|
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
The claim is that
p(fix|line)
is high. Using Bayes' rule, we have that
p(fix|line) * p(line) = p(line|fix) * p(fix),
or
p(fix|line) = p(line|fix) * p(fix) / p(line).
Essentially, what statistics tells us is that the probability we are arguing about is derived from three other probabilities. Let's look at each in turn.
The probability that a given Starcraft match is fixed, p(fix). One could narrow this down to the probability that a given MK match is fixed, but that requires arguing that MK is particularly susceptible or immune to match fixing, which I don't think is a productive road to take. Anyway, the key point is that you should be more suspicious of MK if you think that match fixing is rampant in Starcraft; likewise, should you believe that a Starcraft player fixing a match is astronomically unlikely, it makes sense that this prior probability is able to outweigh the immense unlikelihood of the betting lines for this match.
The probability that the betting lines become this skewed in a hypothetical fixed match, p(line|fix). This is where people wonder how a match fixer could be dumb enough to bet enough money on a match that the betting company voids all bets on the match. The answer "doesn't matter, obviously they are that dumb" is insufficient and circular; it assumes that a fixed match is the most likely explanation for the betting line, which is the very thing we are debating. If, for instance, you believe that an elaborate troll is more likely to produce such lines than a genuinely fixed match, you argue that p(line|troll) is so much higher than p(line|fix) that it outweighs any counterbalancing difference between p(troll) and p(fix). This doesn't seem like an illogical position to take, especially if you believe that p(troll) is as high as p(fix), which goes back to the previous paragraph.
The probability, in general, that betting lines are skewed this far, p(line). This is where "I don't know much about betting, but..." comes into play. A lot of smart people have spent a lot of time in these threads trying to explain just how beyond comprehension it is that lines would skew this far on their own. I don't wish to take anything away from those arguments, but I think that too many people have used them to argue for MK's guilt without acknowledging the rest of the picture. When weighing explanations, e.g., p(fix|line) vs. p(troll|line), we can see from the equation that we're dividing by p(line) in both cases. If we want to determine the most likely explanation for the evidence before us, the likelihood of the line skew by itself is totally irrelevant. What matters is the likelihood of a given event skewing the betting line this much, and the probability of such an event occurring. That's what the math tells us. And most importantly, for all intents and purposes these probabilities are unknown, and therefore open to debate.
So yeah, a lot of people here are defending MK based on an incomplete understanding of probability. At the same time though, a number of self-appointed correctors of public ignorance are using pretty strong language while failing to give credit to the degree of statistical uncertainty surrounding the issue. There are idiots on both sides, and some of them are incurable. Ignore those, and do your best to approach the discussion from an empirical and unbiased standpoint.
|
|
|
On April 03 2015 17:55 ICT wrote:
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
The claim is that
p(fix|line)
is high. Using Bayes' rule, we have that
p(fix|line) * p(line) = p(line|fix) * p(fix),
or
p(fix|line) = p(line|fix) * p(fix) / p(line).
Essentially, what statistics tells us is that the probability we are arguing about is derived from three other probabilities. Let's look at each in turn.
The probability that a given Starcraft match is fixed, p(fix). One could narrow this down to the probability that a given MK match is fixed, but that requires arguing that MK is particularly susceptible or immune to match fixing, which I don't think is a productive road to take. Anyway, the key point is that you should be more suspicious of MK if you think that match fixing is rampant in Starcraft; likewise, should you believe that a Starcraft player fixing a match is astronomically unlikely, it makes sense that this prior probability is able to outweigh the immense unlikelihood of the betting lines for this match.
The probability that the betting lines become this skewed in a hypothetical fixed match, p(line|fix). This is where people wonder how a match fixer could be dumb enough to bet enough money on a match that the betting company voids all bets on the match. The answer "doesn't matter, obviously they are that dumb" is insufficient and circular; it assumes that a fixed match is the most likely explanation for the betting line, which is the very thing we are debating. If, for instance, you believe that an elaborate troll is more likely to produce such lines than a genuinely fixed match, you argue that p(line|troll) is so much higher than p(line|fix) that it outweighs any counterbalancing difference between p(troll) and p(fix). This doesn't seem like an illogical position to take, especially if you believe that p(troll) is as high as p(fix), which goes back to the previous paragraph.
