Couch Talk Episode 38 (play)
Guest: Kartikeya Date, “A Cricketing View”
Subscribe to Couch Talk podcast on iTunes.
–Hello, and welcome to Couch Talk! Today’s guest is Kartikeya Date, of the blog – The Cricketing View. Kartikeya, through his relentless and diligent work, has written a vast number of insightful posts on the Decision Review System (DRS), and he believes that DRS is fundamentally changing how cricket is played.
Welcome to the show, Kartikeya!Kartikeya Date – Thank you, Subash. Thanks for inviting me again!
SJ – Pleasure! You were one of the early guests on this podcast show. And that was in July, last year. We discussed the DRS back then too. How do you think the DRS, the arguments and the discussions around it has changed in the alst 9 months?
KD– The quality of discussions about DRS has substantially improved. In the last 12 months, Cricinfo has this In Focus series called “Technology in Cricket”, in which they’ve been writing carefully reported articles, much more frequently in the last 9 months than they did before that. You’ve begun to see reasonably skeptical articles – for example, Greg Baum had an article about DRS at the end of December. Some players have come out and talked about it. We’ve had much more diverse opinion about DRS, and ball tracking particularly.
I think the debate has progressed substantially in the last 9 months. When India were in England, there were two debates going on. One was institutional debate – BCCI vs The Universe. The other was the far more substantially interesting debate, which I would call- the debate between the technophiles and people who are more skeptical of technology. The technophile position,(and more than 1 person actually said this) that flawed technology is better than an unreliable human being, which is to say that umpires make mistakes and so technology is better than umpires. The skeptical position, which I’m more sympathetic to, would say well, “What we are measuring, and what is actually being measured, and given the difference between these two, what are the unintended consequences of what we measure?”
From here, I think several conceptual issues arise. How do you determine whether the technology is reliable? How do you determine whether it is accurate? What is the difference between something being “reliable” and something being “accurate”? What is the umpire’s call, exactly? All these interesting conceptual issues come up.
SJ – When you say reliability and accuracy, do you mean to say technology is reliable, but we don’t know if it is accurate?
KD– Yes. Basically, if you type in a query, ask a computer to say what is “47 x 3”, it is going to say the exact same answer every time. But, with a human being, you don’t know. If you ask them to multiply two numbers, even a person who knows multiplication will make a mistake (from time to time). There is a basic difference between errors and mistakes that needs to be understood while talking about accuracy and reliability. A disembodied technical system can be reliably inaccurate, and very consistently inaccurate. Whereas, a human being can be both inaccurate and also can make a mistake. For example, if an umpire sees an lbw decision in real time and makes one decision, then there is a very good chance that if he sees it again on a replay, he might change his mind about it. If that’s the case, then his original decision is a mistake. Whereas, if the umpire looks at an on field decision on a replay, and agrees with the original decision that he made on the field, then that (the original decision), I think, is not a mistake. The umpire, despite having many looks, has come to one conclusion about it. Now, if you take the same ball and put it through a ball tracking, and the ball tracking might say it had missed the leg stump by a whisker. The umpire is absolutely sure that in whatever mode of judgement he uses, he’s right. And, the ball tracking is equally sure that it is right. No matter how many times you process the video image, it is going to give you the same result – missing by a whisker. They are going to disagree, make opposite calls.
This, to anybody who understands marginality, not too surprising. Whenever you make a rule, by definition rules are restrictive, there are going to be margins. So, what’s happening is that, errors and mistakes, in my opinion, are not being distinguished well enough in the DRS. This is my conceptual diagnosis of what’s happening. What we’ve seen in the last 9 months which has also added to our corpus of empirical evidence about DRS is that this seems to be the matter of dispute. If you see Jacques Kallis’ recent comment, that (paraphrasing Kallis here) you look at something and it looks close, and then DRS says it is not.That’s like a classic statement of the difference two different modes of judgement and reaching two different modes conclusions. So, to say that we are going to judge if one is more accurate than other – its hard to do that. In order to accept that it is hard to that, I think, it is necessary to accept the conceptual difference between the two modes of judgement. There is no reasons to thin that human umpires process reality in the same way that the disembodied video camera apparatus does.
