April 10, 2014

The problem with low catches

The cricket cognoscenti is largely eager to give the fielder the benefit of the doubt when a catch is referred. This should not be the case
18

Video evidence often creates doubt about low catches because of the requirements of the law
Video evidence often creates doubt about low catches because of the requirements of the law © Getty Images

Mahela Jayawardene was spectacularly caught at square cover by Michael Lumb in the World T20 game between England and Sri Lanka on March 27, only to have the decision reversed by TV umpire Steve Davis upon review. Davis was universally condemned. "Absolute Shocker!!!" tweeted England Test wicketkeeper Matthew Prior. In the Cricketer's view Jayawardene "should have been caught - in fact he was caught". On the BBC the commentator Jonathan Agnew and former Sri Lanka batsman Russel Arnold were in agreement that the catch was clean. Andrew McGlashan summarised the matter when he wrote: "Steve Davis ruled there was doubt, as is so often the case with TV pictures, but on this occasion few who viewed the images could quite fathom the decision."

After the ICC released the exact footage used by Davis to make his decision, most publications shared this new information with fans without comment, but the great New Zealand batsman Martin Crowe stood his ground about the catch having been legitimate. In doing so, he touched on a deeper truth about how we watch cricket and how we view evidence in it. Before the clarifying picture became available, the episode was seen as yet another in a long list of perfectly legitimate catches denied because the video replay inserted doubt where there ought to have been none.

I was not convinced that Davis was wrong. First, I have for long been a supporter of the TV review being used extensively for low catches. Given what the current law states (Law 32 and Law 19 are both essential to fully define a catch), given that cricket is not played on astroturf, given the comparative size of the cricket ball and the adult human hand (specifically the fingers), basic geometry and physics suggest that low catches should be legally completed very rarely. It is impossible to say with confidence, unless the fingers are all together and the ball is deep in the palm, that a ball lying in two hands that are resting on a grass outfield does not touch the grass at all. For this is what the law requires in a fair catch.

As a general rule, it is reasonable to assume that a ball that is caught with the hand or hands (more likely the fingers) simultaneously touching the ground and the ball is, most times, unlikely to have been caught fairly.

This is at odds with observations often heard on commentary. How often do we hear the remark, "Everybody knows the catch was clean. Taking it to the review is always going to throw up at least one doubtful picture." This is misplaced. The video evidence creates doubt because of the requirements of the law, not simply because of the limits of high-definition video technology.

The second reason for my scepticism is that there is no reason to think that the broadcasters show every available camera angle. The broadcast is produced by a director who puts together pictures from multiple cameras. The feed viewers receive is not the same as the one available to the TV umpire. This has been the case for a number of seasons now. The ICC publishes a drawing specifying where broadcasters are required to place cameras for broadcast, and where cameras for umpires are to be placed. Here is the 2013-14 playing handbook (pdf, see p. 5.34). These layouts have been published by the ICC since 2008-09 (pdf, see p. 98).

A reasonable observer ought not to be sure that every available angle has been shown. One can understand viewers not knowing the laws and playing conditions, but surely it is not too much to expect from professional broadcasters in TV and radio commentary and press boxes? If you were the ICC and one of your employees faced the kind of ridicule and abuse that Steve Davis faced simply because the professional commentariat did not know the laws and playing conditions, how would you feel about it? The ICC's restraint has been admirable.

It has to be emphasised here that all the camera angles were not available because of editing decisions made by the director of the broadcast, not by the ICC or by the TV umpire. Not only do professional commentators and reporters (who travel from series to series, year after year) seem to be unaware of the rules, they also seem unaware, in the case of the live commentators, of the precise role their colleagues play in bringing us the broadcast and in providing pictures to the TV umpire.

It is not the case that the bias against the evidence of video replays is part of a general scepticism towards technologically produced evidence about the precise nature of events on the field. While most commentators are sceptical about the use of video reviews for low catches, there is another area in which many of these same commentators have become slavishly devoted to what the technology shows.

