When Hawkeye isn't enough - or is it? [Archive] - MensTennisForums.com

When Hawkeye isn't enough - or is it?

oz_boz
03-05-2007, 08:51 AM
Discussion started in Youzhny def Nadal in Dubai thread, about the ball that was overruled by Hawkeye and won Youzhny the first set (7-5 in tb). I think the subject is worthy of a separate thread.

As the Hawkeye has a margin of error of 3.6 mm (source unknown to me), it can't make 100% safe calls less than this. Apemant came up with an alternative solution: let Hawkeye have three alternatives IN/OUT/UNKNOWN, and replay the point if the UNKNOWN call occurs. Maybe more fair than today's rules.

The problem with this approach is that it can't be used for ordinary line calling without significantly delaying play. Lets assume that an average line judge as a margin for error about 5 cm. Judging from a diagram I found on the web

http://images.google.com/imgres?imgurl=http://www-out.bell-labs.com/project/lucentvision/half.agassi-sampras.final.landing.jpg&imgrefurl=http://www-out.bell-labs.com/project/lucentvision/&h=288&w=358&sz=25&hl=sv&start=5&tbnid=wm5Y8N0izg6KsM:&tbnh=97&tbnw=121&prev=/images%3Fq%3Dagassi%2Bservice%26svnum%3D10%26hl%3D sv%26client%3Dsafari%26rls%3Dsv-se%26sa%3DG

it seems as if there were about 30 1st serves called "in" from Sampras-Agassi TMC final 1999 that landed so close to the lines that they couldn't have been safely called by a human eye.

To replay all these points would have caused significant delay, and there probably were a lot of serves called "out" that would also have caused replay, in addition to that a number of ordinary rally points.

One could of course use the unknown option only when Hawkeye is at hand, since a lot fewer balls would be uncertain, but it would make little sense IMO to use it when the calls are more accurate.

One way to solve it would be to say: if the ball according to measurement techniques were LIKELY in, then it is in. (The ball that started this discussion was more likely in than not.)

That is the present paradigm, and IMO it is enough. Particularly considering that getting favourable line calls is probably roughly equally likely for every player.

Sean.J.S.
03-05-2007, 08:54 AM
Just because it's a machine it doesn't mean it won't make mistakes. It is more accurate than the human eye.

ServeAlready81
03-05-2007, 08:59 AM
I say if it's more accurate that the human eye, then more power to Hawk Eye ;) Especially with serves bombing in a 120-130, I think it's so hard to tell for sure if it's good or not when it's so close to the lines.

The only problem I have with the Hawk Eyes system is the b*tch ass chair umpires who won't overrule and make the players use challenges. WTF are you there for then?

Deivid23
03-05-2007, 09:10 AM
Good attempt at starting an interesting topic, making sense won´t work in here though so be ready to get dumb answers

scarecrows
03-05-2007, 09:16 AM
I say if it's more accurate that the human eye, then more power to Hawk Eye ;) Especially with serves bombing in a 120-130, I think it's so hard to tell for sure if it's good or not when it's so close to the lines.

The only problem I have with the Hawk Eyes system is the b*tch ass chair umpires who won't overrule and make the players use challenges. WTF are you there for then?

but now chair umpires are scared of overruling since the player might challenge and if the challenge is right makes the umpire look like an ass

Adler
03-05-2007, 09:24 AM
Just because it's a machine it doesn't mean it won't make mistakes. It is more accurate than the human eye.
Exactly. Stop bitching hawkeye, even if it makes errors it does it not as often as the human eye

oz_boz
03-05-2007, 09:38 AM
The problem is not about some machine making errors or not, the problem is what to do when a conclusive line call cannot be attained. Replay point or take the possibly wrong call knowing that it is probably right? Or any other alternative?

Deboogle!.
03-05-2007, 03:27 PM
I think adding the "unknown" possibility is an interesting suggestion. If Hawkeye can't determine the accuracy to a certain point, the original call should stand. But I personally don't have a problem with it being the tiniest bit inaccurate because, as you say, it's inaccurate for everyone the same because it's a machine, whereas the human judges are not necessarily inaccurate for everyone the same.

If we lived in a perfect world it would be 100% accurate, but we don't, and it's not. As several have said, it's still better than the human eye. I also agree that it seems to have rendered the umps impotent, but I believe that's up to the ATP/ITF/WTA/whatever to make these umps officiate as if there were no hawkeye.

To me, the most important things are getting it at every tournament and getting it on more match courts. For example, you have a tournament like San Jose where every single singles match is played on one court. No reason not to have Hawkeye there. Many smaller tourneys are played exclusively on 2 courts, etc. Let's get it on more courts. Hopefully someone will come along with a system just as accurate as Hawkeye that will also drive down the price :)

gusman890
03-05-2007, 03:41 PM
A chair umpire should do their job first, since thats their duty to begin with. Its not secret that they have gotten a little bit lazy since Hawk-eye has been introduced, but none the less, that doesnt make the situation 100% perfect.

If a player runs out of challenges, then it would be like it was for the longest time for someone aruging a call.

The only time I see a problem with the Hawk-eye/Replay in general is when the Umpire makes a call to award the point or replay it. Since there have been many examples through out the year that this technology has been used to say wether the ball was in, then the umpire decides they should replay the point given the different circumstances.

^^ thats the only guff I have with this process, other then that, I enjoy it :)

Apemant
03-05-2007, 04:35 PM
I think adding the "unknown" possibility is an interesting suggestion. If Hawkeye can't determine the accuracy to a certain point, the original call should stand.

Can you explain your position on this?

If Hawkeye can't make the call, it means the ball landed somewhere from 3.6mm outside to 3.6mm inside (or 1.8/1.8, I'm not sure what they mean by '3.6 mm error') . And if that's the case, if it is THAT close to the outer side of the line, it means the 'original call' is as worthless as Hawkeye's, as there is no way they got it right with any decent probability. So why keep it? With Hawkeye we can at least estimate the chance of it being wrong, i.e. we can say something like, there is 60% chance it got it right. With linespeople, you can't even estimate that probability; so it's nothing more than a blind guess.

The bottom line is, if even Hawkeye can't be sure, then sure as heck no umpire can be sure either. What do we do then? Maybe it really wasn't such a good idea to begin with. Earlier, the umpire would say his final judgement and everyone had to abide by it, right or wrong. And now all this.

Deivid23
03-05-2007, 05:40 PM
With Hawkeye we can at least estimate the chance of it being wrong, i.e. we can say something like, there is 60% chance it got it right. With linespeople, you can't even estimate that probability; so it's nothing more than a blind guess.

Way WRONG assumption. In such calls, deviation within the instrumentīs precision is random so thereīs no way you can trust it more than a blind personīs call

R.Federer
03-05-2007, 05:53 PM
Discussion started in Youzhny def Nadal in Dubai thread, about the ball that was overruled by Hawkeye and won Youzhny the first set (7-5 in tb). I think the subject is worthy of a separate thread.

As the Hawkeye has a margin of error of 3.6 mm (source unknown to me),

I posted this in the Youzhny roasts thread

The error rate during the actual test was 3.6 millimeters.That higher distribution isn’t even across all the test results but the way the system was set up — and there are technical reasons for this — we have a lower error on the really, really close ones compared to…say….10 centimeters.
(link to interview etc with the creator http://www.menstennisforums.com/showpost.php?p=4963449&postcount=184)

mickymouse
03-05-2007, 05:57 PM
Is this the first time that there's been such a close call? I seem to recall many other balls that require 'zooming in' in order for the mechanism to determine whether it's 'in' or 'out'. In such instances, luck plays a role. Since the tournament has adopted Hawkeye, then decisions should be left to Hawkeye and no matter how close the call is, players should respect the outcome. You win some, you lose some. Hawkeye may not be enough but I still trust it more than the human eye. Moreover, unlike linespeople, Hawkeye will not be intimidated by nasty players who glare you down nor influenced by a player's status or ranking. In that sense, it's completely unbiased.