SJ-Dave Richardson, the man in the middle of everything – he is ICC’s General Manager of Cricket. He had said, in an interview with Cricinfo’s George Dobell, “DRS has affected the game slightly more than we thought it would.” This was in the wake of the England Pakistan test series. Also, it came to my knowledge that you had tried to contact Dave Richardson and tried to get some answers in the follow up of this interview with Cricinfo. What’s your take on the statement, and what have you learnt henceforth?
KD– The impression that I got from Dave Richardson’s interview, was that they have shifted their narrative about what the purpose of the DRS is. The traditional understanding is that the DRS is to fix howlers, like obvious umpiring mistakes (not errors, but, mistakes).It came across in the interview that they are saying that one of the ambitions was to question the whole notion of ‘benefit of doubt’ and to see whether that could help redress the balance between bat and ball. I felt, at that time, that this way of reading the interview was a little bit harsh on Dave Richardson. So, I tried to ask him.
What Richardson says, is that DRS has had impacts at multiple levels. For instance, what happened at one instance is that, before DRS was used, ball tracking was still available, right? So, umpires were being assessed by ball tracking even before DRS was being used. So, if an umpire gave something not-out, and the ball track was showing it to be hitting, then he would look at that and say, “Maybe I should give that out next time I see a similar LBW appeal.” But he also says that the trend (in giving more LBWs) started before DRS for this reason that umpires were looking at ball tracking results and rectifying their conventional judgement, of what they would consider to be a safe measure.
The second thing that happened is that once DRS came in, umpires started to feel, “If I give this out and it is not, it can easily be rectified. So, I don’t have to be as cautious as I used to be.” That was another mode that came in after DRS, in the thinking.
The third, and the most interesting thing he told me, was what you might call the “Globalization of umpiring” (my phrase), with the Elite Panel coming in. It takes a little while understanding this because you have to be used to Elite Umpires umpiring all over the world to understand this. What happens is, the umpiring conventions get developed in different countries. A long time ago, it used to be that umpires in England used to judge front foot LBWs differently from the umpires in Australia. This was a kind of conventional view because wickets in Australia are bouncier.
SJ– You mean to say umpires from sub-continent would judge LBWs differently form umpires from non-subcontinent?
KD– Yes. But then, when an English umpire came to India to umpire as an Elite umpire, he would apply the same standard to India. So, maybe some LBW that an Indian umpire might have given, the foreign umpire may not give, and vice-versa. You carry your conventions with you, which are dependent on pitches and general trend of pitches (that you are used to). And, this was especially apparent in the England-Pakistan series where, he said the bounce was relatively low, which was why there was an increase in LBWs which the DRS is picking because of the way the review is designed. From what he said, it is true to say that the point about bringing in DRS was to eliminate howlers. But as any technology skeptic would say is there are unintended consequences, and these are the unintended consequences. And in this sense, he is kind of right. He also said, “If we are totally honest, DRS has affected the game slightly more than we thought it would.” And this is more or less what he told George Dobell as well.
SJ– The England – Pakistan series that Richardson was referring to, that series saw 45 review/ challenges of umpiring decisions, out of which 43 were LBW appeals. Of the 45, only 8 of the on-field calls were overturned and none of those would be what people would’ve termed as “howlers. And, Kevin Pietersen mentioned during the series that batters are not getting the benefit of doubt any more, especially with the front foot LBW. Three of four years ago, they would never be given out, but now they are given out. And hence, the way they approach spin bowling in different conditions had to change. Meaning, essentially, batting technique had to change. A batsman who had been batting for ten-fifteen years, born with a technique has to suddenly adapt to something completely new. Your thoughts on the change in batting techniques?