Since the advent of the ball-tracking animation, cricketers with vast international experience have seemingly lost the ability to judge lbws. "Let's wait for the ball-tracking prediction" is the mantra. It is the stray commentator nowadays who will say, "That looks close", upon seeing a strong lbw appeal. These days, if the ball track shows the ball missing leg stump by a whisker, and the umpire had given it out, then the umpire is wrong, without caveats.

Contrary to common perception, the ball track does not offer a prediction. A prediction is a forecast. Forecasts cannot be made about things that have already taken place. The delivery being modelled has already been completed by the time the occasion to surmise on the hypothetical path of the ball in the absence of the pad arises. The ball track offers a supposition or conjecture. Just as the umpire does. Predictions have the valuable property of being proven right or wrong by ensuing reality; this makes good predictions extremely valuable. The fact that the ball-tracking simulation is called a prediction and not a conjecture provides a clue as to why it is taken seriously to the extent that it is used to verify the umpire's original decision. Unlike in the case of the video replay, the computer model is not a record of the actual event, but an idealised virtual reconstruction of it.

Today an idealised virtual reconstruction is accepted with far greater readiness than a high-definition video replay that breaks down an actual event into hundreds of frames per second. Professional commentators and supporters of the use of ball-tracking and DRS ought to reflect on this peculiarity in their attitude to evidence in our sport, which is increasingly reliant on technology. The nicest, surest-looking picture need not be the least questionable one.

Those who are paid to explain the sport to the public and report on it have a duty to respect umpires. At the very least, they ought not to undermine them. What is remarkable about the Lumb-Jayawardene-Davis episode is that there has not been a single mea culpa, apart from Mr Crowe's (and even this did not extend to the substantive point that Davis was right about the catch), in the professional press.

Cricket deserves better.

Kartikeya Date writes at A Cricketing View and tweets here

Comments have now been closed for this article

  • on April 10, 2014, 15:50 GMT

    To add to my points from the post -

    Any simulation like a ball tracker has to be statistical. I understand the ball has to obey the laws of physics. The location of the ball in 3D space is a matter for statistics, as is the point of impact of the ball on pad (which is done differently and with different levels of automation by the two vendors who provide this technology). So while the ball obeys the laws of physics, where the ball is, is only available in the form of a range or distribution of likelihoods. This is why it is idealized. The surest looking picture is imprecise because scientifically, a precise answer would be a statistical range about the possible trajectories of the ball with explicitly given margin of error.

    I have no problem with using ball tracking, because even if it is inaccurate, it is systematically and reliably so. Engineers or scientists will understand exactly what I mean here.

    The picture we see is an idealized reconstruction for this reason.

  • on April 11, 2014, 17:35 GMT

    @inswing - The following two claims are true in my view. Unless you are trying to dispute either of them, the absurd extreme that "everything is a reconstruction" is moot. Of course everything is a reconstruction.

    1. A 3D reconstruction (which depends on coordinate geometry) and a video record (which depends on light) are categorically different types of reconstruction.

    2. A video record is a record of an actual event, while the conjecture about the path in a ball tracker describes a hypothetical event.

    There is no reason to believe one more readily than the other. This is the extent of my point. At no point have I suggested at ball trackers are inaccurate/accurate.

    As for the point about reviews, Paul Hawkins, the founder of Hawkeye told me in an interview in 2011 that there had been no field tests of Hawkeye's technology which he would be willing to hang his hat on. Ian Taylor of Virtual Eye has been consistently more cautious than the Hawkeye people in these very pages.

  • inswing on April 11, 2014, 15:22 GMT

    "Idealized reconstruction" does not mean inaccurate. Being an engineer and a scientist both, I know that _everything_ is an idealized reconstruction. Nothing you see, hear, or feel is "as is". Everything is a matter of statistics. The earth's position around the sun is a matter of statistics too. It is a idealized reconstruction and there is a range of possibilities. It can be predicted with a high degree of accuracy nonetheless and be used with great success. Because the margin of error is so small, that for practical purposes (e.g. to position a satellite) you can assume that we can correctly predict the position of the earth. Date confuses "statistical" with "inaccurate", which is a hallmark of someone who lacks science education. If the margin of error in the ball trajectory prediction sufficiently small (e.g. a couple of mm) then practically the prediction is correct. To say that "well, it is all statistical so it is inaccurate and can't be trusted" is wrong.