Rogiman
03-05-2007, 05:59 PM
Engineer Tool23 is working on a new device.

He's going to install Nadal in every court to make the calls :lol:

R.Federer
03-05-2007, 06:04 PM
If it is such a CLOSE call that it lands within Hawkeye's margin of error, well we know Hawkeye is not 100% reliable and we know the linespeople and umpires are not 100% reliable either. Basically, for CLOSE CALLS, it's a coin toss (50-50) whose call is correct.

On such a call, as a player : who are you MORE LIKELY TO TRUST, a human who has POSSIBLY vested interests (maybe your opponent has intimidated them throughout the match, like Capriati did in her 03 USO semi final against Henin) or a machine that is NOT calibrated to call either in your favor or against it?

I think the answer to that is : the unbiased machine.

Deivid23
03-05-2007, 06:06 PM
Engineer Tool23 is working on a new device.

He's going to install Nadal in every court to make the calls :lol:

Both you and your boyfriend Ron have the same :retard: sense of humour, congrats, did u go to a bad circus instead of school? :lol:

Rogiman
03-05-2007, 06:09 PM
Both you and your boyfriend Ron have the same :retard: sense of humour, congrats, did u go to a bad circus instead of school? :lol:
I'm sorry we're not significant physicists like you are :bowdown:

Care to teach us quantum mechanics?

Deivid23
03-05-2007, 06:13 PM
If it is such a CLOSE call that it lands within Hawkeye's margin of error, well we know Hawkeye is not 100% reliable and we know the linespeople and umpires are not 100% reliable either. Basically, for CLOSE CALLS, it's a coin toss (50-50) whose call is correct.

On such a call, as a player : who are you MORE LIKELY TO TRUST, a human who has POSSIBLY vested interests (maybe your opponent has intimidated them throughout the match, like Capriati did in her 03 USO semi final against Henin) or a machine that is NOT calibrated to call either in your favor or against it?

I think the answer to that is : the unbiased machine.


MTF consensus for those calls:

"Let the machine overrule or confirm the call, after all itīs a machine so somehow itīs better than the human eye"


Geniuses at work in here I must say

Deivid23
03-05-2007, 06:15 PM
I'm sorry we're not significant physicists like you are :bowdown:

Care to teach us quantum mechanics?

U should have done your job when younger, but itīs never too late. Go read some books and learn sth :D

Rogiman
03-05-2007, 06:17 PM
U should have done your job when younger, but itīs never too late. Go read some books and learn sth :DWould you send me areference to your latest books, Prof. Schrödinger?

R.Federer
03-05-2007, 06:18 PM
MTF consensus for those calls:

"Let the machine overrule or confirm the call, after all itīs a machine so somehow itīs better than the human eye"

Geniuses at work in here I must say

It is NOT better. If you took the time to read the post you would see that it is 50-50 who is correct on a CLOSE call, for a call within the error rate of the machine.

If it is 50-50 on a close call, but only the machine is always unbiased and the human isn't, give one good reason why they should go with the umpire on that one.

R.Federer
03-05-2007, 06:23 PM
Is this the first time that there's been such a close call? I seem to recall many other balls that require 'zooming in' in order for the mechanism to determine whether it's 'in' or 'out'.

Good point!

No it is definitely NOT the first time there has been a close call. All those people who have never seemed to have a problem with Hawkeye suddenly have grave issues with it because of one call they did not like on one point of one player.

Every player has benefited from Hawkeye, EVEN ON CLOSE CALLS. No one seemed to have a problem with it then. Every player can also get unfavorable overturns from Hawkeye.

It happened on SP to Nadal here which is tough for him, but he has benefited in winning a set in Melbourne against Murray, in the last game of the fourth set as well, which the linesperson called out, the ump overruled the linesperson, Murray challenged, and Hawkeye overuled the linesperson. Do you remember seeing any of the same people opposing Hawkeye now say something negative about it then?? ;)

Deivid23
03-05-2007, 06:23 PM
It is NOT better. If you took the time to read the post you would see that it is 50-50 who is correct on a CLOSE call, for a call within the error rate of the machine.

If it is 50-50 on a close call, but only the machine is always unbiased and the human isn't, give one good reason why they should go with the umpire on that one.

:lol:

That´s a good one, I will add it to MTF consensus about this issue

"1) Let the machine overrule or confirm the call, after all it´s a machine so somehow it´s better than the human eye.

2) Machine should be used to correct these calls and prevail even if it can´t intrinsically give anything but an useless measure just bc it´s unbiased"

:haha:

R.Federer
03-05-2007, 06:31 PM
:lol:

Thatīs a good one, I will add it to MTF consensus about this issue

"1) Let the machine overrule or confirm the call, after all itīs a machine so somehow itīs better than the human eye.

2) Machine should be used to correct these calls and prevail even if it canīt intrinsically give anything but an useless measure just bc itīs unbiased"

:haha:

Thanks yes, :lol: 2) about summarizes it nicely.

Btw, where were you and your great opposition to Hawkeye when nadal benefited from it during Melbourne? ;) I will have to go back and check, but I'm quite sure you didn't mind it then :D

This is not about Hawkeye for you is it. It is about one point, one loss to one player. Sorry that you have taken that loss so badly you are looking for rule changes to the entire game itself! :haha:

sarciness
03-05-2007, 06:33 PM
I feel I have a little expertise in statistics and error analysis. I'm a 3rd year undergrad physicist, so I do know at least some. Firstly, errors are ususally quoted to be + or - (some value)... so an error of 3.6 mm means that hawkeye could be as much as 3.6mm either side or the real bounce position.

Without having seen the data, an absolute error of 3.6mm is most probably the largest error that hawkeye produces. Random errors (like this one) ususally work on a bell curve (normal distribution) and therefore a large majority of calls are likely to be much less than 3.6mm inaccurate. I could guess at an average error of ą1mm. Like I said, I haven't seen the data, so I could be wrong.

R.Federer
03-05-2007, 06:36 PM
Still waiting for the geniuses to come up with just ONE good reason to go with the ump on CLOSE CALLS.
:lol:

Deivid23
03-05-2007, 06:42 PM
Thanks yes, :lol: 2) about summarizes it nicely.

Btw, where were you and your great opposition to Hawkeye when nadal benefited from it during Melbourne? ;) I will have to go back and check, but I'm quite sure you didn't mind it then :D

This is not about Hawkeye for you is it. It is about one point, one loss to one player. Sorry that you have taken that loss so badly you are looking for rule changes to the entire game itself! :haha:

http://www.menstennisforums.com/showpost.php?p=4965457&postcount=297

sarciness
03-05-2007, 06:46 PM
Forgot to say, I think it would undermine the public image of hawkeye to have an "unknown" call.

Deivid23
03-05-2007, 06:49 PM
I feel I have a little expertise in statistics and error analysis. I'm a 3rd year undergrad physicist, so I do know at least some. Firstly, errors are ususally quoted to be + or - (some value)... so an error of 3.6 mm means that hawkeye could be as much as 3.6mm either side or the real bounce position.

Without having seen the data, an absolute error of 3.6mm is most probably the largest error that hawkeye produces. Random errors (like this one) ususally work on a bell curve (normal distribution) and therefore a large majority of calls are likely to be much less than 3.6mm inaccurate. I could guess at an average error of ±1mm. Like I said, I haven't seen the data, so I could be wrong.


