×
  • remind me tomorrow
  • remind me next week
  • never remind me
Subscribe to the ANN Newsletter • Wake up every Sunday to a curated list of ANN's most interesting posts of the week. read more

Forum - View topic
EP. REVIEW: Megalobox


Goto page Previous  1, 2, 3, 4, 5, 6, 7, 8, 9  Next

Note: this is the discussion thread for this article

Anime News Network Forum Index -> Site-related -> Talkback
View previous topic :: View next topic  
Author Message
Gina Szanboti



Joined: 03 Aug 2008
Posts: 11306
PostPosted: Sun Jun 03, 2018 8:43 am Reply with quote
NeverConvex wrote:
This is missing the point, though. We already predict systems in which we know there are limits to our ability to forecast; this is what forecasting the expected behavior of probabilistic systems is all about.

The deeper issue is whether there is any room for 'free will' inside of the unpredictable variance in human behavior...

Actually, that is the point in the context of the discussion. Meiam produced a formula that "showed" that Joe couldn't win. I said that you couldn't predict sporting events with perfect accuracy by mathematical calculations, and meiam said we'll be able to in ~20 years. I maintain that it's not possible with human players because there are too many variables.

Yes, we predict things about very, very complicated systems, but we're not always right because we a) don't know all the variables at play, and b) because we can't know precisely how they will all interact at any moment even if we did know them. Maybe in 1000 years with dedicated supercomputers receiving nanosecond by nanosecond telemetry of each player's physical and mental state and data on every divot on the field and every molecule of dust in the playing area, etc., etc., this could be calculated with near perfect accuracy (but it wouldn't be much of a prediction for all that, since it would be predicting in nearly real time), but as long as people and not machines are playing, predictions will never be perfect and there will still be times that Joe wins, despite any math that says it should be impossible (and yeah, sometimes he would lose fights he should easily win, even if he weren't throwing them).

Any given Sunday, y'know? Smile "...one half a step too late or too early and you don't quite make it. One half second too slow, too fast and you don't quite catch it. The inches we need are everywhere around us. They're in every break of the game, every minute, every second." How can algorithms ever hope to perfectly predict that? Sometimes Joe's going to win.

If I'm wrong though, and we ever are able to predict outcomes with such certainty, it's the end of professional sports, since if the underdog never won, they would be pointless and boring as hell.

Yes, free will is one intrinsic part of the unpredictable variance in human behavior. I don't know what you mean by asking if there's room for it.
Back to top
View user's profile Send private message
NeverConvex
Subscriber



Joined: 08 Jun 2013
Posts: 2267
PostPosted: Sun Jun 03, 2018 11:09 am Reply with quote
From what I could tell, meiam didn't produce any such formula, just made some off-hand remarks about Joe "mathematically losing," whatever that's intended to mean. But your suppositions about people being special remain somewhat goofy; given perfect input data on a person's neural state, for example, there is no reason we couldn't predict what they will do. In fact folks have already carried out experiments to this effect.
Back to top
View user's profile Send private message
Gina Szanboti



Joined: 03 Aug 2008
Posts: 11306
PostPosted: Mon Jun 04, 2018 1:11 am Reply with quote
Because dozens of parameters are in constant flux. As I said, the best you might do with a complete data set of human and environmental factors is predict nano- or microseconds ahead, which is pretty useless as a prediction, unless you're a quantum physicist.

I think the experiments you're referring to only predict intent, not success. They can read some brain activity that indicates someone will look in a certain direction before the person has consciously decided to do so, but that can't predict if their eyes will focus on a particular thing or whether the A/C kicks on at that moment and their bangs fall over their eyes and block their vision. Smile

meiam's formula was "Joe's opponent: Work hard, have mechanical gear that enhance there strength and defense
Joe: Work hard

See how Joe is just mathematically inferior to his opponent. Yet he still win."

