hours ago, jackwebster said:

The monkey + typewriters + time = Hamlet bit is built on the assumption that the universe and time will keep going indefinitely. But, the universe won't keeping going. It will end at some point. More importantly for our Shakespeare monkeys, it will end before a single act is written.

Why do the monkeys always get it after tons of attempts? If it is random why not on the first attempt? I think it is time to acquire some monkeys and some typewriters. I worry that it may be harder to get the typewriters.

[MPhillips]

6 hours ago, Plasmodium said:

Sometimes while walking or biking I pencil out a simulation using 3d models to develop or improve wrestling techniques.

Me too...

[Leg rider]

I feel like things like most effective arm angle and such like have been suggested on here would be tough to really work, way to much fluidity in positioning to make that useful it seems to me

[ILLINIWrestlingBlog]

7 hours ago, Leg rider said:

I feel like things like most effective arm angle and such like have been suggested on here would be tough to really work, way to much fluidity in positioning to make that useful it seems to me

This is a valid point. 

Would it work? I think in the example that I gave, using the half and arm bar, once the initial holds are obtained, and you use the AI suggested torque with your legs and torso, it wouldn't matter what the other person did. Moreover, even if you failed, at least you gave yourself the best chance to succeed. 

That's the bottom line for me. It may not work. But doesn't every wrestler want to know the absolute best way to get the takedown? To get the pin? To escape? Especially if that absolute best way was computationally tailored to your and your opponent's physical attributes and abilities. 

Also, I would think AI in the future would suggest only holds that it can logically foresee you being able to accomplish. That will likely be way in the future, though. 

Here's an AI render of a photograph of Coach Poeta on a recruiting visit with Seth Mendoza. I thought this one looked fantastic: 

[jackwebster]

20 hours ago, Wrestleknownothing said:

Why do the monkeys always get it after tons of attempts? If it is random why not on the first attempt? I think it is time to acquire some monkeys and some typewriters. I worry that it may be harder to get the 

 As a betting man, it's wiser to push all in on the monkeys' first attempt.

Edited October 19, 2023 by jackwebster

[CHROMEBIRD]

Will wrestling ever be Moneyballed? AI could be used to analyze tape and match data, then generate gameplans for attacks and counters, what to watch out for, best positions to choose. That already happens today, it's just a very manual and arduous process.

[Wrestleknownothing]

1 hour ago, CHROMEBIRD said:

Will wrestling ever be Moneyballed? AI could be used to analyze tape and match data, then generate gameplans for attacks and counters, what to watch out for, best positions to choose. That already happens today, it's just a very manual and arduous process.

I think you can make a strong case that PSU has already moneyballed wrestling.

Moneyball was about finding mis-priced attributes, i.e. inefficiencies, in baseball players so that you can buy valuable players on the cheap. If I recall correctly the big one for them was on base percentage.

I think PSU is very good at identifying inefficiencies and figuring out how to exploit them.

A technique example would be the ankle ride that drives every non-PSU fan on this board to distraction. Now that the rule has changed I would not be surprised if the coaching staff at PSU finds another technique or advantage to replace it.

An off the mat example would be their early adoption of a sports psychologist like Bonnie (I do not know her last name). Many wrestlers speak glowingly of how she has helped them.

[KCMO2]

But can AI take into account batsh!t crazy moves or behavior like a Ben Askren, or DeSanto, or Paulie Robbins (a "special" kid on my hs team) might do?  I think not.

[Wrestleknownothing]

10 minutes ago, KCMO2 said:

But can AI take into account batsh!t crazy moves or behavior like a Ben Askren, or DeSanto, or Paulie Robbins (a "special" kid on my hs team) might do?  I think not.

Agreed. AI is a big data operation. One of one wrestlers would be an enigma.

[GreatWhiteNorth]

On 10/18/2023 at 5:27 AM, jackwebster said:

The monkey + typewriters + time = Hamlet bit is built on the assumption that the universe and time will keep going indefinitely. But, the universe won't keeping going. It will end at some point. More importantly for our Shakespeare monkeys, it will end before a single act is written.