This is very probably true, but anyway, it still stands the fact that deviation within its precision is random so as an expert in statistics and error analysis, do u consider correct using a physical measuring system to correct a call in the range of its precision?

R.Federer
03-05-2007, 06:52 PM
http://www.menstennisforums.com/showpost.php?p=4965457&postcount=297

What is this supposed to show, that you were always against Hawkeye? Sure you were. That is why you didn't post anything about it for over a year. Even on close calls on any player, no comment. Why all this interest all of a sudden in revamping its position in the game.

Still waiting for one good reason why umpire's call should be chosen over a Hawkeye challenge answer on CLOSE calls.

Deivid23
03-05-2007, 06:52 PM
Forgot to say, I think it would undermine the public image of hawkeye to have an "unknown" call.

Very true, but this doesnīt imply it shouldnīt have it in order to be a fairer method

R.Federer
03-05-2007, 06:53 PM
Forgot to say, I think it would undermine the public image of hawkeye to have an "unknown" call.

Not unless it was balanced with umpires who would say into the microphone "Like Hawkeye, my human eye also has a margin of error and I am too uncertain to overrule" :D

Ariadne
03-05-2007, 06:58 PM
This is not about Hawkeye for you is it. It is about one point, one loss to one player. Sorry that you have taken that loss so badly you are looking for rule changes to the entire game itself! :haha:

:yeah:

There have been many uses of Hawkeye now and everyone knew that it was going to be the final decision. It's been hailed as just the thing to keep the fans happy and even though Fed has disliked it from the start, he's accepted it, including it being used to make him lose an apparent winning match point.

We now have Nadal complaining bitterly "stupid rule" because it went against him. He lost the set, Fed lost a match point. How differently they both handled it.

Now, it's being called into question. Why? Does it get banned because a petulant player lost a set? He still had the rest of the match to play. It's not as if there have been stacks of disagreement with the outcome before, thus seriously undermining the system's integrity.

Live by the Hawkeye, Die by the Hawkeye. If you accept it, you have to accept it with all it comes.

Deivid23
03-05-2007, 07:05 PM
It's been hailed as just the thing to keep the fans happy and even though Fed has disliked it from the start, he's accepted it, including it being used to make him lose an apparent winning match point.



Hawkeye has already gone in his favour on a matchpoint also

Deivid23
03-05-2007, 07:07 PM
What is this supposed to show, that you were always against Hawkeye? Sure you were. That is why you didn't post anything about it for over a year. Even on close calls on any player, no comment. Why all this interest all of a sudden in revamping its position in the game.

Timing ;)

Still waiting for one good reason why umpire's call should be chosen over a Hawkeye challenge answer on CLOSE calls.

:lol:

If you guys can´t get a simple concept why am I supposed to even try to go further? :confused:

sarciness
03-05-2007, 07:12 PM
This is very probably true, but anyway, it still stands the fact that deviation within its precision is random so as an expert in statistics and error analysis, do u consider correct using a physical measuring system to correct a call in the range of its precision?

I think it is logical to use hawkeye for anything over 3.6mm from the lines definitely and probably for all calls, as it's clearly better than linesmen or an umpire 99% of the time (plus unbiased as some people have said). An alternative would be to use super-slow-motion cameras (like MacCam at the US). Even then, you'll never get 100% accuracy.

I say go on using Hawkeye as it's being used currently.

p.s. I'm only a layman, not an expert in stats and error analysis.

Deivid23
03-05-2007, 07:16 PM
I think it is logical to use hawkeye for anything over 3.6mm from the lines definitely and probably for all calls, as it's clearly better than linesmen or an umpire 99% of the time (plus unbiased as some people have said). An alternative would be to use super-slow-motion cameras (like MacCam at the US). Even then, you'll never get 100% accuracy.

I say go on using Hawkeye as it's being used currently.

p.s. I'm only a layman, not an expert in stats and error analysis.

I agree HE is a correct mothod in probably more than 99% cases but summarizing and to make it clearer to you. what Iīve been talking about all this time is this:

"Correcting a measure within HEīs precision with sth that is intrinsecally useless in that range is a complete nonsense"

R.Federer
03-05-2007, 07:17 PM
If you guys can´t get a simple concept why am I supposed to even try to go further? :confused:

Sure ;) The real genius here is the one who doesnt have an answer but tries to belittle others' without an answer. And we losers, so sad that we can't see through the genius :sad:

Lets see again, when did Hawkeye become so terrible? Seems like one point in one match in the quarterfinals last week. And it wasn't the Federer quarter final, that's for sure. What a coincidence :haha: :rolls:

Deivid23
03-05-2007, 07:23 PM
Sure ;) The real genius here is the one who doesnt have an answer but tries to belittle others' without an answer. And we losers, so sad that we can't see through the genius :sad:


Itīs much funnier this way trust me;)

How are u expected to comprehend deeper thoughts when you think Hawkeye should be used to correct calls within its precision just bc itīs unbiased in its error? :lol:

sarciness
03-05-2007, 07:24 PM
I agree HE is a correct mothod in probably more than 99% cases but summarizing and to make it clearer to you. what Iīve been talking about all this time is this:

"Correcting a measure within HEīs precision with sth that is intrinsecally useless in that range is a complete nonsense"

Not meaning to sound rude... but eh? With sth? A measure within HE's precision?

I think you're trying to ask whether it's intrinsically useless to overrule a call where it was closer than its accuracy range? This is an interesting question. Obviously HE will not get 100% of the calls right in this case. I still think using it is a better alternative than linejudges or umpires, even in this case. To draw an arbitrary line, and say "linejudge's calls or umpire's calls will overrule HE in the case where its within 3mm of the line (or whatever number) becasue HE isn't 100% sure" is ludicrous if you accept the notion that it's more accurate than people.

R.Federer
03-05-2007, 07:34 PM
It´s much funnier this way trust me;)

How are u expected to comprehend deeper thoughts when you think Hawkeye should be used to correct calls within its precision just bc it´s unbiased in its error? :lol:

It is indeed funnier. Because so many people on this thread are laughing at you. ;) What is really funny though is how much one call in one game for one player can affect you, that you are now looking for some support in your one man tirade against Hawkeye. I'm sorry that you are waging a loser's battle --but maybe you are used to that ;)

It might be a good idea to differentiate your life from a tennis player's fortunes. Or we would all be gravely concerned about your health should someone not be able to defend his clay points. :D Still waiting for a genius answer to why the umpire's call, possibly with biased error, should be chosen over Hawkeye on close calls :lol: :haha:

Deivid23
03-05-2007, 07:35 PM
Not meaning to sound rude... but eh? With sth? A measure within HE's precision?

I think you're trying to ask whether it's intrinsically useless to overrule a call where it was closer than its accuracy range? This is an interesting question. Obviously HE will not get 100% of the calls right in this case. I still think using it is a better alternative than linejudges or umpires, even in this case. To draw an arbitrary line, and say "linejudge's calls or umpire's calls will overrule HE in the case where its within 3mm of the line (or whatever number) becasue HE isn't 100% sure" is ludicrous if you accept the notion that it's more accurate than people.

I´m not trying to sound rude, but, how can you say it won´t get 100% calls right in these cases as it would hit them almost all? :confused:

Despite being able to represent the HE´s error distribution in this range by a normal distribution the fact that it´s a RANDOM deviation after all STILL stands, therefore it´s theoretically useless to use it as a correcting method in this range

Deivid23
03-05-2007, 07:48 PM
It is indeed funnier. Because so many people on this thread are laughing at you. ;) What is really funny though is how much one call in one game for one player can affect you, that you are now looking for some support in your one man tirade against Hawkeye. I'm sorry that you are waging a loser's battle --but maybe you are used to that ;)

It might be a good idea to differentiate your life from a tennis player's fortunes. Or we would all be gravely concerned about your health should someone not be able to defend his clay points. :D Still waiting for a genius answer to why the umpire's call, possibly with biased error, should be chosen over Hawkeye on close calls :lol: :haha:

:lol:

1) I´m the one who is laughing at the ignorance of many people (curiosly, most of them are Fedtards :D ) in some basic concepts during these discussions.

