Tuesday, October 05, 2010

On the UDRS debate

The BCCI is the last line of defense that the game of cricket has to protect it from being saddled with a very bad technology solution to the problem of improving umpiring decisions. The reasons for BCCI's opposition to the use of Umpires Decision Review System (UDRS) probably do not form a coherent whole - they range from contractual problems with TV Broadcasters, to the discomfort of a few players (mainly Tendulkar) with the idea of players having to ask for a review. But nevertheless they reveal a discomfort that ought to be taken seriously in my view. Everytime there is a situation where an Umpiring decision is seen to be mistaken in a series where UDRS is not in use (Mohali being the obvious case in point), it is automatically seen as a case which make the necessity of the UDRS blindingly obvious.

This isn't the case in my view. The UDRS is a technology of adjudication. It comprises not just of Hawkeye or HotSpot or Snickometer - each of which has been shown to be fallible (and not foolproof), but of the Umpires on the field, the Umpire the Pavilion, the fielding captain, the batsmen and the protocols of communication between them. If we go by the basic definition of a technology - it is the knowledge of using or applying a tool or technique to a given problem, and not by the common usage of the term, in which "technology" refers to only the non-human parts of the apparatus, it is inescapable that the human and non-human parts of this apparatus constitute each other in addition to constituting the system as a whole. So the video replay, or the hawkeye or hotspot or snickometer simulations are as good as the communication protocol regime in which they are used. The extent of their fallibility depends on this.

Is Hawkeye essential for the LBW decision? Are we making more accurate LBW decisions thanks to a simulation which has it's problems - both computational and representational? The answer to this question is a complicated one, and in my view, it is that if we want to believe that we are making more accurate decisions, then yes, we are. But there is nothing inherently more accurate about Hawkeye based LBW decisions. The 'accuracy' LBW decision has a history. The instrumental definition of the LBW law which lists a finite checklist of criteria does not totally define the LBW law, because it's implementation has been dependent on the tradition of umpiring on view in a given game. The great English Umpire Harold 'Dickie' Bird, for example was a renowned "not-outer", especially towards the end of his career. He was not wrong, he just interpreted the benefit-of-doubt more generously in favor of the batsman. The great present-day Umpire, Simon Taufel of Australia, is decidedly more generous to bowlers. In general, Umpires have become far more likely to give LBWs on the front foot. This has changed batting techniques, especially against spin bowling.

The UDRS-supported LBW decision will not therefore be more accurate than the old decisions, it will merely shift the location of the conjecture involved in the decision. But in what direction will this shift take place? This is where the most problematic aspect of the UDRS comes to surface in my view.

Lets consider a situation and apply the UDRS to it. It's the 4th morning of a Test Match. The not out batsmen at the start of play in the third innings are the number 4 batsman, and a nightwatchman. The score is 55/3. At the stroke of the first drinks break, the nightwatchman pads up to an offbreak that hits him marginally outside off stump, on the knee roll. He's forward, but not fully so, the way tailenders tend to be when they're uncertain of the length. The ball has been turning sharply out of the rough for the first 50 minutes of the morning session up to this delivery. The Umpire gives the nightwatchman out. The nightwatchman is 6ft tall, and both he and the nonstriker feel that the Umpire's decision was a bold one. But, the batting side have already used up one review, and are wondering whether it is worth using up another. They decide not to. The Hawkeye simulation shows the ball to be just missing the outer corner of leg stump by a whisker. But, since there was no review, the Umpire's original conjecture about the trajectory of the ball stands.

So what was the LBW based on in this instance? At least 4 distinct bits of judgment - the Umpire's judgment, the non-striker's judgment, the batting team's judgment of the match situation, hawkeye's judgment of the flight of the ball. The original decision was marginal, and even though Hawkeye showed the ball to be missing leg-stump by the slimmest possible margin, it's hard to be certain that the Umpire was in fact wrong. Would the decision have been overturned on review? Yes, since the ball was shown to be missing. If you opted for a more lenient margin of error, then probably No, because then the on-field call would have stood.