  • U.A.1985 on April 11, 2014, 10:53 GMT

    @lanky1 - how can you tell what really is ground then? If grass is not frpund then what part of ground is ground. Perhaps little bit of soil above ground is loose and maybe that also spreads out from the hands? In my opinion catch in a way is not fair bcoz fielder is taking assistance of ground. A cstch should be completely completed in air and then it should be classed as fair.

  • OldBertie on April 11, 2014, 2:55 GMT

    I think we have to be careful with describing ball tracking as a conjecture. While it isn't technically a prediction, as the event it's describing has no chance of happening, it functions the same way as a prediction. I say this because the system is provided with the information up to impact, and then it assumes no impact and predicts where the ball would have gone. I have no problem comparing this to a prediction, because in testing of the technology (I assume) they will have bowled a whole lot of balls at a set of stumps, then fed only two-thirds of the information to the system, allowing it to make what is functionally a prediction from the view of the system. The prediction can either be verified or disproved quite simply, and I assume that, since we're still using the technology after a bunch of independent testing, that the system has passed with flying colours.

  • McWheels on April 10, 2014, 20:07 GMT

    Grass isn't in the laws, the ground is. When fingers are on the ground, in the grass, the ball will bounce on them in a very similar way when completing a catch. I have no sympathy for the umpire or the article writer.

    The decision was plain wrong, the users don't understand it or the game sufficiently to make a sound judgement.

  • inswing on April 10, 2014, 14:58 GMT

    About the pointless and nit-picky claim that "it is not a prediction", it is absolutely a prediction. An event that did not take place, and you are predicting what would have happened, so it is a prediction. But terminology aside, just because it is a prediction does not mean that it is unreliable and untrustworthy. The ball does not behave randomly, but must obey certain laws. The projection is not some kind of guess in "idealized" conditions, it is very accurate in current conditions, and this can be verified easily. There is always a margin of error, but the point is that that margin is small enough so that one can consider the projected path to be "correct" in practice. The coverage of an event that took place, like a catch, can absolutely be less reliable than that, given that we are tying to figure out the position of the ball within a few millimeters from footage that does not have that resolution. It is very logical and sensible to trust projections more than catch footage.

  • Sigismund on April 10, 2014, 12:46 GMT

    Another great article, but I am firmly with py0alb: the grass is not the same thing as the ground. When the ball hits the ground, it bounces; this is not true of the grass on its own. This is how everyone playing in all the millions of non-DRS cricket matches that ever happenend could easily tell what was a fair catch and what was not; and this is clearly what is intended by law 32. Delighted you have raised the issue of LBWs. I have been carping on about this for ages - the DRS simply misundertands the rules. The half-ball condition is supposed to cover the margin of error but is too narrow. Not Out should only be overturned if the prediction shows it was utterly plumb (which could be defined); Out should only be overturned if pitching well outside leg or missing by a full ball at least.

  • py0alb on April 10, 2014, 10:21 GMT

    Law 32. b)

    "the ball does not touch the ground even though the hand holding it does so in effecting the catch."

    We could probably do with a clearer definition of "touch the ground" and "in effecting the catch".

    The MCC clarifies: A fielder catches the ball when he takes it in the air from a stroke of the bat. He does not complete the catch until he has established both forms of complete control

    (i) over the ball

    and (ii) over his own

    which seems to sort out the second question, but there is no clarification for what touching the ground means.

  • py0alb on April 10, 2014, 10:11 GMT

    "a ball lying in two hands that are resting on a grass outfield does not touch the grass at all. For this is what the law requires in a fair catch. "

    The law makes no mention of the grass. the law mentions the ground. The grass is not the ground. The grass is what covers the ground.

  • on April 10, 2014, 15:50 GMT

    To add to my points from the post -

    Any simulation like a ball tracker has to be statistical. I understand the ball has to obey the laws of physics. The location of the ball in 3D space is a matter for statistics, as is the point of impact of the ball on pad (which is done differently and with different levels of automation by the two vendors who provide this technology). So while the ball obeys the laws of physics, where the ball is, is only available in the form of a range or distribution of likelihoods. This is why it is idealized. The surest looking picture is imprecise because scientifically, a precise answer would be a statistical range about the possible trajectories of the ball with explicitly given margin of error.