2) No worries, I´ve never lost and will never lose any sleep bc the outcome of a tennis match, don´t panic ;) . I guess you all parrots whose usernames are all the same (R.Fed, RogerF, Federergreat, FedtardIam, IlikeFed´sass, etc) think all people are like you :D

3) To make it even better, you also have reading comprehension problems :lol:

I´ve never said human eye should prevail over Hawkeye, this is not the same thing that defending Hawkeye shouldn´t be used as a correcting method in ranges where it simply doesn´t happen to be an useful thing.

sarciness
03-05-2007, 07:49 PM
I´m not trying to sound rude, but, how can you say it won´t get 100% calls right in these cases as it would hit them almost all? :confused:

Despite being able to represent the HE´s error distribution in this range by a normal distribution the fact that it´s a RANDOM deviation after all STILL stands, therefore it´s theoretically useless to use it as a correcting method in this range

The normal distribution is not random in every sense of the word. It is more likely to be one thing (in this case accurate) than another (inaccurate). Without knowing the numbers, I can't say how much more likely it is to be accurate than inaccurate, but the fact that it has made thousands of correct calls with only a very small number of errors speaks volumes. To be explicit even within 3.6mm of the lines, HE is more than 50% likely to be correct. In the limit where hawkeye says the ball was in by 0.00000001mm, it is almost exactly 50% accurate. This is the worst case for hawkeye, and this is still as good as random guessing, which is at best what people can do in this range (less than 50% possibly due to mental presure exerted from players)

What I'm arguing is that it is better than any human judgment alternative (other than using people to judge super high framerate camreas).

R.Federer
03-05-2007, 07:53 PM
This is an interesting question. Obviously HE will not get 100% of the calls right in this case. I still think using it is a better alternative than linejudges or umpires, even in this case. To draw an arbitrary line, and say "linejudge's calls or umpire's calls will overrule HE in the case where its within 3mm of the line (or whatever number) becasue HE isn't 100% sure" is ludicrous if you accept the notion that it's more accurate than people.

Excellent points sarciness. And coming from someone who is putting in clear terms things that you are very obviously trained in, I think your points are absolutely correct. :yeah:

And since the creators of this technology, like you, have obvious training in the material, this is probably the kind of thinking that will make Hawkeye stay. A few unfavorable calls, and a lot of whining that follows, after forgetting all those points won on favorable challenges, not withstanding ;) Especially ridiculous ideas from coming from posters with no training claiming to know the technical side :haha:

Thank you for your excellent post.

Deivid23
03-05-2007, 07:58 PM
The normal distribution is not random in every sense of the word. It is more likely to be one thing (in this case accurate) than another (inaccurate). Without knowing the numbers, I can't say how much more likely it is to be accurate than inaccurate, but the fact that it has made thousands of correct calls with only a very small number of errors speaks volumes. To be explicit even within 3.6mm of the lines, HE is more than 50% likely to be correct. In the limit where hawkeye says the ball was in by 0.00000001mm, it is almost exactly 50% accurate. This is the worst case for hawkeye, and this is still as good as random guessing, which is at best what people can do in this range (less than 50% possibly due to mental presure exerted from players)

What I'm arguing is that it is better than any human judgment alternative (other than using people to judge super high framerate camreas).


No matter if HE´s error can be represented by a normal distribution, it´s still a random error despite having a bigger chance to hit it right in the center of the bell distribution, so by this randomness, Hawkeye can perfectly call OUT a ball that is really IN if the call is closer than its precision and that´s a fact, so this is where it loses all credibility for me in these particular cases as a method of correction

sarciness
03-05-2007, 08:03 PM
No matter if HEīs error can be represented by a normal distribution, itīs still a random error despite having a bigger chance to hit it right in the center of the bell distribution, so by this randomness, Hawkeye can perfectly call OUT a ball that is really IN if the call is closer than its precision and thatīs a fact, so this is where it loses all credibility for me in this particular cases as a method of correction

So essentially, you won't be happy until you have a delta-dirac distribution on line calls (i.e. in or out, 100% accuracy)? I see. Interesting. So using it becasue it's the best judgement we have isn't enough? Your alternative?

R.Federer
03-05-2007, 08:05 PM
The normal distribution is not random in every sense of the word. It is more likely to be one thing (in this case accurate) than another (inaccurate). Without knowing the numbers, I can't say how much more likely it is to be accurate than inaccurate, but the fact that it has made thousands of correct calls with only a very small number of errors speaks volumes. To be explicit even within 3.6mm of the lines, HE is more than 50% likely to be correct. In the limit where hawkeye says the ball was in by 0.00000001mm, it is almost exactly 50% accurate. This is the worst case for hawkeye, and this is still as good as random guessing, which is at best what people can do in this range (less than 50% possibly due to mental presure exerted from players)

What I'm arguing is that it is better than any human judgment alternative (other than using people to judge super high framerate camreas).


Sarciness, I don't believe the hawkeye error distribution can be normal. For one thing, the size of the area around the tennis court is finite. So any spot that hawkeye can pick up is a finite distance away from the side line or the base line and gives zero probability to a wide length of numbers on both tails.

Regardless, I think this part of your statement summarizes the problem perfectly

In the limit where hawkeye says the ball was in by 0.00000001mm, it is almost exactly 50% accurate. This is the worst case for hawkeye, and this is still as good as random guessing, which is at best what people can do in this range (less than 50% possibly due to mental presure exerted from players)

sarciness
03-05-2007, 08:06 PM
Excellent points sarciness. And coming from someone who is putting in clear terms things that you are very obviously trained in, I think your points are absolutely correct. :yeah:

And since the creators of this technology, like you, have obvious training in the material, this is probably the kind of thinking that will make Hawkeye stay. A few unfavorable calls, and a lot of whining that follows, after forgetting all those points won on favorable challenges, not withstanding ;) Especially ridiculous ideas from coming from posters with no training claiming to know the technical side :haha:

Thank you for your excellent post.

Thanks, I do try :wavey:

R.Federer
03-05-2007, 08:11 PM
So essentially, you won't be happy until you have a delta-dirac distribution on line calls (i.e. in or out, 100% accuracy)? I see. Interesting. So using it becasue it's the best judgement we have isn't enough? Your alternative?

Does not exist :lol: There will be belittling of everyone else's ideas (which do not match his non-ideas), but no sane response on why on close calls (50-50) the umpire's should be preferable to Hawkeye.

Previously, there was hiding under "you cannot comprehend such deep ideas" to everyone, but now since you certainly understand the material (and wayyy better than anyone else on the thread), lets see his response to you. ;)

Deivid23
03-05-2007, 08:11 PM
So essentially, you won't be happy until you have a delta-dirac distribution on line calls (i.e. in or out, 100% accuracy)? I see. Interesting. So using it becasue it's the best judgement we have isn't enough? Your alternative?

Make a better system than this, it could be done by having different locations of the cams courtside or modifying the number of the current 8 cams they work with, probably investing more money to get a sharper technology. I´m not much into scientific magazines these past years but I´m sure it can be done much better than these 0.36 cm without having to invest too much

Deivid23
03-05-2007, 08:12 PM
Does not exist :lol: There will be belittling of everyone else's ideas (which do not match his non-ideas), but no sane response on why on close calls (50-50) the umpire's should be preferable to Hawkeye.

