What Is A Robbery? Devin Haney - Vasyl Lomachenko Scored By Unbaised Artificial Intelligence.

[Cypocryphy]

I took a look, I'm not going to watch the whole thing but get the premise and don't like it. It's hy over complicated for no reason, we can all see what lands and judge the effectiveness of the punches.

I'm not sure why you wouldn't like it. Clearly people can't judge what lands and what doesn't land because we wouldn't have people crying robbery several times a month. This is currently being used for amateur fights and to great success. Apparently, 80 percent of the results are comparable to the judges, and the 20 percent that are not tend to side with the publics perception of a fight and not the judges.

I went out of my way to find mistakes that the AI made in the Loma fight, and I could only find one (maybe two). One was for Loma in round three that was missed and one was for Devin in round nine, where Devin seemed to land but it wasn't counted. Regardless, many punches that I thought missed actually did land when I slowed the fight down. And some that I thought landed didn't actually land. Many of these boxers are so fast that if you blink you can miss it. With DEEPSTRIKE, it catches 99 percent of the action and accurately reports it.

I can't see how anyone would not want an accurate and fair assessment of a fight. These fighters deserve better, and this is a way to give everyone a fair shake.

By the way, check out the other thread I posted. I have Hagler vs Duran. I don't know if you remember how you scored that fight, but you can see how the AI scored it. You can look at Jabbr's YouTube channel for some other fights, but I thought the Hagler vs Duran would be interesting.

[Spray_resistant]

I'm not sure why you wouldn't like it. Clearly people can't judge what lands and what doesn't land because we wouldn't have people crying robbery several times a month. This is currently being used for amateur fights and to great success. Apparently, 80 percent of the results are comparable to the judges, and the 20 percent that are not tend to side with the publics perception of a fight and not the judges.

I went out of my way to find mistakes that the AI made in the Loma fight, and I could only find one (maybe two). One was for Loma in round three that was missed and one was for Devin in round nine, where Devin seemed to land but it wasn't counted. Regardless, many punches that I thought missed actually did land when I slowed the fight down. And some that I thought landed didn't actually land. Many of these boxers are so fast that if you blink you can miss it. With DEEPSTRIKE, it catches 99 percent of the action and accurately reports it.

I can't see how anyone would not want an accurate and fair assessment of a fight. These fighters deserve better, and this is a way to give everyone a fair shake.

By the way, check out the other thread I posted. I have Hagler vs Duran. I don't know if you remember how you scored that fight, but you can see how the AI scored it. You can look at Jabbr's YouTube channel for some other fights, but I thought the Hagler vs Duran would be interesting.

I would be interested in seeing how it scored clear as day notable robberies, I would prefer fan online scoring to official judges, most people really do want to see the right fighter win. This was a very close fight and how exactly is AI determining effectiveness which is what is key here?

[Cypocryphy]

I would be interested in seeing how it scored clear as day notable robberies, I would prefer fan online scoring to official judges, most people really do want to see the right fighter win. This was a very close fight and how exactly is AI determining effectiveness which is what is key here?

Yeah. That's a good question. I'm going to see if I can do some research and find out what is considered "effectiveness." I'm curious about that too.

[Cypocryphy]

I would be interested in seeing how it scored clear as day notable robberies, I would prefer fan online scoring to official judges, most people really do want to see the right fighter win. This was a very close fight and how exactly is AI determining effectiveness which is what is key here?

Effectiveness falls in the category of the "Aggression" metric. So here is how these other two metics are defined, those being "Pressure" and "Aggression."

Pressure: The pressure metric is measured by assessing forward movement, opponent moving backwards, having opponent in the corner or on ropes, staying in the mid-range or inside. Ultimately, this alludes to "ring generalship."

Aggression: The aggression metric uses indicators such as throwing punches with a high power commit value, throwing combinations, throwing the first and/or last punch in an exchange, etc. This metric alludes to who is being the "effective aggressor."

Both of these metrics have to be considered along with "clean punches" and the quality and quantity of those punches.

[Brother Mouzone]

A robbery to me is one fighter clearly and decisively winning at least 8 of 12 rds and doesn't get the decision. Boxing is subjective and 7-5 leaves too much doubt even if it appears they clearly won. I am discounting kds so lets say they need to be a minimum of 4 points ahead.

In the case of Loma/Haney, neither fighter clearly won more than 1 or 2 rds, it was a close highly contested chess match and I am ok with Haney getting the nod but will respect anyone's opinion that Loma won close.

This ..............I mentioned in the live fight thread that 115-113 for either fighter was fair and thought it would be a majority draw .Was not scoring round for round but did mention there was a ton of swing rounds that were difficult to score, I did give the 11th to Loma and mentioned it was the most cleary won round of the fight. Gave the 12th to Devin barely..

[Spray_resistant]

Effectiveness falls in the category of the "Aggression" metric. So here is how these other two metics are defined, those being "Pressure" and "Aggression."

Pressure: The pressure metric is measured by assessing forward movement, opponent moving backwards, having opponent in the corner or on ropes, staying in the mid-range or inside. Ultimately, this alludes to "ring generalship."