The probability, in general, that betting lines are skewed this far, p(line). This is where "I don't know much about betting, but..." comes into play. A lot of smart people have spent a lot of time in these threads trying to explain just how beyond comprehension it is that lines would skew this far on their own. I don't wish to take anything away from those arguments, but I think that too many people have used them to argue for MK's guilt without acknowledging the rest of the picture. When weighing explanations, e.g., p(fix|line) vs. p(troll|line), we can see from the equation that we're dividing by p(line) in both cases. If we want to determine the most likely explanation for the evidence before us, the likelihood of the line skew by itself is totally irrelevant. What matters is the likelihood of a given event skewing the betting line this much, and the probability of such an event occurring. That's what the math tells us. And most importantly, for all intents and purposes these probabilities are unknown, and therefore open to debate.
So yeah, a lot of people here are defending MK based on an incomplete understanding of probability. At the same time though, a number of self-appointed correctors of public ignorance are using pretty strong language while failing to give credit to the degree of statistical uncertainty surrounding the issue. There are idiots on both sides, and some of them are incurable. Ignore those, and do your best to approach the discussion from an empirical and unbiased standpoint.
The visual evidence is only "polarizing" in the sense that some people are willing to be silly enough to defend MarineKing. The visual evidence combined with the betting lines make the probability of matchfixing very high, despite the comparatively insignificant unknowns.
|
On April 04 2015 00:10 Doodsmack wrote:Show nested quote +On April 03 2015 17:55 ICT wrote:
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
The claim is that
p(fix|line)
is high. Using Bayes' rule, we have that
p(fix|line) * p(line) = p(line|fix) * p(fix),
or
p(fix|line) = p(line|fix) * p(fix) / p(line).
Essentially, what statistics tells us is that the probability we are arguing about is derived from three other probabilities. Let's look at each in turn.
The probability that a given Starcraft match is fixed, p(fix). One could narrow this down to the probability that a given MK match is fixed, but that requires arguing that MK is particularly susceptible or immune to match fixing, which I don't think is a productive road to take. Anyway, the key point is that you should be more suspicious of MK if you think that match fixing is rampant in Starcraft; likewise, should you believe that a Starcraft player fixing a match is astronomically unlikely, it makes sense that this prior probability is able to outweigh the immense unlikelihood of the betting lines for this match.
The probability that the betting lines become this skewed in a hypothetical fixed match, p(line|fix). This is where people wonder how a match fixer could be dumb enough to bet enough money on a match that the betting company voids all bets on the match. The answer "doesn't matter, obviously they are that dumb" is insufficient and circular; it assumes that a fixed match is the most likely explanation for the betting line, which is the very thing we are debating. If, for instance, you believe that an elaborate troll is more likely to produce such lines than a genuinely fixed match, you argue that p(line|troll) is so much higher than p(line|fix) that it outweighs any counterbalancing difference between p(troll) and p(fix). This doesn't seem like an illogical position to take, especially if you believe that p(troll) is as high as p(fix), which goes back to the previous paragraph.
The probability, in general, that betting lines are skewed this far, p(line). This is where "I don't know much about betting, but..." comes into play. A lot of smart people have spent a lot of time in these threads trying to explain just how beyond comprehension it is that lines would skew this far on their own. I don't wish to take anything away from those arguments, but I think that too many people have used them to argue for MK's guilt without acknowledging the rest of the picture. When weighing explanations, e.g., p(fix|line) vs. p(troll|line), we can see from the equation that we're dividing by p(line) in both cases. If we want to determine the most likely explanation for the evidence before us, the likelihood of the line skew by itself is totally irrelevant. What matters is the likelihood of a given event skewing the betting line this much, and the probability of such an event occurring. That's what the math tells us. And most importantly, for all intents and purposes these probabilities are unknown, and therefore open to debate.