KD– If you are given a thought experiment, saying “What if you introduce something like DRS to cricket? What will be the consequences?” Then, this is one of the things that you will think about. I thought this would happen, way back in 2009. I have a post on it on my blog way back then. that is an avoidable problem, what Kevin Pietersen is saying. And, in my view, the problem is a function of the fact that even though (and,I can say this categorically, since I asked Dave Richardson about the whole error vs mistake problem. I defined what I meant by a mistake to him, and asked what ICC is trying to do. Is it trying to fix errors or is it trying to fix mistakes?) He told me, “By your definition, we are trying to fix mistakes.” But, having said that, the way the protocol is designed, it does not do this very well. And that is because it basically fits the umpire’s mode of judgement, not the umpire’s actual actual judgement but the method or mode of judgement, with ball tracking’s mode of judgement and it uses the ball tracking system to verify the umpire’s mode of judgement. It does not similarly allow umpire to verify the mode of ball tracking mode of judgement.
Did you read the interview that Ian Taylor gave after Jacques Kallis’ comments, in which he said “Well, umpires should be allowed to over-rule the ball tracking predictions.”? He was limiting his comments to the point that the ball tracking system has insufficient information. I think the argument can be extended conceptually without stretching its logic, to say that umpires use a different mode of judgement than the ball tracking system. So, if the (TV) umpire thinks the (on-field) decision is close, even if the “umpire’s call” zone in the ball tracking system disagrees with the umpire’s judgment, (TV) umpires should have the right to uphold the on-field umpire’s decision. So there needs to be an intermediate step, where a human umpire, after looking at the replay, looks and determines whether possible error has been made or whether a mistake has been made.
This is hard to do. So, the way I will do it is as follows. Suppose there is a decision being made in the field. It’s a marginal LBW decision. It is an off-break that pitches on off stump, turns a little bit, the batsman is caught at the crease and when the batsman is hit, you can sort of see a part of middle stump, and the whole of off stump. The ball hits the pad and rolled to the off side. There is some doubt about whether it would hit the leg stump or go past the leg stump. Suppose the umpire on the field gives it out, and you know what example I am referring to.
SJ – yes, the World Cup semi-final.
KD – Yes, the World Cup semi-final Sachin Tendulkar LBW decision. Suppose that’s the case and the umpire on the field gives it out. What would a TV umpire look at on the replay? He should have a look at that and first determine if this is a mistake. the TV umpire makes a judgement that if an umpire saw this ball, it is reasonable that he would rule it the way an on-field umpire did? If you ask this question, and say “No, it is not reasonable. The decision is obviously wrong.” There are instances where you look at the replays and say “This looks very marginal on whether it pitched on leg stump or not.” In that case, you go to the technology and verify that, you can verify that. But the first judgement, about the predictive part should be made by the umpire, about whether it is obviously wrong or not. If it is determined that a mistake is not being made, then the decision given on the fieldshould stand as the correct decision.
SJ– So, you are saying that the predictive path from the point of impact to the stumps should not even enter the equation, when the second umpire looking at this thinks a reasonable decision has been made, then the predictive path of whether the ball would have hit the stumps, the graphical recreation of it doesn’t enter the equation?
And now, how would this be an error? If the umpire determines the mistake has not been made, and lets the decision stand. But, because the broadcaster owns Hawk Eye and they show Haw Eye for every delivery, and later on they show Hawk Eye for the delivery, and say that it is missing leg stump by a whisker, then you technically says that the umpire has made an error. You can also say, symmetrically, that the ball tracking made an error. There is a symmetry of error there because there are different modes of judgement made there. There is no scientific or philosophical reason to believe one is superior to the other, because both are making predictions an umpire will make an informed prediction based on his experience and expertise and based on institutional standard of correctness that emerges in umpiring community.
SJ– Here is my question – There is no way to test the effectiveness of the ball tracking technology, whether the ball will go on to hit, because you are actually only judging the umpire based on the assumption that the ball tracking is perfect. In the interview we talked about earlier, Richardson talked about the company called “Computer Vision Consulting Ltd.” Which is staffed by post graduates from Cambridge universities that is carrying out tests on some of the aspects of the DRS. If my understanding of the article was correct, then it was the ball tracking. What do you think is being done? Because, it still doesn’t make sense what they are testing ball tracking against.