It's not much of a formula, but it still counts as "(Value A + Value B) > Value A ∴ Joe loses." It fails to predict the outcome because it fails to factor in all the relevant variables, not because the series is wrong about it being possible for Joe to win.
Back to top
View user's profile Send private message
CrowLia



Joined: 24 Feb 2012
Posts: 5500
Location: Mexico
PostPosted: Mon Jun 04, 2018 1:54 am Reply with quote
I don't know if this has been brought up before, but I'm puzzled about how there were no ethical/sportsmanship related controversies regarding Ace within the show's universe. I mean, if the computer is predicting you opponent's moves, is it even a sport anymore? With the gear you can argue that, although people with better tech are at an advantage, the same can be said about real life sports. But once it's a computer making the moves, not only does it become an unfair advantage, the whole point of the sport -a test of peak human physical prowess- becomes completely moot
Back to top
View user's profile Send private message My Anime My Manga
Gina Szanboti



Joined: 03 Aug 2008
Posts: 11306
PostPosted: Mon Jun 04, 2018 2:40 am Reply with quote
I think the reason it hasn't been brought up is that the show has failed (see, I can admit it's not perfect Smile) to define what Gear is supposed to do, beyond "enhance" a boxer's performance. We don't know what that means, or if people in this world think there's a limit on it besides what the boxer's body can tolerate. Likewise, it's not been made clear why Yuri's Gear is so awesome, other than it's been implanted in his body like cybernetics. What is the advantage to that? We don't know. ¯\_(ツ)_/¯
Back to top
View user's profile Send private message
NeverConvex
Subscriber



Joined: 08 Jun 2013
Posts: 2267
PostPosted: Mon Jun 04, 2018 4:25 pm Reply with quote
I think the ambiguity over what gear is supposed to do wasn't a mistake, but was part of the point of the Joe vs Mikio bout, and of Mikio's story as a whole. Adding robo-gear into the mix at all already calls into question what exactly the Megalobox matches are testing; use of Ace to actively propose approximately optimal strategies is just taking that process one step further, displacing not only physical strength & stamina but also much of the boxer's need to calculate and plan with the use of machine-learning supplements. Joe vs Mikio's bout therefore forces into view more directly the question: where does the machine end and the man begin, and what exactly is it these contests measure & celebrate?

I also think the show largely came down on the side of validating (via Yuri's commentary) Mikio's performance exactly because he chose not to rely fully on Ace. I think if Joe v Mikio had ended with Mikio never questioning Ace's predictions that Yuri would not have spoken up to Yukiko in Mikio's defense.

Gina wrote:
Because dozens of parameters are in constant flux. As I said, the best you might do with a complete data set of human and environmental factors is predict nano- or microseconds ahead, which is pretty useless as a prediction, unless you're a quantum physicist.


A large number of parameters being present isn't necessary or sufficient to make a system's behavior hard to predict. A better criterion is something like Warren Weaver's notion of organized complexity, which describes problems for which stochastic or unknown factors cannot simply be averaged out: "All these are certainly complex problems, but they are not problems of disorganized complexity, to which statistical methods hold the key. They are all problems which involve dealing simultaneously with a sizable number of factors which are interrelated into an organic whole."

There are certainly some aspects of human behavior that are like this, and other that aren't at all like this (e.g. there are some very good models for predicting average frequencies of choice by humans in simple lever-pressing psychological experiments). Predicting the next punch a boxer is going to throw a second or so ahead of time doesn't seem like it's likely to fall into the very difficult category for prediction-making, though. Given a suite of high-fidelity sensors reading your opponent's muscle contractions, the pattern of biochemical firings within & between their neurons, and both fighters' relative fighting postures over time, it seems like it should be perfectly feasible to predict an opponent boxer's next move with very high fidelity. If we're willing to posit high-fidelity sensors operating in real time (as the show has), it's hard to see why there would be any major barriers left to this kind of prediction; it's not like a boxer can easily twist his/her arm out of position at the last fraction of a second --- a boxer's moves would be largely determined well in advance of signal receipt.
Back to top
View user's profile Send private message
Gina Szanboti



Joined: 03 Aug 2008
Posts: 11306
PostPosted: Mon Jun 04, 2018 8:15 pm Reply with quote
I feel like you're shifting the goalposts on me here a bit (and I've bought into it along the way). My original response was to this statement: "people are trying to develop computer model to plug in stats to determine outcome." I understood this to mean the final outcome, predicted ahead of the match, rather than the result of any particular blow or play in real time. Since then I've been thinking in terms of sports in general, not just boxing. And I certainly haven't been thinking of only Gear boxing. Smile

I will agree with your latest post, but it still seems a useless endeavor, since you're positing AI Gear vs AI Gear, where the boxer is just a scaffold and transportation for the Gear. The only limiting variable seems to be the boxer's stamina and the ability of their body to withstand the stress of the sync, so I imagine you could make some pretty good predictions on the outcome of a fight based on that. Basically, you've got AIs swapping perfectly aimed and perfectly defended blows until someone's body gives out, or the crowd gets bored and goes home and they call it a draw.