The UDRS in this instance would add nothing to quality of the umpire's decision, despite introduce multiple additional judgments to the process. The economy of the review has it's own by-products - the most disturbing of these being the tactical use of the review by teams. The tactical use of the review also brings home the issue of marginal decisions being overturned, for no reason other than the fact that a side decided to ask for a review. The UDRS defines the marginal decision very narrowly in actual fact - in the case of the LBW, this is limited to the margin of error in Hawkeye. It also complicates the marginal decision by removing it from being the sole prerogative of the Umpire, to making players complicit in it. There is no place for this complication in the rhetoric of accuracy which is used to market the UDRS (see Ricky Ponting and Virender Sehwag's comments in favor of UDRS at Mohali)

The UDRS is a technology of adjudication, but evidence shows that it is a deeply problematic technology of adjudication. What new information do the players bring to the situation? Limiting the reviews to two is a pyrrhic solution to a problem of UDRS's own making - the fear that teams might bring the game to a grinding halt by questioning everything that goes against them. If it is really a matter of teams using it well, then is it also not introducing a new element to the contest? Does cricket want cricketers to now also be experts at questioning the umpire's decision? If it is put this way, I think most players would be horrified, even after some reflection. It is never put this way.

As a technology then, UDRS's problem is that it has too many moving parts, too many competing interests. What might a simpler system look like? One of the most underestimated aspects of the UDRS is the communication protocols it sets up. Players can initiate a review, even though they don't always have good information. The third umpire, the one entity in the entire apparatus who is information-rich - on three important counts - he's an expert as an umpire, he has video equipment at his disposal, and he has the expertise to interpret what he sees intelligently and come to a conclusion, is not allowed to initiate anything. The on-field umpire is not allowed to initiate an inquiry under UDRS, but is allowed to do so under other circumstances.

A simpler situation might involve on the On-field Umpires and the Third Umpire. Both would have the ability to initiate communication about a decision (either out or not-out). The Third Umpire is best placed to determine whether an on-field decision was marginal (in which case he will not interfere) or reasonably certainly disputable (in which case he will). There are fewer moving parts here, and no conflicts of interest on account of the match situation. A possible protocol would run as follows.

1. The fielding side makes an appeal.
2. The umpire on the field, as usual, makes judgment as to the merit of the appeal. If he's sure, he gives it Out or Not Out.
3. The TV Umpire is simultaneously looking at it, trying to determine whether one of the following three applies: (a) the umpire is clearly right, (b) the decision is marginal or (c) the umpire is clearly wrong.
4. If it is (c), the TV Umpire can point this out to the on-field umpire, who can reverse his decision. If it is (a) or (b), he does nothing.
5. If the Umpire on the field is unsure, for example about an inside edge - it looks like a good LBW shout, but for this doubt, he could make a provisional decision, and then consult the TV Umpire to either consult or reverse his decision.

This is a better way to separate the marginal decisions from the obvious errors. This separation is a crucial (even central) problem that has to be addressed in any technology of adjudication, and the UDRS only addresses it in a very cumbersome, roundabout way by setting up an economy of reviews.

There's no necessity for these decisions to be communicated through the big screens - wireless communications are sufficiently advanced for the umpires to communicate directly in a very reliable way. This way, broadcasters will not be able to waste time building up suspense through gratuitous graphics which end with "Out" or "Not Out". It will also withdraw the batsman right to appeal, and limit the fielding side to asking the question only once. It will eliminate the tactical economy of the review system.

The UDRS is facing resistance not just for contractual reasons (Existing contracts between broadcasters and individual host boards will have to be revised to accomodate UDRS), but also because it is unnecessarily complicated technological solution to the problem. The ICC would still have to mandate what software technology should be available to the Umpires. This is a separate problem, one that UDRS does not address. The UDRS is not going to produce more accurate umpiring decisions. It may correct some obvious errors within it's limitations (2 reviews etc.), but it will also radically reconstruct the way decisions are made on the cricket field, and radically relocate authority.