So yeah, a lot of people here are defending MK based on an incomplete understanding of probability. At the same time though, a number of self-appointed correctors of public ignorance are using pretty strong language while failing to give credit to the degree of statistical uncertainty surrounding the issue. There are idiots on both sides, and some of them are incurable. Ignore those, and do your best to approach the discussion from an empirical and unbiased standpoint. The visual evidence is only "polarizing" in the sense that some people are willing to be silly enough to defend MarineKing. The visual evidence combined with the betting lines make the probability of matchfixing very high, despite the comparatively insignificant unknowns.
Yeah, it's polarizing because there are people who don't draw the same conclusions as you. That's how the whole polarizing thing works.
|
On April 04 2015 01:14 blackone wrote:Show nested quote +On April 04 2015 00:10 Doodsmack wrote:On April 03 2015 17:55 ICT wrote:
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
The claim is that
p(fix|line)
is high. Using Bayes' rule, we have that
p(fix|line) * p(line) = p(line|fix) * p(fix),
or
p(fix|line) = p(line|fix) * p(fix) / p(line).
Essentially, what statistics tells us is that the probability we are arguing about is derived from three other probabilities. Let's look at each in turn.
The probability that a given Starcraft match is fixed, p(fix). One could narrow this down to the probability that a given MK match is fixed, but that requires arguing that MK is particularly susceptible or immune to match fixing, which I don't think is a productive road to take. Anyway, the key point is that you should be more suspicious of MK if you think that match fixing is rampant in Starcraft; likewise, should you believe that a Starcraft player fixing a match is astronomically unlikely, it makes sense that this prior probability is able to outweigh the immense unlikelihood of the betting lines for this match.
The probability that the betting lines become this skewed in a hypothetical fixed match, p(line|fix). This is where people wonder how a match fixer could be dumb enough to bet enough money on a match that the betting company voids all bets on the match. The answer "doesn't matter, obviously they are that dumb" is insufficient and circular; it assumes that a fixed match is the most likely explanation for the betting line, which is the very thing we are debating. If, for instance, you believe that an elaborate troll is more likely to produce such lines than a genuinely fixed match, you argue that p(line|troll) is so much higher than p(line|fix) that it outweighs any counterbalancing difference between p(troll) and p(fix). This doesn't seem like an illogical position to take, especially if you believe that p(troll) is as high as p(fix), which goes back to the previous paragraph.
The probability, in general, that betting lines are skewed this far, p(line). This is where "I don't know much about betting, but..." comes into play. A lot of smart people have spent a lot of time in these threads trying to explain just how beyond comprehension it is that lines would skew this far on their own. I don't wish to take anything away from those arguments, but I think that too many people have used them to argue for MK's guilt without acknowledging the rest of the picture. When weighing explanations, e.g., p(fix|line) vs. p(troll|line), we can see from the equation that we're dividing by p(line) in both cases. If we want to determine the most likely explanation for the evidence before us, the likelihood of the line skew by itself is totally irrelevant. What matters is the likelihood of a given event skewing the betting line this much, and the probability of such an event occurring. That's what the math tells us. And most importantly, for all intents and purposes these probabilities are unknown, and therefore open to debate.
So yeah, a lot of people here are defending MK based on an incomplete understanding of probability. At the same time though, a number of self-appointed correctors of public ignorance are using pretty strong language while failing to give credit to the degree of statistical uncertainty surrounding the issue. There are idiots on both sides, and some of them are incurable. Ignore those, and do your best to approach the discussion from an empirical and unbiased standpoint. The visual evidence is only "polarizing" in the sense that some people are willing to be silly enough to defend MarineKing. The visual evidence combined with the betting lines make the probability of matchfixing very high, despite the comparatively insignificant unknowns. Yeah, it's polarizing because there are people who don't draw the same conclusions as you. That's how the whole polarizing thing works.