Previously, there was hiding under "you cannot comprehend such deep ideas" to everyone, but now since you certainly understand the material (and wayyy better than anyone else on the thread), lets see a response to you. ;)

STFU u parrrot :lol:

R.Federer
03-05-2007, 08:14 PM
STFU u parrrot :lol:

STFU you ass :)

Sarciness exposed your stupidity nicely huh :lol:

sarciness
03-05-2007, 08:15 PM
Sarciness, I don't believe the hawkeye error distribution can be normal. For one thing, the size of the area around the tennis court is finite. So any spot that hawkeye can pick up is a finite distance away from the side line or the base line and gives zero probability to a wide length of numbers on both tails.

Regardless, I think this part of your statement summarizes the problem perfectly

It can still be normal, I think. Think of coke bottles. The amount of liquid they have is discrete and finite, as is their capacity. However, you'll notice that they aren't filled with the exact same amount of coke as each other (although they're very close). It's the same with hawkeye.

By the way, the normal distribution never falls to 0. So in theory, hawkeye could say a ball's a mile out, though this probability is so tiny, it's not worth thinking about!

R.Federer
03-05-2007, 08:18 PM
It can still be normal, I think. Think of coke bottles. The amount of liquid they have is discrete and finite, as is their capacity. However, you'll notice that they aren't filled with the exact same amount of coke as each other (although they're very close). It's the same with hawkeye.

By the way, the normal distribution never falls to 0. So in theory, hawkeye could say a ball's a mile out, though this probability is so tiny, it's not worth thinking about!

The range of the normal is -inf to +inf with prob appr. zero on both tails, but the range of Hawkeye is finite, with probability probably appr. zero on both tails. I guess you can say that theoretically Hawkeye's range is +inf to -inf since in principle if it was installed all the way to Timbuctoo it could measure a large value with small probability, but in practice --ie, on a court-- it is a distribution with bounded range (in fact, I think it only reports absolute errors so it really has only positive range).

sarciness
03-05-2007, 08:20 PM
Make a better system than this, it could be done by having different locations of the cams courtside or modifying the number of the current 8 cams they work with, probably investing more money to get a sharper technology. Iīm not much into scientific magazines these past years but Iīm sure it can be done much better than these 0.36 cm without having to invest too much

Right, so the millions invested are not enough? You think this is cheap easy stuff? When will it be accurate enough, 0.3cm, 0.1cm?

As far as I'm concerned if it's better than the current system of line judges (which it is far more often than not), it's worth implementing. I'm sure it will get more accurate as time goes on and improvements are made and we'll laugh at our 0.36cm accuracy. But for now it's better than anything else we have.

Deivid23
03-05-2007, 08:22 PM
STFU you ass :)

Sarciness exposed your stupidity nicely huh :lol:

:lol:

Reading comprehension problems again?

Sarciness also agrees Hawkeye canīt correct calls within its precisionīs range so heīs away from MTFīs consensus, not your case, though :D

sarciness
03-05-2007, 08:26 PM
The range of the normal is -inf to +inf with prob appr. zero on both tails, but the range of Hawkeye is finite, with probability probably appr. zero on both tails. I guess you can say that theoretically Hawkeye's range is +inf to -inf since in principle if it was installed all the way to Timbuctoo it could measure a large value with small probability, but in practice --ie, on a court-- it is a distribution with bounded range (in fact, I think it only reports absolute errors so it really has only positive range).

Of course, it's all theoretical. The normal is just a good model to start with, I'm not saying it's a perfect model of hawkeye. They've probably programmed it to produce error messages rather than report spurreous results and other clever technical things that I don't know about. They probably have bounded the range of it so it will never report "out by 10.0m"... Anyhow, I feel I'm done with this thread for now, nice to see some decent opinion rather than "it's got an uncertainty, we can't use it" :)

Deivid23
03-05-2007, 08:30 PM
Right, so the millions invested are not enough? You think this is cheap easy stuff? When will it be accurate enough, 0.3cm, 0.1cm?

As far as I'm concerned if it's better than the current system of line judges (which it is far more often than not), it's worth implementing. I'm sure it will get more accurate as time goes on and improvements are made and we'll laugh at our 0.36cm accuracy. But for now it's better than anything else we have.

I´m also sure those 0.36 cm will be dramatically improved in the next future and I think it also could be done now. As everything in business, I know they have likely just reached a compromise between accuracy and money invested, I was just expressing that it´s a completely wrong system to me in these particular cases as they´re giving the power of deciding to a system that can call a ball IN as OUT or viceversa just bc its own definition and limits as a physical measuring system.

stebs
03-05-2007, 08:36 PM
Iīm also sure those 0.36 cm will be dramatically improved in the next future and I think it also could be done now. As everything in business, I know they have likely just reached a compromise between accuracy and money invested, I was just expressing that itīs a completely wrong system to me in these particular cases as theyīre giving the power of deciding to a system that can call a ball IN as OUT or viceversa just bc its own definition and limits as a physical measuring system.

What you have completely missed Deivid is that the average error rate is 0.36 cm. The closer the ball to the line, the less margin for error there is meaning that the closest and most contraversial calls are those which are most likely to be correct. I think this fact makes a huge difference to your argument as it shows that the machine will always be more accurate in these circumstances than the human eye. Of course there are times when it may make an error in such cases where the call is unbelievably close but in a call as close as the one in the Nadal - Youzhny match the margin for error would have been much smaller than 0.36 cm.

Deivid23
03-05-2007, 08:40 PM
What you have completely missed Deivid is that the average error rate is 0.36 cm. The closer the ball to the line, the less margin for error there is meaning that the closest and most contraversial calls are those which are most likely to be correct. I think this fact makes a huge difference to your argument as it shows that the machine will always be more accurate in these circumstances than the human eye. Of course there are times when it may make an error in such cases where the call is unbelievably close but in a call as close as the one in the Nadal - Youzhny match the margin for error would have been much smaller than 0.36 cm.

Sorry but I can´t understand your point, I read the precision of the HE system was +- 0.36 cm in another thread, so the closer the call is, the bigger chance it has to miss:confused:

R.Federer
03-05-2007, 08:48 PM
Anyhow, I feel I'm done with this thread for now, nice to see some decent opinion rather than "it's got an uncertainty, we can't use it" :)
Thanks sarciness, very nice to get your opinions too. Refreshing change from the deflect and move tactics of others :)

I'm done with the thread too.

Apemant
03-06-2007, 08:01 AM
Sorry but I canīt understand your point, I read the precision of the HE system was +- 0.36 cm in another thread, so the closer the call is, the bigger chance it has to miss:confused:

Wrong. According to the link posted by R.Federer (I think), Paul Hawkins claims that the machine is set up to be more accurate with really close calls, than those further away (like 10 cm or more). And 3.6 mm is the average, for both really close ones and easy ones. Which means it's calibrated to be more accurate than 3.6 mm, for calls really close to the outer edge of a line.

But that's besides the point. Where you really are wrong is in the assumption that a machine can ever be 100% correct. 3.6 mm isn't some magic number, where all the calls outside that margin are 100% correct while those inside are less than 100%. In fact, both inside and outside of the 'magic' 3.6 mm range, the calls are less than 100% correct. So, if only 100% accuracy satisfies you, then Hawkeye is no good at all. For me, it's the other way around. I don't care about 100%, nothing in life is 100%. Umpires overrules were never 100% in the history of tennis. So, what I care for is the higher percentage of the probability that the call is correct. If Hawkeye says IN, and there is 90%, or even 80% chance (according to its accuracy and the distance from the edge) that it really was in, it's good for me - since the linespeople in such cases have about 50% chance of getting it right.