So I don't think you can convince me that infallible predictions can be made on the final outcomes of most sports ahead of the game, no matter how good the data going in might be. And I remain doubtful of the utility of in-game real-time predictions as far as making any money off of it, and if it doesn't pay, why bother? Smile

Btw, I'm not saying that you could never correctly predict the outcome of a sporting event, just that you can never always correctly predict the outcome of a sporting event as long as people are the ones playing. I can flip a coin and correctly predict the outcomes of many events. You might even achieve 95% correct predictions. But no matter how good your algorithms or your data, you'll never get to 100%, which is why sometimes Joe will win.
Back to top
View user's profile Send private message
NeverConvex
Subscriber



Joined: 08 Jun 2013
Posts: 2267
PostPosted: Mon Jun 04, 2018 8:29 pm Reply with quote
Fair enough. I was mostly focused on evaluating how plausible a thing Ace is, given what Megalobox seems to hypothesize about its sensors and so on; I don't remember where I initially replied to you, but if that's not what you had in mind then I guess I did get us off-track.

I don't think we know enough to determine whether boxing matches can be perfectly predicted before the match has begun. In principle that kind of prediction should be possible, assuming boxing matches are driven by purely classical (in the physics sense) phenomena, but there's plenty of speculation about quantum phenomena and its relevance to peoples' decision-making, and I'm no physicist, so hell if I'm going to pretend I have any authoritative knowledge there.

In practice, it's just very hard to say without trying, and how successful we can be will depend on how much information we're willing to pretend is available to the hypothetical AI. But if we're literally talking about where the best predictive models of boxing match outcomes will be in 20 years, I would guess that they'll modestly outperform most human commentators while still leaving substantial remaining variation that they'll fail to predict. I think that'll be due to two factors: A) we're not likely to ever consistently measure enough dimensions of boxers & boxing matches to ever get enough data to drive a truly 'big data' approach to modeling it. And B) boxing matches can behave in 'path dependent' fashion (or so I assume; I'm not a boxing fan), where small variations in whether a punch lands can be incredibly important to the outcome but very hard to gather enough information to predict ahead of time. These kinds of high-influence almost-a-coin-flip events are unlikely to ever be overcome by real-world analytics when predictions are required far ahead of being able to measure the most important events; for practical purposes they appear as if they have a large, truly random component.
Back to top
View user's profile Send private message
Gina Szanboti



Joined: 03 Aug 2008
Posts: 11306
PostPosted: Fri Jun 08, 2018 2:15 am Reply with quote
I'm thinking the situation with Sachio's father is going to be their way out. I might be mistaken, but I was under the impression that the Shirato Group had him killed, not just that he died after his work was stolen. If that's the case, then it doesn't seem unlikely that Fujimaki might've carried out the contract, in which case, he might be the one pushing up daisies courtesy of Yukiko by the time this is over. On the other hand, I'd be kinda surprised if it was that easy. I suppose even easier would be if she paid off Nambu's debt, but I don't see that happening.

Anyway, I'm glad we finally learned what that was about. I noticed him staring at Sachio back when, and figured he might be his long lost dad or something. Close enough.
Back to top
View user's profile Send private message
NeverConvex
Subscriber



Joined: 08 Jun 2013
Posts: 2267
PostPosted: Fri Jun 08, 2018 10:50 am Reply with quote
Hm - maybe. I'm beginning to wonder whether we may not get a happy ending after all -- that Joe and Nanbu may one or both be dead by the end of this journey.
Back to top
View user's profile Send private message
jdnation



Joined: 15 May 2007
Posts: 1994
PostPosted: Fri Jun 08, 2018 5:18 pm Reply with quote
Even if a computer is punching for you, I guess the point is still that one must have the physical stamina to keep up with it.

There's a trade-off.

The AI Gear can predict your opponent's moves with some degree of accuracy, but the user can't pace themselves or cancel a move.

Versus the normal gear where like a boxer, you still have control over your own body and can make your own decisions and pace yourself properly. There is also the human instinct, which computers cannot solve, and as Joe showed, it was flawed and trick-able merely by letting down one's guard.