One flaw in your logic is that there is no reason to believe the universe won't keep going. No evidence to support it.

[GreatWhiteNorth]

On 10/18/2023 at 2:52 PM, Wrestleknownothing said:

Why do the monkeys always get it after tons of attempts? If it is random why not on the first attempt? I think it is time to acquire some monkeys and some typewriters. I worry that it may be harder to get the typewriters.

You're right. Truly random means some monkeys may get it on only a few attempts, maybe even the first.

Your second point. Sadly, as most monkey experiments go - the monkey's skull is cut open and the brain is connected electronically to the test equipment. That way the unreliable interface between monkey hands and typewriter is bypassed (and also the unreliable typewriters in general.) Just digital.

This could have been the original idea behind the Planet of the Apes.

 

[jackwebster]

4 hours ago, GreatWhiteNorth said:

One flaw in your logic is that there is no reason to believe the universe won't keep going. No evidence to support it.

Maybe, maybe not. So, I'll hedge and say that the Earth and its monkeys and typewriters and people to read Hamlet won't last long enough to see the first dirty joke 

[ILLINIWrestlingBlog]

13 hours ago, CHROMEBIRD said:

Will wrestling ever be Moneyballed? AI could be used to analyze tape and match data, then generate gameplans for attacks and counters, what to watch out for, best positions to choose. That already happens today, it's just a very manual and arduous process.

That's what I'm wondering, CHROMEBIRD.

I have done Moneyball looks at Zane Richards' opponents before, as well as Zac Braunagel's. That was a very manual and arduous process going through videos. Stopping and starting a hundred times for each video. As you note, there is a service that is available that looks very labor intensive, using humans to catalog what the wrestlers are doing, but it also looks like it provides valuable data for coaches. 

According to one company, AI is capable at this time to do some work "watching" video:

"AI can not only observe data but can also be trained with large volumes of video footage to detect, identify, categorize and automatically tag specific objects. Overall, it is a tool used to assist humans in understanding video content, and make automated decisions based on observations made from the data collected."

What is left unclear in my mind is what "objects" are being detected. Facial recognition software can map faces, but can it learn to follow movements? How would it handle a Nashon Garrett/Jesse Delgado scramble? Someday, I think. 

Again, I turn to AI art renderings. 

What the program I used on IMAR's photo did was in a number of cases, inexplicable. Below is the AI render next to Tony Rotundo's original photo. IMAR's right foot is wearing Sideshow Bob-size shoes. His right hand has a finger bending at an impossible angle. Then, there's the scarf worn by the ILLINOIS coach. A nice touch, but it makes zero sense. We can see now why the AI put a Coke and a hot dog on the mat: The red ankle band left on the mat magically became a refreshing snack! Strangest of all, the ref has the green Infinity Time Stone on his forehead. There is NOTHING in the original photo to suggest that the ref should be wearing the Eye of Agamotto! 

If AI can't get still photographs right, I don't think it's ready for videos. 

[RYou]

On 10/18/2023 at 11:42 AM, alex1fly said:

I think you're on to something re: new moves and opponent analysis.

Wouldn't true Ai offer your opponent the same detail about you ? And, given your use of trends and tactics and how to attack your opponent, wouldn't AI inform your opponent of how you will be attack and how to counter what is likely to be expected ?

It seems to me, that with both using similar strategic programming we would be saddled with an inordinate number of OT's, 1-0, and 2-1 bouts except for those where one opponent is clearly dominant.  NCAA championships would become boring.

[CHROMEBIRD]

20 hours ago, KCMO2 said:

But can AI take into account batsh!t crazy moves or behavior like a Ben Askren, or DeSanto, or Paulie Robbins (a "special" kid on my hs team) might do?  I think not.

Imo that's what AI and analytics are especially good at. No human is completely random. Even seasoned poker players will have tendencies and "tells", conscious or subconscious, that become evident with enough data. Desanto might wrestle a batshit crazy style but I think an AI would quickly pick up on his tricep pinch, reaching over to counter his shot, and who knows what else. I would love to see what would come out of an analysis of Askren or Welch's matches, like they tend to favor a side or have greater success scrambling through certain positions and situations.