Hey if you want to believe that he would go 3 cc and not defend himself after seeing a proxy hatch in front of his nose, I guess there's no helping you.
|
On April 04 2015 01:14 blackone wrote:Show nested quote +On April 04 2015 00:10 Doodsmack wrote:On April 03 2015 17:55 ICT wrote:
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
The claim is that
p(fix|line)
is high. Using Bayes' rule, we have that
p(fix|line) * p(line) = p(line|fix) * p(fix),
or
p(fix|line) = p(line|fix) * p(fix) / p(line).
Essentially, what statistics tells us is that the probability we are arguing about is derived from three other probabilities. Let's look at each in turn.
The probability that a given Starcraft match is fixed, p(fix). One could narrow this down to the probability that a given MK match is fixed, but that requires arguing that MK is particularly susceptible or immune to match fixing, which I don't think is a productive road to take. Anyway, the key point is that you should be more suspicious of MK if you think that match fixing is rampant in Starcraft; likewise, should you believe that a Starcraft player fixing a match is astronomically unlikely, it makes sense that this prior probability is able to outweigh the immense unlikelihood of the betting lines for this match.
The probability that the betting lines become this skewed in a hypothetical fixed match, p(line|fix). This is where people wonder how a match fixer could be dumb enough to bet enough money on a match that the betting company voids all bets on the match. The answer "doesn't matter, obviously they are that dumb" is insufficient and circular; it assumes that a fixed match is the most likely explanation for the betting line, which is the very thing we are debating. If, for instance, you believe that an elaborate troll is more likely to produce such lines than a genuinely fixed match, you argue that p(line|troll) is so much higher than p(line|fix) that it outweighs any counterbalancing difference between p(troll) and p(fix). This doesn't seem like an illogical position to take, especially if you believe that p(troll) is as high as p(fix), which goes back to the previous paragraph.
The probability, in general, that betting lines are skewed this far, p(line). This is where "I don't know much about betting, but..." comes into play. A lot of smart people have spent a lot of time in these threads trying to explain just how beyond comprehension it is that lines would skew this far on their own. I don't wish to take anything away from those arguments, but I think that too many people have used them to argue for MK's guilt without acknowledging the rest of the picture. When weighing explanations, e.g., p(fix|line) vs. p(troll|line), we can see from the equation that we're dividing by p(line) in both cases. If we want to determine the most likely explanation for the evidence before us, the likelihood of the line skew by itself is totally irrelevant. What matters is the likelihood of a given event skewing the betting line this much, and the probability of such an event occurring. That's what the math tells us. And most importantly, for all intents and purposes these probabilities are unknown, and therefore open to debate.
So yeah, a lot of people here are defending MK based on an incomplete understanding of probability. At the same time though, a number of self-appointed correctors of public ignorance are using pretty strong language while failing to give credit to the degree of statistical uncertainty surrounding the issue. There are idiots on both sides, and some of them are incurable. Ignore those, and do your best to approach the discussion from an empirical and unbiased standpoint. The visual evidence is only "polarizing" in the sense that some people are willing to be silly enough to defend MarineKing. The visual evidence combined with the betting lines make the probability of matchfixing very high, despite the comparatively insignificant unknowns. Yeah, it's polarizing because there are people who don't draw the same conclusions as you. That's how the whole polarizing thing works.
You have the odds that are almost evidence by their own.
You have the game where MKP ignores everything and seems to do every desicion he can to lose the game
You can consider MKP has not been fielded since that game, probably for a reason..
|
On April 04 2015 02:58 Glorfindel! wrote:Show nested quote +On April 04 2015 01:14 blackone wrote:On April 04 2015 00:10 Doodsmack wrote:On April 03 2015 17:55 ICT wrote:
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
The claim is that
p(fix|line)
is high. Using Bayes' rule, we have that
p(fix|line) * p(line) = p(line|fix) * p(fix),
or
p(fix|line) = p(line|fix) * p(fix) / p(line).
Essentially, what statistics tells us is that the probability we are arguing about is derived from three other probabilities. Let's look at each in turn.