KD– That is not yet clear to me, and I asked them that. When I interviewed Paul Hawkins, founder of the Hawk Eye, he told me that you could actually design an interesting field test which could maybe look at how ball tracking performs in field conditions, somewhat simulating match conditions. He said that you could do that, but you (ICC) have to pay for it. ICC seems to have gone in the direction of testing how well the prediction is made by the ball tracking system using some sort of computer vision related technology. I asked Dave Richardson about this, and he told me that they are asking them to look at the accuracy and reliability of Hawk Eye. They do this by using intensive computation to reproduce a finely calibrated model of the pitch and the path of the ball using all the available HD video. There are many different angles of HD video available for each delivery. The reason why this is not used in Hawk Eye itself is that, whatever they do, it is going to take about 45 minutes to an hour of processing time to actually compute the path of one ball. My best guess is that they are going to use a much more accurate but much less efficient algorithm to judge how well the Hawk Eye algorithm works.
And this is fine, it is not the problem. The problem is not that Hawk Eye is not accurate. The problem is that it does things in a different way, and it is not surprising that batting techniques are changing because of it. Now, whether you want batting techniques to change – that’s a choice. That is a normative choice for the ICC. Do they want cricket to change this way? If they do, then that is fine! They could just use Hawk Eye as it is. If you think about it, when Ian Taylor said (that sometimes umpires should be able to over-rule ball tracking conclusions) what he is saying is that if the ball tracking makes a mistake, not an error but a mistake, then umpires must have the power to overrule. Another option is that, if the ball tracking has insufficient data, it should be able to return that as a result. It must be able to say “Well, in this instance, you are on your own, there’s not enough data to compute.” You can imagine the problems it could create. Suppose at a crucial moment, the ball tracking goes off and the batsman goes on to make a century, or gets out and the batting collapses, etc. When these things go into the world, the choices have different valences. You have to be careful. Increasingly, I don’t particularly care care about whether they use DRS or not. What I care about is whether they understand what they are doing by using DRS. Whether they understand what choices they are making.
SJ– From the evidences that we have had from the last three or four years, I’m not really sure that they understand it completely.
KD – That is because they are still chasing the most accurate decision, and that is not the right thing to be chasing, I think
SJ – Recently, in the SA-NZ series, damning of DRS pretty much happened and it didn’t catch most of the cricket enthusiast’s view because there was something more interesting happening at the other side of the world far too big. Jacques Kallis said that he was not convinced by the predictive path of the delivery shown by the ball tracking technology. Doug Bracewell said that members of the NZ squad have also been iffy about the accuracy of the DRS. What does this do to the DRS argument. We have had the Indian players, back in 2008, not being sold on the efficiency of the system. England have raised some questions, Andy Flower has raised some questions now. You have Jacques Kallis, one of the preeminent cricketers of the generation, questioning it. And now you have New Zealand, another team not being sold on it completely. What does this do to the DRS argument?
KD– Jacques Kallis actually went even further and said that he did think 99% of the cricketers did not believe in the DRS.
SJ– I would like to know whether Jacques Kallis did a poll of 100% of the cricketers to come to that conclusion.
KD – It is interesting that he said that. If you see that these players have played together for years and have known each other, and that they are having conversations about DRS. I’m sure they are, it would be incredible that the players don’t talk about the DRS amongst themselves. If Kallis, in some poetic sense, (in the same sense that you have the 1% and the 99%), feels using 99% to refer to his perception that the majority of cricketers don’t like the ball tracking prediction, then, I think it is significant. But then, I think it goes back to the issue of accuracy, to the issue the ICC is still chasing the most accurate solution. It is not clear as to what the consequences of this so called accuracy is going to be. In the beginning we were talking about how Dave Richardson thought ball tracking was affecting umpiring.