    I have no problem with using ball tracking, because even if it is inaccurate, it is systematically and reliably so. Engineers or scientists will understand exactly what I mean here.

    The picture we see is an idealized reconstruction for this reason.

  • on April 11, 2014, 17:35 GMT

    @inswing - The following two claims are true in my view. Unless you are trying to dispute either of them, the absurd extreme that "everything is a reconstruction" is moot. Of course everything is a reconstruction.

    1. A 3D reconstruction (which depends on coordinate geometry) and a video record (which depends on light) are categorically different types of reconstruction.

    2. A video record is a record of an actual event, while the conjecture about the path in a ball tracker describes a hypothetical event.

    There is no reason to believe one more readily than the other. This is the extent of my point. At no point have I suggested at ball trackers are inaccurate/accurate.

    As for the point about reviews, Paul Hawkins, the founder of Hawkeye told me in an interview in 2011 that there had been no field tests of Hawkeye's technology which he would be willing to hang his hat on. Ian Taylor of Virtual Eye has been consistently more cautious than the Hawkeye people in these very pages.

  • inswing on April 11, 2014, 15:22 GMT

    "Idealized reconstruction" does not mean inaccurate. Being an engineer and a scientist both, I know that _everything_ is an idealized reconstruction. Nothing you see, hear, or feel is "as is". Everything is a matter of statistics. The earth's position around the sun is a matter of statistics too. It is a idealized reconstruction and there is a range of possibilities. It can be predicted with a high degree of accuracy nonetheless and be used with great success. Because the margin of error is so small, that for practical purposes (e.g. to position a satellite) you can assume that we can correctly predict the position of the earth. Date confuses "statistical" with "inaccurate", which is a hallmark of someone who lacks science education. If the margin of error in the ball trajectory prediction sufficiently small (e.g. a couple of mm) then practically the prediction is correct. To say that "well, it is all statistical so it is inaccurate and can't be trusted" is wrong.

  • U.A.1985 on April 11, 2014, 10:53 GMT

    @lanky1 - how can you tell what really is ground then? If grass is not frpund then what part of ground is ground. Perhaps little bit of soil above ground is loose and maybe that also spreads out from the hands? In my opinion catch in a way is not fair bcoz fielder is taking assistance of ground. A cstch should be completely completed in air and then it should be classed as fair.

  • OldBertie on April 11, 2014, 2:55 GMT

    I think we have to be careful with describing ball tracking as a conjecture. While it isn't technically a prediction, as the event it's describing has no chance of happening, it functions the same way as a prediction. I say this because the system is provided with the information up to impact, and then it assumes no impact and predicts where the ball would have gone. I have no problem comparing this to a prediction, because in testing of the technology (I assume) they will have bowled a whole lot of balls at a set of stumps, then fed only two-thirds of the information to the system, allowing it to make what is functionally a prediction from the view of the system. The prediction can either be verified or disproved quite simply, and I assume that, since we're still using the technology after a bunch of independent testing, that the system has passed with flying colours.

  • McWheels on April 10, 2014, 20:07 GMT

    Grass isn't in the laws, the ground is. When fingers are on the ground, in the grass, the ball will bounce on them in a very similar way when completing a catch. I have no sympathy for the umpire or the article writer.

    The decision was plain wrong, the users don't understand it or the game sufficiently to make a sound judgement.

  • inswing on April 10, 2014, 14:58 GMT

    About the pointless and nit-picky claim that "it is not a prediction", it is absolutely a prediction. An event that did not take place, and you are predicting what would have happened, so it is a prediction. But terminology aside, just because it is a prediction does not mean that it is unreliable and untrustworthy. The ball does not behave randomly, but must obey certain laws. The projection is not some kind of guess in "idealized" conditions, it is very accurate in current conditions, and this can be verified easily. There is always a margin of error, but the point is that that margin is small enough so that one can consider the projected path to be "correct" in practice. The coverage of an event that took place, like a catch, can absolutely be less reliable than that, given that we are tying to figure out the position of the ball within a few millimeters from footage that does not have that resolution. It is very logical and sensible to trust projections more than catch footage.