Regarding the IN/OUT/UNKNOWN, it would be even better if Hawkeye actually gave the estimation of the probability that the call is what it says it is. For example, it would say: IN - 100% (99.9% but rounded to nearest integer). Or it would say: IN - 75%. Then the umpire would decide how important the point was, and whether to accept Hawkeye's ruling or maybe replay the point or whatever. For example, IN - 75% obviously isn't good enough for a match point or even a set point, at least in my opinion.

But like someone said, that would really undermine the 'image' of Hawkeye's accuracy. People would actually realize that it can't be trusted 100%, which of course, never was in doubt as far as I'm concerned - but I guess people want the illusion of certainty, as if any certainty ever existed in real world.

oz_boz
03-06-2007, 08:42 AM
Apemant :yeah:

Discussing whether a new camera setting or some other technical adjustments can make for better calls makes little sense unless you're actaully involved in the making of Hawkeye - we don't know enough about it. Heck, I've even taken a grad course on computer vision, but that doesn't mean I would be able to evaluate the setup sytem without digging deep into the books, and probably not even then.

It would be possible with a high speed camera to get a distribution of HE:s error (probably looks something like a normal distribution). Measuring the distribution would make for calls being more or less LIKELY, like Apemant suggests. Then all we have to do is setting a threshold for acceptable probability - say 90% - and if the call is tighter replay the point.

I am an advocate of the 51% limit, basically because favour of wrong calls will even out; replaying points in the name of fairness would require a lot of replay without Hawkeye (see first post); that 100% calls is an impossibility, we have to settle with "likely" anyway. Letting the umpire decide percentage might work, but I guess he would have to be the only one with access to the HE call. Easy to see him say 90% is enough and players want 99, and we are back to were we started.

Apemant
03-06-2007, 09:14 AM
I am an advocate of the 51% limit, basically because favour of wrong calls will even out; replaying points in the name of fairness would require a lot of replay without Hawkeye (see first post); that 100% calls is an impossibility, we have to settle with "likely" anyway. Letting the umpire decide percentage might work, but I guess he would have to be the only one with access to the HE call. Easy to see him say 90% is enough and players want 99, and we are back to were we started.

Exactly... this Hawkeye thing might prove to be a Pandora's box. The more accuracy you attain, the less satisfied people are, because they can't accept the notion of not being absolutely certain... always.
This technology only proves that whether a ball is IN or OUT is actually non-deterministic. I mean, it is an elastic sphere hitting a non-perfectly straight line. There is no definitive answer what 'IN' really means in these circumstances. The same way as you can't really say 'how long the coastline really is' or 'how long this stick exactly is' etc. No more than you can write the exact value of Pi.

Deivid23
03-06-2007, 10:20 AM
Wrong. According to the link posted by R.Federer (I think), Paul Hawkins claims that the machine is set up to be more accurate with really close calls, than those further away (like 10 cm or more). And 3.6 mm is the average, for both really close ones and easy ones. Which means it's calibrated to be more accurate than 3.6 mm, for calls really close to the outer edge of a line.


Obviosuly Hawkins tries to defend his system but this theory folds in front of logic (Hawkeye doesn´t need to be that precise in calls 10cm or more off the line so that comparison is absolutely demagogic and useless) or in cases like the setpoint in Nadal-Youzhny match, where umpire, Nadal and Misha all saw the mark of the bounce OUT and Hawkeye claimed it IN.

"The mark of the ball was still on court and it was outside but in the challenge it was in, so that's unbelievable."

"I say to him, 'look, the ball is out', and he say 'I know'.

Even Youzhny agreed the ball was out.

"It looked like it was out," said Youzhny. "I saw the mark, but I just took the challenge because it was a very important point. When it showed it was good I was a little bit shocked."




But that's besides the point. Where you really are wrong is in the assumption that a machine can ever be 100% correct. 3.6 mm isn't some magic number, where all the calls outside that margin are 100% correct while those inside are less than 100%. In fact, both inside and outside of the 'magic' 3.6 mm range, the calls are less than 100% correct. So, if only 100% accuracy satisfies you, then Hawkeye is no good at all. For me, it's the other way around. I don't care about 100%, nothing in life is 100%. Umpires overrules were never 100% in the history of tennis. So, what I care for is the higher percentage of the probability that the call is correct. If Hawkeye says IN, and there is 90%, or even 80% chance (according to its accuracy and the distance from the edge) that it really was in, it's good for me - since the linespeople in such cases have about 50% chance of getting it right.

Regarding the IN/OUT/UNKNOWN, it would be even better if Hawkeye actually gave the estimation of the probability that the call is what it says it is. For example, it would say: IN - 100% (99.9% but rounded to nearest integer). Or it would say: IN - 75%. Then the umpire would decide how important the point was, and whether to accept Hawkeye's ruling or maybe replay the point or whatever. For example, IN - 75% obviously isn't good enough for a match point or even a set point, at least in my opinion.

But like someone said, that would really undermine the 'image' of Hawkeye's accuracy. People would actually realize that it can't be trusted 100%, which of course, never was in doubt as far as I'm concerned - but I guess people want the illusion of certainty, as if any certainty ever existed in real world.

Meanwhile Hawkeye has these black holes in such a close calls it´s a deluded thought to defend it has a big chance to hit the call right and a complete nonsense to have it as a method of correction in that particular range I insist

oz_boz
03-06-2007, 10:37 AM
Meanwhile Hawkeye has these black holes in such a close calls itīs a deluded thought to defend it has a big chance to hit the call right and a complete nonsense to have it as a method of correction in that particular range I insist

What percentage do you accept for a "big chance" then? 100% isn't reachable. Why not the 51% rule, when it evens out in the long run? Replaying uncertain points could take a lot of time, and furthermore a player in Misha's situation could say "Hey, the ball was PROBABLY ok, so why should I have to replay it?"

Castafiore
03-06-2007, 11:13 AM
Just a quote from Tennisreporters.net:

NEWSFLASH: HAWKEYE IS NOT FOOLPROOF: Two lines people told TennisReporters.net recently that Hawkeye can be inaccurate when its cameras are placed at odd angles and that does happen at different tournament locales as the available spots for the cameras change. Also, the system is apparently much weaker during twilight hours. The lines people also say that they don't see as well when the shadows begin to creep in, either.

The ATP rulebook says that the Electronic system can not be appealed. Its decision is final.

However, in this case (a match played late in the evening):
- Apperently, the ball mark was still visible and it showed that it was out
- Nadal saw that it was out
- Youzhny thought that it was out
- More importantly, the umpire saw it out.
So, despite this, the decision by the hawkeye system is final and the system said that it was in.

I'm still for Hawkeye and this match hasn't changed my opinion.
No system made by men is without error and the human eye is not perfect either so you're always going to end up with disputed points IMO.
However, some people have suggested that the umpires take on less responsibilities in decision making just because they're afraid of being proven wrong by a machine.

With hawkeye, the umpire decides to wait for the player to challenge and let the machine make the final decision. Instead of arguing with umpires, players will appeal on Hawkeye.
Without hawkeye, would the umpire have overturned the call by the line judge since he agreed with Rafa that it was out?

We'll never know of course. This may (but even that's far from certain) have cost Rafa the set but that didn't decide the entire match and overall, Youzhny was better. :)
Just like in Federer's match, there was a disputed challenge, one with which Federer did not agree but it cost him a set but he went on to win the match.

Apemant
03-06-2007, 11:53 AM
"The mark of the ball was still on court and it was outside but in the challenge it was in, so that's unbelievable."