The probability that a given Starcraft match is fixed, p(fix). One could narrow this down to the probability that a given MK match is fixed, but that requires arguing that MK is particularly susceptible or immune to match fixing, which I don't think is a productive road to take. Anyway, the key point is that you should be more suspicious of MK if you think that match fixing is rampant in Starcraft; likewise, should you believe that a Starcraft player fixing a match is astronomically unlikely, it makes sense that this prior probability is able to outweigh the immense unlikelihood of the betting lines for this match.
The probability that the betting lines become this skewed in a hypothetical fixed match, p(line|fix). This is where people wonder how a match fixer could be dumb enough to bet enough money on a match that the betting company voids all bets on the match. The answer "doesn't matter, obviously they are that dumb" is insufficient and circular; it assumes that a fixed match is the most likely explanation for the betting line, which is the very thing we are debating. If, for instance, you believe that an elaborate troll is more likely to produce such lines than a genuinely fixed match, you argue that p(line|troll) is so much higher than p(line|fix) that it outweighs any counterbalancing difference between p(troll) and p(fix). This doesn't seem like an illogical position to take, especially if you believe that p(troll) is as high as p(fix), which goes back to the previous paragraph.
The probability, in general, that betting lines are skewed this far, p(line). This is where "I don't know much about betting, but..." comes into play. A lot of smart people have spent a lot of time in these threads trying to explain just how beyond comprehension it is that lines would skew this far on their own. I don't wish to take anything away from those arguments, but I think that too many people have used them to argue for MK's guilt without acknowledging the rest of the picture. When weighing explanations, e.g., p(fix|line) vs. p(troll|line), we can see from the equation that we're dividing by p(line) in both cases. If we want to determine the most likely explanation for the evidence before us, the likelihood of the line skew by itself is totally irrelevant. What matters is the likelihood of a given event skewing the betting line this much, and the probability of such an event occurring. That's what the math tells us. And most importantly, for all intents and purposes these probabilities are unknown, and therefore open to debate.
So yeah, a lot of people here are defending MK based on an incomplete understanding of probability. At the same time though, a number of self-appointed correctors of public ignorance are using pretty strong language while failing to give credit to the degree of statistical uncertainty surrounding the issue. There are idiots on both sides, and some of them are incurable. Ignore those, and do your best to approach the discussion from an empirical and unbiased standpoint. The visual evidence is only "polarizing" in the sense that some people are willing to be silly enough to defend MarineKing. The visual evidence combined with the betting lines make the probability of matchfixing very high, despite the comparatively insignificant unknowns. Yeah, it's polarizing because there are people who don't draw the same conclusions as you. That's how the whole polarizing thing works. You have the odds that are almost evidence by their own. You have the game where MKP ignores everything and seems to do every desicion he can to lose the game You can consider MKP has not been fielded since that game, probably for a reason..
The odds of match fixing are high but you can't say it's so high he's match fixing, until there's conclusive proof (replay showing his screen seeing the spine) there's no reason to hunt him down. Doesn't wolf have the replay or something.
|
I would be shocked if Wolf had a replay. That doesn't seem possible given how the cast is set up / organization.
Plus if he had it and didn't say anything this entire time, that just doesn't make sense to me either given that he's previously commented on this controversy.
|
On April 04 2015 03:17 Wuster wrote:
I would be shocked if Wolf had a replay. That doesn't seem possible given how the cast is set up / organization.
Plus if he had it and didn't say anything this entire time, that just doesn't make sense to me either given that he's previously commented on this controversy.
I saw them cast a game I think it was proleague but it could've been SSL and at the end moongladeau left the game came up meaning sometime or all the time they're using their accounts for casting. Maybe not with proleague but I'm pretty sure Wolf could get that replay?
|
On April 04 2015 03:25 Ovid wrote:Show nested quote +On April 04 2015 03:17 Wuster wrote:
I would be shocked if Wolf had a replay. That doesn't seem possible given how the cast is set up / organization.
Plus if he had it and didn't say anything this entire time, that just doesn't make sense to me either given that he's previously commented on this controversy. I saw them cast a game I think it was proleague but it could've been SSL and at the end moongladeau left the game came up meaning sometime or all the time they're using their accounts for casting. Maybe not with proleague but I'm pretty sure Wolf could get that replay?