It is time for the ICC to take the resistance to the UDRS seriously, and not merely condemn it as contractual intransigence or technophobia. There are simpler solutions available. The BCCI, possibly unintentionally, is doing the game a great service by resisting UDRS. Now it needs to take the next step and constructively propose something simpler.


  1. I saw a lot of the UDRS at three Tests in England this summer. It's not perfect, but it's worth the problems to avoid awful errors like Gambhir's and Hussey's.

    Another factor is the number of televisions around English grounds. Television viewers (rightly - we're paying big money) demand quick and comprehensive replays. 25,000 people at the ground would know there was an edge and one wouldn't - the umpire who raised the finger. Totally untenable.

    If it were Sachin on 98 in Mumbai, there would be a riot. Maybe at Edgbaston too with Ian Bell vs Australia.

  2. hey , nice blog , like it ,
    won’t be nice if i u can clickover to my blog page too ,
    & post some suggestion

  3. Would they improve upon the mediocre umpiring that is visible? Forgetting for a while their lack of accuracy at the nth degree? If the answer is yes or likely, I'm ready to go with it.

    It isn't a perfect system, has its flaws and limitations, but it is an improvement over what exists. Either umpires do a better visible job or take whatever help there is to begin with. And I am sure, the system will evolve once in place.

    Or we could all close our eyes or our HDTVs and slo mos and imagine ourselves to be too far from the action to bother about the handful of poor decisions per innings.

    Maybe cricket broadcasts must include a disclaimer, at the start and through every match, for audiences to ignore replays and such and carry on watching in the spirit of the game.

    The sport and television will then revert back to be truly the opium of masses...we can always imagine we saw a great contest in it.

  4. Easier said than done. What is to happen to the dismissed batsman in each case? What you propose relies on lightning quick confirmation/rejection by the 3rd umpire, or else there will always be a necessary delay for every single decision that is made by an on-field umpire while the 3rd umpire goes through replays, which aren't provided independently by the broadcaster and often take a while to be viewed. During this time the 3rd umpire must still check to see that the ball was legally delivered before moving through at least one camera angle. There are enough stop-go delays in cricket, what UDRS tries to achieve is an economy in referring a decision and it should ideally be used only to correct obvious blunders. I haven't checked the stats, but I'm sure soon teams will realize that making a tactical for a 50% decision is a waste of the system. The system is also open to abuse, say when a team has retained both referrals and uses them up for the sake of using it when they are near the close of an innings. That said, it will still be less than the number unnecessary double-checks by 3rd umpires if we were to see him communicate with the on-field umpires as per the system you propose.

  5. It's too narrow to think of UDRS as an improvement compared to the status quo. It's far more complicated than that. I dislike cost-benefit analyses, mainly because non-marketed things either get priced arbitrarily or ignored altogether, and hence dislike narrow, reduced value judgments that ignore complications.

    My whole point has been to point out the complications of UDRS. I rather hoped that this would be received in that light. That the UDRS has been approved is not in dispute. That it is a method intended to reduce errors is not in dispute either. The questions are as follows:

    1. Is it a good way to reduce errors?
    2. What are the consequences to cricket of this system? How does it change the relationship between players and umpires? How does it change our conception of an "accurate" decision?

    That's what i've tried to address in this post.

  6. On this business of slowing down the game.

    1. If there are frequent appeals, then the game is not in danger of running out of time, because it is possible to conclude that a large number of appeals suggests that batsmen are in trouble.

    2. If the average umpire makes 3-4 obvious errors in a game, why would the third umpire initiate contact more than 3-4 times? If an umpire is in doubt 3-4 more times, then there would be about 10 such communications over the course of a game.

    3. If there aren't frequent appeals, then this would not happen very often at all.

    So I don't think the time-wasting argument is plausible.