"I say to him, 'look, the ball is out', and he say 'I know'.

Even Youzhny agreed the ball was out.

"It looked like it was out," said Youzhny. "I saw the mark, but I just took the challenge because it was a very important point. When it showed it was good I was a little bit shocked."


Contrary to the popular belief, ball marks aren't foolproof even on claycourts, let alone HC. Depending on spin the ball might 'dig' itself into clay and make a real mess out of it's 'mark'; so if the mark is really really close to the edge of the line, like a milimeter, then even on clay you just can't be 100% sure you got it right. It's usually up to the umpire to just decide whether it will be called in or out.

On HC, I assume the ball doesn't leave the mark as wide as it's full profile. In fact, when I think again, I'm not sure if it even matters. Hawkeye obviously displays the 'mark' as wide as the ball's full diameter. But what is the official rule regarding this? Does the ball have to actually touch the line, or its orthogonal projection needs to overlap the line? I don't have a clue about this. If the ball is slow, I guess it doesn't connect with the floor with it's full diameter so it's definitely not the same thing.

Sunset of Age
03-06-2007, 12:11 PM
Contrary to the popular belief, ball marks aren't foolproof even on claycourts, let alone HC. Depending on spin the ball might 'dig' itself into clay and make a real mess out of it's 'mark'; so if the mark is really really close to the edge of the line, like a milimeter, then even on clay you just can't be 100% sure you got it right. It's usually up to the umpire to just decide whether it will be called in or out.

Well said, indeed a fact that is often overlooked.
This thread was an interesting read indeed, but IMHO quite pointless - whatever system is used, Hawkeye, human eye, ball marks or future technology, 100% accuracy is unreachable. Sad but true.

And further: statistics will show that EVERY player gets an equal chance of having a call of his/her being in or out wrongly. You win some, you lose some... that was the case when Hawkeye wasn't yet invented, and sorrily enough, it will stay that way.

Apemant
03-06-2007, 01:11 PM
Well said, indeed a fact that is often overlooked.
This thread was an interesting read indeed, but IMHO quite pointless - whatever system is used, Hawkeye, human eye, ball marks or future technology, 100% accuracy is unreachable. Sad but true.

True, but not sad. :)

100% accuracy is unattainable, but it's not exactly important. Balls don't hit the ground so near the outer edge of a line all that frequently. If we get the accuracy of 1mm 99.9% of the time, that might cover for all but 0.1% of shots. So they could as well be replayed. Earlier, replaying was out of the question as linespeople got them wrong too often for replaying to be practical. I mean, who would want to replay every fifth point or so? Not to mention that linespeople are generally incapable of estimating how sure they are about any particular call. But given enough accuracy, I don't think replaying a single point per match or so would be all that bad, if the machine admits it's too close to make a definite call.

oz_boz
03-06-2007, 02:48 PM
This thread was an interesting read indeed, but IMHO quite pointless - whatever system is used, Hawkeye, human eye, ball marks or future technology, 100% accuracy is unreachable. Sad but true.


IMHO it's exactly the bolded quote that makes this discussion necessary, contrary to being pointless. Since no perfect measurement method exists, what to do with the close calls?

Sunset of Age
03-06-2007, 02:56 PM
IMHO it's exactly the bolded quote that makes this discussion necessary, contrary to being pointless. Since no perfect measurement method exists, what to do with the close calls?

Sorry, English isn't my maiden tongue, so I didn't even attempt to explain what I meant to say (or rather, in which I obviously failed to express myself accurately). My point is that since 100% accurateness isn't attainable, we should keep things the way the way they are, i.e. either human eye-judgement and/or hawkeye.
Mistakes will always occur, and as I said, every player will have the same chance of getting a wrong call in his or her favour or disadvantage in due time. No need for extra judgement possbilities IMHO, it would only make things even more difficult.

Problem is also - if nearly every close call will be reviewed, and the disadvantaged player doesn't agree with the decision - wouldn't the next step in this direction be the need for a Tennis High Court for Appeal? I surely hope not...

oz_boz
03-06-2007, 03:12 PM
I got your opinion right ClayBuster, just wanted to point out that dicussion is necessary before decision. I wasn't even sure I liked the present situation, but after some consideration, I do.

Maybe it is thinking too high about the tournament heads, but I believe they have thought a little about this before implementing Hawkeye, without going public - maybe to sustain the "perfect" image of machine calls like some have suggested.

Rogiman
03-06-2007, 03:23 PM
Oh, god, Tool23 is even more stupid than he first seemed, scary thought...

How are you going to convince people you're a physicist if you don't even have a clue about elemntary probability?

Image Processing and Stochastic Processes experts, most of them PhD holders in Physics and Mathematics, have worked their asses off to come up with a device whose average error is 3.6mm, and Tool23 thinks all it takes to improve it is "different locations of the cams courtside or modifying the number of the current 8 cams they work with, probably investing more money to get a sharper technology" :speakles:

Boy, that is laughable, the guy doesn't even have an elemntary school knowledge in physics OR in mathematics.

Tool23, with a mind as thick as yours, the best thing one can do is to shut up...

Rogiman
03-06-2007, 03:27 PM
What percentage do you accept for a "big chance" then? 100% isn't reachable. Why not the 51% rule, when it evens out in the long run? Replaying uncertain points could take a lot of time, and furthermore a player in Misha's situation could say "Hey, the ball was PROBABLY ok, so why should I have to replay it?"Are you seriously expecting an answer from this tool?
The guy doesn't know what Random Variable is.

R.Federer
03-06-2007, 03:30 PM
Just a quote from Tennisreporters.net:



The ATP rulebook says that the Electronic system can not be appealed. Its decision is final.

However, in this case (a match played late in the evening):
- Apperently, the ball mark was still visible and it showed that it was out
- Nadal saw that it was out
- Youzhny thought that it was out
- More importantly, the umpire saw it out.
So, despite this, the decision by the hawkeye system is final and the system said that it was in.
Just like in Federer's match, there was a disputed challenge, one with which Federer did not agree but it cost him a set but he went on to win the match.

Thanks for digging that up. I wondered if the umpire had any options once Hawkeye had made the call. It lends weight to the argument that the ball was out, if the linesperson as well as the umpire saw it out. However, Youzhny's view on the matter should be irrelevant. He was on the far side of the court. If he had claimed he saw it was in, people would have said that he was too far to make a judgement. So it's just convenient to take his opinion as relevant at all when he says its out.
In the Federer match, I did not see that Federer did not agree with the call. How do you mean? Did he say it looked good but was overturned by Hawkeye? I don't think that happened. He took the decision of Hawkeye and did not complain.

oz_boz
03-06-2007, 03:35 PM
It lends weight to the argument that the ball was out, if the linesperson as well as the umpire saw it out.

It does, but negligible weight compared to Hawkeye's decision.

Rogiman
03-06-2007, 03:42 PM
No human being can make a clear decision about a ball landing within 3.6 mm off the line, period.

It means nothing if the players thought it was out.

oz_boz
03-06-2007, 04:01 PM
No human being can make a clear decision about a ball landing within 3.6 mm off the line, period.

It means nothing if the players thought it was out.

It means SOMETHING, you know that ;). But most surely not enough to tip the favour over to OUT.

Rogiman
03-06-2007, 04:07 PM
It means SOMETHING, you know that ;). But most surely not enough to tip the favour over to OUT.Well that is exactly what Tool23 can't understand, nowadays most of the algorithms in the industry are accepted or rejected according to their probability of success.
Yes, both the human eye's error and the machine's arror are random variables (and no, Tool23, not all random variables have uniform distributions, some events may have greater probability than others :retard:), but the human eye's error's expectation is much greater.