If you login to your account on another computer, you only have access to the replay from that computer no?
I also doubt KeSPA's going to let the casters take them even if they ask given how guarded replays are in S. Korea.
|
On April 04 2015 03:13 Ovid wrote:Show nested quote +On April 04 2015 02:58 Glorfindel! wrote:On April 04 2015 01:14 blackone wrote:On April 04 2015 00:10 Doodsmack wrote:On April 03 2015 17:55 ICT wrote:
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
The claim is that
p(fix|line)
is high. Using Bayes' rule, we have that
p(fix|line) * p(line) = p(line|fix) * p(fix),
or
p(fix|line) = p(line|fix) * p(fix) / p(line).
Essentially, what statistics tells us is that the probability we are arguing about is derived from three other probabilities. Let's look at each in turn.
The probability that a given Starcraft match is fixed, p(fix). One could narrow this down to the probability that a given MK match is fixed, but that requires arguing that MK is particularly susceptible or immune to match fixing, which I don't think is a productive road to take. Anyway, the key point is that you should be more suspicious of MK if you think that match fixing is rampant in Starcraft; likewise, should you believe that a Starcraft player fixing a match is astronomically unlikely, it makes sense that this prior probability is able to outweigh the immense unlikelihood of the betting lines for this match.
The probability that the betting lines become this skewed in a hypothetical fixed match, p(line|fix). This is where people wonder how a match fixer could be dumb enough to bet enough money on a match that the betting company voids all bets on the match. The answer "doesn't matter, obviously they are that dumb" is insufficient and circular; it assumes that a fixed match is the most likely explanation for the betting line, which is the very thing we are debating. If, for instance, you believe that an elaborate troll is more likely to produce such lines than a genuinely fixed match, you argue that p(line|troll) is so much higher than p(line|fix) that it outweighs any counterbalancing difference between p(troll) and p(fix). This doesn't seem like an illogical position to take, especially if you believe that p(troll) is as high as p(fix), which goes back to the previous paragraph.
The probability, in general, that betting lines are skewed this far, p(line). This is where "I don't know much about betting, but..." comes into play. A lot of smart people have spent a lot of time in these threads trying to explain just how beyond comprehension it is that lines would skew this far on their own. I don't wish to take anything away from those arguments, but I think that too many people have used them to argue for MK's guilt without acknowledging the rest of the picture. When weighing explanations, e.g., p(fix|line) vs. p(troll|line), we can see from the equation that we're dividing by p(line) in both cases. If we want to determine the most likely explanation for the evidence before us, the likelihood of the line skew by itself is totally irrelevant. What matters is the likelihood of a given event skewing the betting line this much, and the probability of such an event occurring. That's what the math tells us. And most importantly, for all intents and purposes these probabilities are unknown, and therefore open to debate.
So yeah, a lot of people here are defending MK based on an incomplete understanding of probability. At the same time though, a number of self-appointed correctors of public ignorance are using pretty strong language while failing to give credit to the degree of statistical uncertainty surrounding the issue. There are idiots on both sides, and some of them are incurable. Ignore those, and do your best to approach the discussion from an empirical and unbiased standpoint. The visual evidence is only "polarizing" in the sense that some people are willing to be silly enough to defend MarineKing. The visual evidence combined with the betting lines make the probability of matchfixing very high, despite the comparatively insignificant unknowns. Yeah, it's polarizing because there are people who don't draw the same conclusions as you. That's how the whole polarizing thing works. You have the odds that are almost evidence by their own. You have the game where MKP ignores everything and seems to do every desicion he can to lose the game You can consider MKP has not been fielded since that game, probably for a reason.. The odds of match fixing are high but you can't say it's so high he's match fixing, until there's conclusive proof (replay showing his screen seeing the spine) there's no reason to hunt him down. Doesn't wolf have the replay or something.