Even before DRS was implemented, before 2008, the umpires were going back and looking at the ball tracking predictions of the decisions they made, and finding that they needed to be corrected. Now, we know for a fact that the ball tracking predictions were very, very poor during the mid-2000s. We know for a fact that for a long time that ball tracking prediction used a generic point of release for every delivery. They did not actually track whether the bowler is tracking from wide of the crease or not, etc, in the early days. The fact that you had the animation, and the fact that it looked so convincing, who knows if the umpires were actually wrong, and whether that bad quality ball tracking system actually merited the attention that it got from them. Umpiring conventions change and did so even before the advent of DRS. The advent of Shane Warne and Anil Kumble and the revival of significant spin bowling that happened in the early 90s, that has an effect on the number of front foot LBWs the umpires have been giving. I think the rise of DRS also has an effect.
What is considered a “reasonable decision”, what is the convention for a “reasonable decision” is fairly stable, even though it has a history. You very rarely have a problem where one umpire’s decisions are totally at odd from other umpires’ decisions. They generally follow the same convention, and agree that a ball is missing the leg or off stump or the ball turned too much or too little. These are sort of the nuts and bolts of an umpire’s mode of judgement.
SJ– Where do you think, in your opinion, as someone who has spent inordinate amount of time understanding various sides of the DRS equation, both technically and philosophically, where we are as a sport, and where this DRS is headed to? And more importantly, where does this leave the umpires?
KD– If it goes in in the direction it is going now, I think the umpire is already obsolete.. Because if the umpire’s judgement is being verified by the DRS or the ball-tracking, then the umpire might as well not make the decision in the first place. Why not, whenever an LBW appeal is made, just look it up on the DRS and determine? Why have the umpire make a decision on the field in the first place? What is “Umpire’s Call” in DRS right now? Umpire’s Call is basically this. The Umpire makes a decision, the DRS looks at it and says that it cannot decide because too many things are on the margin to decide whether this is out or not; and so, let us go with the umpire’s decision. But, that’s not the umpire’s call. The umpire’s call is determined in terms of the ball tracking system. Now, there is no reason why this should not be reversed. Why is the umpire’s call not determined by an umpire? Right now, it is not! Right now, the umpire is obsolete in LBW decisions. So, whether an umpire’s decision is right or not depends on whether a player questions it and secondly, what the DRS says.
There is no umpire’s call anymore. The umpire is already obsolete, LBW wise. And it is not a surprise. You cited the statistic of Pakistan- England series where 43 of the 45 decisions were LBW. There was a statistic posted by S. Rajesh on Cricinfo about the World Cup, which had similar statistics. And that was One Day cricket. About 80% of the DRS requests were about the LBW appeal. I think it is not a surprise that most of the disputed issues are LBWs and it is not a good sign for the umpire that the vast majority of the DRS requests relate to an umpire’s judgement about LBW, which is then verified by the technology. This is the key-point – why should the technology verify an umpire’s decision? Why should an umpire not verify an umpire’s decision? This has always bothered me about the DRS. Because, if you really think that umpires can make mistakes, then obviously the solution must be to let another umpire have as many looks at the decision as he wants. And I don’t buy the argument about time, because the DRS takes time as well. But, they have not chosen to do that. They have chosen to directly have the umpire’s decision reviewed in terms of the ball tracking system. And I think that’s basically, the end of the umpire, right there.
SJ – So, where are we headed, in terms of the cricket, and the effect of cricket in terms of the DRS on the game itself.
KD – I don’t know. It’s hard to say, but we are head to a situation where the umpire is basically just counting balls. Effectively, that is what they are doing right now. The protocol has several other issues, like it sets up an economy of error by limiting two reviews per innings, but those are all small things. The basic philosophical point that they have made is to say that they are going to use this technology to verify whether the umpire is right or not. If that’s the way they are going to go, if I was an umpire, I’d be very upset about it.
SJ – on that bright note, thank you, Kartikeya. Thanks for coming on the show and it’s always a pleasure talking to you!
KD – Thank you. My pleasure.
SJ – Thank you all for listening.
Episode transcribed by: Bharathram Pattabiraman