In that respect, the AI gear is inefficient and not something for the average boxer.

As to the point of the gear at all. I suppose it was just to up the spectacle. And also it seems as of the latest episode, the military is involved, so it seems that they are using essentially the consumer/entertainment sports industry to test their wares. We do not know much about the world of Megalobox, except that there does seem to be a war going on, or recently ended.

So if the entire Megalobox industry was built as a cost effective way to develop fighting gear technology for the battlefield, it would make sense, akin to how it is speculated that the American Government provided seed funding to places like Facebook and cell phone companies to develop technology that would be both of use to consumers as well as provide the means of data, tracking, surveillance and social engineering.

I think that is a subtext we will discover in the latter episodes.
Back to top
View user's profile Send private message
Gina Szanboti



Joined: 03 Aug 2008
Posts: 11306
PostPosted: Fri Jun 08, 2018 6:06 pm Reply with quote
NeverConvex wrote:
Hm - maybe. I'm beginning to wonder whether we may not get a happy ending after all

Well, it's not like its predecessor got one, exactly. I have no idea which way the writers will jump this time, but All Dead Now and Never Die both seem like plausible titles. Smile
Back to top
View user's profile Send private message
Engineering Nerd



Joined: 24 Apr 2008
Posts: 897
Location: Southern California
PostPosted: Fri Jun 15, 2018 1:10 pm Reply with quote
Just want to make sure, Megalo Box has 13 episodes instead of 12 right? I sure hope the final fight is a 2-parter, it deserves that much of screen time.
Back to top
View user's profile Send private message
meiam



Joined: 23 Jun 2013
Posts: 3442
PostPosted: Fri Jun 15, 2018 1:16 pm Reply with quote
musouka wrote:
meiam wrote:
Joe goes up against Yuri, the champion (once again the champion went and fought against a no name fighters for no other reason that said no name fighters was the main character) using gear. He lose. So what does he do then? He remove his gear... and start winning. The gear were literally just dragging him down. And of course within 3 month he's already way past Yuri (who need to use gear just to keep up with Joe in there inevitable fight). Cause, hey what's a lifetime of training (couple with an entire company R&D division) and slow grinding up as a fighter without publicity gimmick next to Joe raw talent?


So, you literally think that if Joe had removed his gear with his first fight with Yuri, then he would have won that fight in which he got KO'd in the first round? Like, the gear was some sort of limiter and if Joe had just ripped it off his arms from the very beginning he would have won all those fights (that he deliberately threw)?

Come on, man. Your criticism in this respect is downright incoherent. Correlation is not causation, and if you got the impression that the series was trying to tell you that it was...well, you are literally the only person I've come across that thinks Joe's gear was some anchor keeping him tethered to the bottom.


Laughing Laughing Laughing Laughing
Back to top
View user's profile Send private message
musouka



Joined: 09 Sep 2003
Posts: 705
PostPosted: Fri Jun 15, 2018 2:20 pm Reply with quote
Man, I really loved this episode, flaws and all. It actually made me cry at the end, and I don't really consider myself to be particularly emotional when it comes to anime.

I loved Sachio's flashback to Joe telling him what he does to ground himself when he's scared, especially accompanied by watching Joe run his fingers against the wall. (I'm pretty sure we've seen Joe make the motion more than once in prior episodes, but I haven't had time to go back and check.) Between that and watching Joe go into the ring in his shitty old gear, knowing what that said about what he planned to do, and how he must have felt about it...yep, this show sure has its hooks sunk into me.

The moment that got to me was Nanbu shouting for Joe to stand up, though. Nanbu has always been the strongest voice in telling Joe to stay down, to give up, to play his role--so it was absolutely electrifying to see him throw all caution to the wind and fully become a coach to Joe, even if it meant giving up his eyesight. (The pulling out his eye and offering it up as repayment works better as an emotional beat than a logical one, but I really do prefer it to Nanbu dying.)

Overall just an excellent show. I'm completely in love--hell, I might even buy the Japanese release.
Back to top
View user's profile Send private message
Display posts from previous:   
Reply to topic    Anime News Network Forum Index -> Site-related -> Talkback All times are GMT - 5 Hours
Goto page Previous  1, 2, 3, 4, 5, 6, 7, 8, 9  Next
Page 7 of 9

 


Powered by phpBB © 2001, 2005 phpBB Group