Even though he intentionally lost, it doesn't necessarily mean he's involved in matchfixing. It is a possibility that he had said to someone that he didn't want to win, for whatever reason, and this got leaked to someone who took advantage of it.
|
On April 03 2015 17:55 ICT wrote:
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
Sure, if you assume that the a priori probability of any game being fixed is almost zero then this evidence is hardly convincing. Or if you assume that suspicious line movements are actually evidence for FAIR games not FIXED ones then again, there's nothing to worry about.
The problem with this approach is that you can't just chose the value of these probabilities at will. There's plenty of evidence that fixing happens in sports in general and Starcraft in particular. Using your notation you can't claim that P(fix) < 10^-6 for example.
Same goes for the suspicious line movements. There is evidence that match fixers sometimes do mess up. Pinnacle's policy doesn't exist in a void: it is based on previous experience on how rigged matches are bet on. You can argue that sometimes you get false alarms.but to go as far as to claim that fair games are actually more likely to produce suspicious betting patterns is just silly.
There's a level of judgement and guessing involved but that doesn't mean that all guesses are reasonable.
|
On April 04 2015 05:04 cheekymonkey wrote:Show nested quote +On April 04 2015 03:13 Ovid wrote:On April 04 2015 02:58 Glorfindel! wrote:On April 04 2015 01:14 blackone wrote:On April 04 2015 00:10 Doodsmack wrote:On April 03 2015 17:55 ICT wrote:
I've noticed a lot of posts claiming that people who believe MK to be innocent don't understand probability. Without taking sides, I'd like to point out that the degree of confidence we have that match fixing occurred is based on a number of unknowable quantities, and there's plenty of room for subjective interpretation.
To illustrate this simply, let's ignore the visual evidence (which seems to be pretty polarizing anyway) and focus on the betting line movement. When people say "Look at the line, you moron. He's clearly cheating!", what they mean is that given the line movement, the probability that the match is fixed is very high. I'm going to incorporate some statistical notation; if that distresses you, skip over it and I'll do my best to explain what it means.
The claim is that
p(fix|line)
is high. Using Bayes' rule, we have that
p(fix|line) * p(line) = p(line|fix) * p(fix),
or
p(fix|line) = p(line|fix) * p(fix) / p(line).
Essentially, what statistics tells us is that the probability we are arguing about is derived from three other probabilities. Let's look at each in turn.
The probability that a given Starcraft match is fixed, p(fix). One could narrow this down to the probability that a given MK match is fixed, but that requires arguing that MK is particularly susceptible or immune to match fixing, which I don't think is a productive road to take. Anyway, the key point is that you should be more suspicious of MK if you think that match fixing is rampant in Starcraft; likewise, should you believe that a Starcraft player fixing a match is astronomically unlikely, it makes sense that this prior probability is able to outweigh the immense unlikelihood of the betting lines for this match.
The probability that the betting lines become this skewed in a hypothetical fixed match, p(line|fix). This is where people wonder how a match fixer could be dumb enough to bet enough money on a match that the betting company voids all bets on the match. The answer "doesn't matter, obviously they are that dumb" is insufficient and circular; it assumes that a fixed match is the most likely explanation for the betting line, which is the very thing we are debating. If, for instance, you believe that an elaborate troll is more likely to produce such lines than a genuinely fixed match, you argue that p(line|troll) is so much higher than p(line|fix) that it outweighs any counterbalancing difference between p(troll) and p(fix). This doesn't seem like an illogical position to take, especially if you believe that p(troll) is as high as p(fix), which goes back to the previous paragraph.
The probability, in general, that betting lines are skewed this far, p(line). This is where "I don't know much about betting, but..." comes into play. A lot of smart people have spent a lot of time in these threads trying to explain just how beyond comprehension it is that lines would skew this far on their own. I don't wish to take anything away from those arguments, but I think that too many people have used them to argue for MK's guilt without acknowledging the rest of the picture. When weighing explanations, e.g., p(fix|line) vs. p(troll|line), we can see from the equation that we're dividing by p(line) in both cases. If we want to determine the most likely explanation for the evidence before us, the likelihood of the line skew by itself is totally irrelevant. What matters is the likelihood of a given event skewing the betting line this much, and the probability of such an event occurring. That's what the math tells us. And most importantly, for all intents and purposes these probabilities are unknown, and therefore open to debate.