  • Sigismund on April 10, 2014, 12:46 GMT

    Another great article, but I am firmly with py0alb: the grass is not the same thing as the ground. When the ball hits the ground, it bounces; this is not true of the grass on its own. This is how everyone playing in all the millions of non-DRS cricket matches that ever happenend could easily tell what was a fair catch and what was not; and this is clearly what is intended by law 32. Delighted you have raised the issue of LBWs. I have been carping on about this for ages - the DRS simply misundertands the rules. The half-ball condition is supposed to cover the margin of error but is too narrow. Not Out should only be overturned if the prediction shows it was utterly plumb (which could be defined); Out should only be overturned if pitching well outside leg or missing by a full ball at least.

  • py0alb on April 10, 2014, 10:21 GMT

    Law 32. b)

    "the ball does not touch the ground even though the hand holding it does so in effecting the catch."

    We could probably do with a clearer definition of "touch the ground" and "in effecting the catch".

    The MCC clarifies: A fielder catches the ball when he takes it in the air from a stroke of the bat. He does not complete the catch until he has established both forms of complete control

    (i) over the ball

    and (ii) over his own

    which seems to sort out the second question, but there is no clarification for what touching the ground means.

  • py0alb on April 10, 2014, 10:11 GMT

    "a ball lying in two hands that are resting on a grass outfield does not touch the grass at all. For this is what the law requires in a fair catch. "

    The law makes no mention of the grass. the law mentions the ground. The grass is not the ground. The grass is what covers the ground.

  • on April 10, 2014, 10:01 GMT

    Harsha immediately reacts saying, 'Now, where did that pitch?' evenn if it has pitched on middle.

  • Lanky1 on April 10, 2014, 10:01 GMT

    The point is that the blades of grass are not part of the 'ground'. So the fingers may be spread with the grass poking through them but so long as they are under the ball preventing the ball from touching the soil the catch is made. In the palms down case where the ball touches the soil that is plainly not at catch.

  • sabihoff12 on April 10, 2014, 8:56 GMT

    Good article. Loved reading it

  • FahadBashir on April 10, 2014, 8:00 GMT

    While your article is well thought out and explained, I find it difficult to deal with the unnecessary criticism of ball tracking technology. Ball tracking technology and its predictive path are not based on some computer mumbo jumbo recently indoctrinated into science, but undeniable laws of physics widely accepted since Sir Isaac Newton's opportune discovery of the laws of motion and gravity 400 years ago.

  • Insightful2013 on April 10, 2014, 6:21 GMT

    Umpires are the same as referees. I suggest if you cannot recall, use, You tube. England was knocked out of the 2010 World Cup by having a goal disallowed that was at least a yard over the line. Technology all the way, all the time. USA, same tournament, knocked out by a bent Ref from Mali, I think. Humans can never be objective and tools are created to help,assist,benefit, so use them. This is the silliest debate ever! The error was made by Davis, reinforcing the use of better tools and limiting human involvement.

  • waspsting on April 10, 2014, 4:26 GMT

    I couldn't agree more.

    It seems to me that most - almost all, honestly - "experts" don't even seem to know what the law about clean catches are.

    The Jaywardena incident aside, frequently fielders take a catch and then fall to the ground with their palms down, i.e. the ball in hand but touching the ground.

    This is not a catch, as per the rules, but NO ONE - I mean the commentators - even mention the possibility of it not being a catch, which I find astounding.

    For example, the day after the Eng-Sri Lanka game, a Bangldeshi batsmen was caught this way, and not a peep out of the commentators.

    I remember over 10 years ago Jonty Rhodes took a superhuman diving 'catch' but could do nothing other than fall to ground palms down. Geoff Boycott correctly said it was not out but the other commentator thought it was out because there was no other way to have fallen in the circumstances!

    If the 'experts' don't know the rules... there's a serious problem here

  • on April 10, 2014, 3:46 GMT

    @Ozcricketwriter - You seem to be referring to two separate catches in your comment. I only refer to one catch - the one Michael Lumb claimed against Mahela Jayawardene. Which is the second one?