Aggression: The aggression metric uses indicators such as throwing punches with a high power commit value, throwing combinations, throwing the first and/or last punch in an exchange, etc. This metric alludes to who is being the "effective aggressor."

Both of these metrics have to be considered along with "clean punches" and the quality and quantity of those punches.

Effectiveness pertains to how damaged or diminished a fighter is by offense landed on them. How does AI differentiate a punch of the same impact landed on lets say Amir Khan vs GGG? It matters greatly in fights and then we get down to reputable punchers vs featherdusters landing power punches.

[Cypocryphy]

Effectiveness pertains to how damaged or diminished a fighter is by offense landed on them. How does AI differentiate a punch of the same impact landed on let's say Amir Khan vs GGG? It matters greatly in fights and then we get down to reputable punchers vs featherdusters landing power punches.

I'm not entirely clear what you are referring to here. Effectiveness? So you don't mean "effective aggression"? These are all professional boxers, so they all hit hard, more or less. Effectiveness is not explicitly a judging criterium.

You score a fight based on clean punches, effective aggression, ring generalship and defense. There's no judging criterium that says you have to judge based on "effectiveness." That doesn't exist.

Wait ... I kind of see what you mean. So you would say that one of Wilder's punches are far more effective than one of Chocolotitos punches? Is that what you're getting at?

Ultimately, a boxer will gradually lose the ability to defend himself, which means he'll be taking more punches as the fight wears on until one figure is clearly winning, or is dropped or stopped. And if a fighter is dropped, then he automatically loses the round and a point. So the judging criteria already has "effectiveness" assessed by the four criteria, and if one fighter is being more effective, it eventually shows on the other fighter by his taking more punches, slowing down and/or ultimately being dropped or stopped. "Effectiveness" tends to show itself over the entirety of the fight that way.

[Spray_resistant]

I'm not entirely clear what you are referring to here. Effectiveness? So you don't mean "effective aggression"? These are all professional boxers, so they all hit hard, more or less. Effectiveness is not explicitly a judging criterium.

You score a fight based on clean punches, effective aggression, ring generalship and defense. There's no judging criterium that says you have to judge based on "effectiveness." That doesn't exist.

Wait ... I kind of see what you mean. So you would say that one of Wilder's punches are far more effective than one of Chocolotitos punches? Is that what you're getting at?

Ultimately, a boxer will gradually lose the ability to defend himself, which means he'll be taking more punches as the fight wears on until one figure is clearly winning, or is dropped or stopped. And if a fighter is dropped, then he automatically loses the round and a point. So the judging criteria already has "effectiveness" assessed by the four criteria, and if one fighter is being more effective, it eventually shows on the other fighter by his taking more punches, slowing down and/or ultimately being dropped or stopped. "Effectiveness" tends to show itself over the entirety of the fight that way.

Rounds are scored primarily on clean Effective punches, harder punches are more effective than softer ones and weaker chins and fragile skin is more effected by hard punches than someone with a face made of granite.

10 Paulie Mallinaggi flicking jabs = 2 hard Cotto power shots.

How can AI scoring account for the attributes of particular fighters in the way or better than the eye can?

[The D3vil]

Jeez, let it go.

The fight was 115-113 (either way) or a draw by the standards that we judge boxing by today.

In the future A.I. may very well score fights & I support that . . . in the future, when the technology has been thoroughly tested and implemented on lower levels.

They way we judge boxing now, it was not a robbery.in any way by the judges or by Compubox wihich has been around for 40 years now & said Haney outlanded or tied Loma in punches landed in 7 of 12 rounds.

With no knockdowns, that's a fine scorecard.

[Cypocryphy]

Rounds are scored primarily on clean Effective punches, harder punches are more effective than softer ones and weaker chins and fragile skin is more effected by hard punches than someone with a face made of granite.

10 Paulie Mallinaggi flicking jabs = 2 hard Cotto power shots.

How can AI scoring account for the attributes of particular fighters in the way or better than the eye can?

That's why I said that the AI in DeepStrike measures the punches from minimum to maximum, so it goes minimum value (e.g. flicking jab), low value, middle value, high value and maximum value. If you watch the Golovkin and Canelo fight, you can see that it easily separates what is a maximum value strike from what is merely a low value strike. As to the technicalities of how it works, you'd have to ask the creators. But from the fights I've seen, it does very well at assessing the strength or impact of a punch.

If you're saying that a power punch from Paulie is worth far less than a power punch from Cotto, you can't score a fight that way. Regardless as to who hits harder, it will be apparent to the judge or, in this case, the AI because he'll be hurt afterward, dropped or stopped. You can't have boxers get in the ring and because in your opinion one hits harder that every single one of his punches is worth much more than his opponent's. That's what it sounds like you're saying here, and that's not how you judge a fight. If a punch is that much more powerful, then it willl be evident by its effect on the opponent.

But like I said, it already determines how hard a punch is and does so quite accurately. Check out the Canelo and Golovkin fight to very for yourself.