So yeah, a lot of people here are defending MK based on an incomplete understanding of probability. At the same time though, a number of self-appointed correctors of public ignorance are using pretty strong language while failing to give credit to the degree of statistical uncertainty surrounding the issue. There are idiots on both sides, and some of them are incurable. Ignore those, and do your best to approach the discussion from an empirical and unbiased standpoint. The visual evidence is only "polarizing" in the sense that some people are willing to be silly enough to defend MarineKing. The visual evidence combined with the betting lines make the probability of matchfixing very high, despite the comparatively insignificant unknowns. Yeah, it's polarizing because there are people who don't draw the same conclusions as you. That's how the whole polarizing thing works. You have the odds that are almost evidence by their own. You have the game where MKP ignores everything and seems to do every desicion he can to lose the game You can consider MKP has not been fielded since that game, probably for a reason.. The odds of match fixing are high but you can't say it's so high he's match fixing, until there's conclusive proof (replay showing his screen seeing the spine) there's no reason to hunt him down. Doesn't wolf have the replay or something. Even though he intentionally lost, it doesn't necessarily mean he's involved in matchfixing. It is a possibility that he had said to someone that he didn't want to win, for whatever reason, and this got leaked to someone who took advantage of it.
even if that was the case. it's still matchfixing. if he was gonna intentionally lose, for whatever reason, the result of the match was set before it began. even it's not for money, hes still doing a disservice to his fans, his team etc
|
at this point in time you would need evidence he DIDNT match fix to convince me...
thats how blatant it looks to me.
|
the extent of jumping to conclusions done in esports is fucking ludicrous. magicamy doesn't play anymore and now there are 62 pages where any debate is laughed away because Internet Detectives have already decided the case.
because somehow they've made it believable that kespa wouldn't want to investigate this and ban whoever is involved. even though this is ridiculous and known matchfixing would destroy sc2 faster than anything else possibly could.
that's always my favorite part. in all of these things (just like magicamy) there's an included point that makes it so no matter what the people involved say, the internet detectives know they're lying. (with magicamy people wouldn't believe reynad after because somehow he was seen as benefitting by keeping up the lie)
so you have some facts, the conclusions jumped to, and the surrounding opinions that make it impossible to have your conclusions changed. well done!
personally I'll wait for an official response from people who know what they're talking about. if he was matchfixing I have zero doubt he'll be out of sc2 forever.
|
On April 04 2015 07:09 Kevin_Sorbo wrote:
at this point in time you would need evidence he DIDNT match fix to convince me...
thats how blatant it looks to me.
innocent until proven guilty
except on the internet
|
It's 2015. Korean SC2 players have almost nothing to lose. At the time this sort of thing happened in BW it was shocking because the league had a long history and losing your spot in that history had weight to it. Granted the motivations were probably the same, but this time around I imagine most players can see quite clearly that interest in the game is dwindling and that the days of them making money by playing it are numbered. I sometimes wonder how many of them continue playing because they don't know what else to do. MarineKing himself tried to move on but wasn't able to. Sad story all-around.
|
On April 04 2015 10:17 Vari wrote:Show nested quote +On April 04 2015 07:09 Kevin_Sorbo wrote:
at this point in time you would need evidence he DIDNT match fix to convince me...
thats how blatant it looks to me. innocent until proven guilty except on the internet
i think thats fine. most of these people (magicamy, mkp etc) are priveledged and could easily defend themselves/get resourses to help defend them. the innocent till proven helps the noobs who couldnt defend themselves even if they were innocent. prevents corruption which i dont think is the case here.
|
The statement wouldn't take that long if MarineKing was completely clean and if no match fixing had ever taken place that day.
I'll wait a bit before changing my sig though 
|
If Marineking was actually involved in match fixing he wouldn't get paid by the fixers because that was some of the worst and most obvious acting I've ever seen, and the betting got voided anyway.
You need savior level fixing to actually cause some trouble
|
|
|
|
|
|
EPT