  • Ozcricketwriter on April 10, 2014, 3:30 GMT

    That catch clearly touched the ground, so should have been not out. We might think that it is out because he put his hands around the ball and didn't drop it; but the laws as they stand say that if any part of the ball touches the ground it is not out. If you think that this catch is not out, your argument is with the laws, not with the umpire.

  • Ozcricketwriter on April 10, 2014, 3:30 GMT

    That catch clearly touched the ground, so should have been not out. We might think that it is out because he put his hands around the ball and didn't drop it; but the laws as they stand say that if any part of the ball touches the ground it is not out. If you think that this catch is not out, your argument is with the laws, not with the umpire.

  • on April 10, 2014, 3:46 GMT

    @Ozcricketwriter - You seem to be referring to two separate catches in your comment. I only refer to one catch - the one Michael Lumb claimed against Mahela Jayawardene. Which is the second one?

  • waspsting on April 10, 2014, 4:26 GMT

    I couldn't agree more.

    It seems to me that most - almost all, honestly - "experts" don't even seem to know what the law about clean catches are.

    The Jaywardena incident aside, frequently fielders take a catch and then fall to the ground with their palms down, i.e. the ball in hand but touching the ground.

    This is not a catch, as per the rules, but NO ONE - I mean the commentators - even mention the possibility of it not being a catch, which I find astounding.

    For example, the day after the Eng-Sri Lanka game, a Bangldeshi batsmen was caught this way, and not a peep out of the commentators.

    I remember over 10 years ago Jonty Rhodes took a superhuman diving 'catch' but could do nothing other than fall to ground palms down. Geoff Boycott correctly said it was not out but the other commentator thought it was out because there was no other way to have fallen in the circumstances!

    If the 'experts' don't know the rules... there's a serious problem here

  • Insightful2013 on April 10, 2014, 6:21 GMT

    Umpires are the same as referees. I suggest if you cannot recall, use, You tube. England was knocked out of the 2010 World Cup by having a goal disallowed that was at least a yard over the line. Technology all the way, all the time. USA, same tournament, knocked out by a bent Ref from Mali, I think. Humans can never be objective and tools are created to help,assist,benefit, so use them. This is the silliest debate ever! The error was made by Davis, reinforcing the use of better tools and limiting human involvement.

  • FahadBashir on April 10, 2014, 8:00 GMT

    While your article is well thought out and explained, I find it difficult to deal with the unnecessary criticism of ball tracking technology. Ball tracking technology and its predictive path are not based on some computer mumbo jumbo recently indoctrinated into science, but undeniable laws of physics widely accepted since Sir Isaac Newton's opportune discovery of the laws of motion and gravity 400 years ago.

  • sabihoff12 on April 10, 2014, 8:56 GMT

    Good article. Loved reading it

  • Lanky1 on April 10, 2014, 10:01 GMT

    The point is that the blades of grass are not part of the 'ground'. So the fingers may be spread with the grass poking through them but so long as they are under the ball preventing the ball from touching the soil the catch is made. In the palms down case where the ball touches the soil that is plainly not at catch.

  • on April 10, 2014, 10:01 GMT

    Harsha immediately reacts saying, 'Now, where did that pitch?' evenn if it has pitched on middle.

  • py0alb on April 10, 2014, 10:11 GMT

    "a ball lying in two hands that are resting on a grass outfield does not touch the grass at all. For this is what the law requires in a fair catch. "

    The law makes no mention of the grass. the law mentions the ground. The grass is not the ground. The grass is what covers the ground.

  • py0alb on April 10, 2014, 10:21 GMT

    Law 32. b)

    "the ball does not touch the ground even though the hand holding it does so in effecting the catch."

    We could probably do with a clearer definition of "touch the ground" and "in effecting the catch".

    The MCC clarifies: A fielder catches the ball when he takes it in the air from a stroke of the bat. He does not complete the catch until he has established both forms of complete control

    (i) over the ball

    and (ii) over his own

    which seems to sort out the second question, but there is no clarification for what touching the ground means.