R.Federer
03-06-2007, 04:38 PM
It does, but negligible weight compared to Hawkeye's decision.

Precisely, that's what I think as well.

Like others said, no matter what you do, Hawkeye remains a machine. there will always be some error. If it goes to .3mm, it is still error. So it is just a useless idea to invest tonnes more of money, to get improved accuracy of .1mm. If it can be made more accurate with a reasonable investment, go for it, but it still will never be fool proof 100%.

In very close calls, Hawkeye is still preferable to the human eye. It is unbiased whether it is calling for #1 or for #1000. Until, that is, our resident conspiracy theorist (and we all know who that is ;) ) suggests hawkeye is also calibrated in favor of #1.

Castafiore
03-06-2007, 05:50 PM
In the Federer match, I did not see that Federer did not agree with the call. How do you mean? Did he say it looked good but was overturned by Hawkeye? I don't think that happened. He took the decision of Hawkeye and did not complain.

Also from TennisReportersnet:
Federer also claimed that Hawkeye cost him a match point at 6-5 in the breaker. "The ball was out, but it's hard to accept, match point," said Federer. "It was a bit of a pity it went three and it had a bitter taste at the end with the net cord and Hawkeye."

Also in the ATP rulebook, you can read that only ball marks on clay can be inspected by an umpire before making a final decision on a point. So, a ball mark on hard courts can be an indicator perhaps but you can't ask an umpire to come down and check the ball mark on a surface other than clay.

R.Federer
03-06-2007, 07:11 PM
Also from TennisReportersnet:
Federer also claimed that Hawkeye cost him a match point at 6-5 in the breaker. "The ball was out, but it's hard to accept, match point," said Federer. "It was a bit of a pity it went three and it had a bitter taste at the end with the net cord and Hawkeye."



?
Right, so he agreed it was out. He hit the last ball in that set if I remember correctly, and it was on the deuce side of djokovic's half, and it went off the sideline. It wasn't called, until it was challenged by djokovic. At which point it was determined to be out by Hawkeye. And this comment shows he did not dispute it. So I don't think he disagreed with it at all.

Also in the ATP rulebook, you can read that only ball marks on clay can be inspected by an umpire before making a final decision on a point. So, a ball mark on hard courts can be an indicator perhaps but you can't ask an umpire to come down and check the ball mark on a surface other than clay.

Thanks for that. It makes sense.

Deivid23
03-06-2007, 09:59 PM
Oh, god, Tool23 is even more stupid than he first seemed, scary thought...

How are you going to convince people you're a physicist if you don't even have a clue about elemntary probability?

Image Processing and Stochastic Processes experts, most of them PhD holders in Physics and Mathematics, have worked their asses off to come up with a device whose average error is 3.6mm, and Tool23 thinks all it takes to improve it is "different locations of the cams courtside or modifying the number of the current 8 cams they work with, probably investing more money to get a sharper technology" :speakles:

Boy, that is laughable, the guy doesn't even have an elemntary school knowledge in physics OR in mathematics.

Tool23, with a mind as thick as yours, the best thing one can do is to shut up...

:lol:

What a jackass you are :haha:

Wanna bet than in not a long period of time Hawkeye´s precision will be much better and that this will be likely achieved with a slight change in cams location and a sharper technology which could give wider ranges, more sensitive captures in darker conditions, etc? Use a bit your brain, it won´t hurt you, burro :D

The one who doesn´t stand a clue about probabilites and what a physical measure is it´s only you, as it has been clearly shown last days :hug:

Keep bitching, sucking your boyfriend Ron´s ass or doing chilidish avatars as it´s the best you can do, don´t polllute this interesting thread and stick to your clown role out there, thanks ;)

Deivid23
03-06-2007, 10:03 PM
It means nothing if the players thought it was out.

:lol:

Yes, it means nothing both players and the umpire saw both the ball and the mark out :retard:

Sunset of Age
03-06-2007, 10:07 PM
?
Right, so he agreed it was out. He hit the last ball in that set if I remember correctly, and it was on the deuce side of djokovic's half, and it went off the sideline. It wasn't called, until it was challenged by djokovic. At which point it was determined to be out by Hawkeye. And this comment shows he did not dispute it. So I don't think he disagreed with it at all.

that's exactly the way I remember it as well. Of course, Fed wasn't exactly happy about the challenge outcome, but he did not DISPUTE it, which is quite a difference.
And we all know Fed has openly stated his opinion on Hawkeye long before - I happen to disagree with him on this, BTW ;) - but he's accepted that it's around and will accept its consequences, i.e. calls to his disadvantage, as well.

fmolinari2005
03-06-2007, 10:14 PM
Reasons for ending the use of the hawkeye:

- it isnt just as good as ball marks on clay
- isnt as good as Nadal`s naked eye
- is a bad person

So, we should change to a more precise mechanism. The Hawkass system.

Deivid23
03-06-2007, 10:20 PM
What percentage do you accept for a "big chance" then? 100% isn't reachable. Why not the 51% rule, when it evens out in the long run? Replaying uncertain points could take a lot of time, and furthermore a player in Misha's situation could say "Hey, the ball was PROBABLY ok, so why should I have to replay it?"

Looks like Hawkeye isn´t sharper enough to be an useful method of correction in a particular range so why use it in those cases? Of course human eye can´t get it right but I´m afraid HE´s doesn´t imply a good enough of an improvement to let it correct every call. Probably many people prefer this but it´s not a fair method if u ask me when you can hear of cases like that one at setpoint in Youhzny Nadal, where one can almost say even human eye has likely got the better of the system (and that simply implies it is not more than a coin toss in some calls within a particular range)

World Beater
03-06-2007, 11:37 PM
wait...if blake had been gipped instead of nadal. Would ED have reversed the call and the match?

:haha:

uhh...continue..interesting thread.

Peoples
03-06-2007, 11:43 PM
Engineer Tool23 is working on a new device.

He's going to install Nadal in every court to make the calls :lol:

:haha:

:lol: @ the Nadal fans bitching now

Vass
03-07-2007, 03:51 AM
I've been opn the Dubai tournament and i didn't believe the hawk-eye that was installed there even for a bit. On one of the baselines it seemed to have been moving the balls 15 cm into the court. This happened twice in womens matches that i saw, when calls that weren't even close were shown just 3 mm out. And i'm absolutely sure that they weren't even close because i was sitting right behind the line judge and so were the people i was watching the matches with.

Plus we all know of cases when players pointed out to dust prints on the court, that proved the hawk-eye wrong.

The technology really needs more calibrating.
I'm still not against it though, because the majority of the calls it makes seem to be correct.

Castafiore
03-07-2007, 07:38 AM
that's exactly the way I remember it as well. Of course, Fed wasn't exactly happy about the challenge outcome, but he did not DISPUTE it, which is quite a difference.
And we all know Fed has openly stated his opinion on Hawkeye long before - I happen to disagree with him on this, BTW ;) - but he's accepted that it's around and will accept its consequences, i.e. calls to his disadvantage, as well.
Well, in this particular discussion about Hawkeye - is it enough or not - the reaction of players (although interesting) is not that relevant to me since it's not really about how a player reacts to it but more about the system itself. Sure, the players have to accept a decision by Hawkeye so they might as well move on as fast as they can (which is sometimes easier said than done of course).
However, in some cases, players as well as some line judges (...) have casted a doubt on the accuracy of the system + what's interesting to me is to see how umpires have perhaps become more passive as a result of this Electronic eye.

I'm still a fan of Hawkeye, though. But, it's useful to review these systems regularly, right? It's one thing to accept that no system made by humans is perfect but that doesn't exclude the possibility to assess possible errors and review the system on a regular basis.