  7. It is not so much time wasting as it is the necessity to waste time while reviewing each decision with the same degree of care, not just the ones which the on-field umpire is unsure of. What you are proposing is for intervention from the 3rd umpire, but such intervention takes time and demands he reviews all cases irrespective of whether the on-field umpire has made the request (which still happens today). Should he pick and choose about which ones he wants to review we go back to square one where people talk about subjective reviews when this system fails, say events similar to Chris Broad's meting out of punishment. In case the point was put across properly, again I ask, what does the batsman who has just been dismissed do? Stick around for a few minutes in the hope that the third umpire will review the action and spot the bowler over-stepping? Does he walk back to the pavilion only to be re-called in each case? So you see, even for as simplistic a protocol as you recommend, there are complications. UDRS while not perfect, forces (or should force) teams to make judicious use of referring the absolute howlers. In this way it minimizes delays and talk of subjectivity better than most other systems.
    Also note that it is not UDRS that you are picking on in the first few paragraphs of this post. The 3rd umpire gives the on-field umpire an appraisal of the exact turn of events in the case of an LBW, he does not overule it based on what Hawkeye extrapolates. He tells him where the ball pitched, where the impact was and other pertinent information like whether the batsman hit the ball. It is upto the on-field umpire to reverse his decision if he believes there is significant reason to change it, and as you might have noticed in the recent Eng-Pak series, they avoided reversing decisions that were marginal and stuck to the original one.

  8. Ach, it depends how long it takes doesn't it? Suppose hawkeye is effectively immediate (and I believe it is) and there was a method of determining no-balls as outlined in your other post. You could give an umpire a $500 smart-phone running an app connected to the ground's wireless network that flashed up, every ball, four things:
    no-ball (yes/no), pitched outside leg (yes/no/your call), pitched in line (yes/no/your call), hitting stumps (yes/no/your call).

    No need for UDRS at all, as the umpire could have a quick look at their phone to confirm and then use their judgement. That only leaves edges and close catches, but replays have proven to be no better than an umpire at judging these things anyway, so why bother? (If technology existed to determine edges, perhaps a sensor strip on bats, then that would be covered too).

    The problem with UDRS isn't the idea of checking a few things, it is that the method chosen is so cumbersome. The technology exists to convey the information to the umpire immediately, yet they are still out there using walkie-talkies.

  9. Russ, it wouldn't even be too hard to put some sort of graphic on it, would it? If desired, you could show just the observed path of the ball without trajectory, giving the umpire more accurate information than his sight and memory, while still allowing for Kartikeya's concern about an massive change in the effective criteria. (Of course, you could also pursue a similar purpose objectively by choosing a larger "benefit of doubt" margin than the current one, which is actually significantly greater than the reported error. [For Hawkeye at least - anyone seen anything about Virtual Eye?])

    Ach, it seems to me that Kartikeya's definition of marginal decisions and obvious error's is the major difference between him and those who push UDRS at any opportunity. If you have some level of trust in the technology, then any evidence you can accordingly gain from it is an obvious improvement to the decision. Kartikeya's position is a bit closer to affirming any decision that could be justified by an umpire with no extra tools, which is actually more inline with structure of the UDRS.

    Kartikeya, I think I share your concerns about the role of players, but unless we are talking about a time-consuming review for every appeal (assuming that we are still only talking about wickets), then it is relevant that the players are sometimes in a better position to know whether the decision is wrong than someone with a tv. That doesn't fit well with the ability to make tactical requests, but it's hard to avoid that conflict.

  10. To clarify - I didn't mean to imply that Kartikeya has no trust in any of the technology, rather that where some would say any obvious improvement to a decision is a good thing, he doesn't think it is an obvious improvement to the game overall.

  11. Jonathan, it depends how intrusive you want to the technology to be. It would be quite easy to provide a "second look" approach that lets the umpire look at tv replays (admittedly on a 4 inch screen) for all decisions. But watching the central umpire arse around on his smart phone while he makes a decision would be no more entertaining than waiting for the third umpire to second guess him.

    I'd prefer a "quick glance" approach that basically lights up whether the decision is clearly out/not out, but otherwise lets the umpire get on with it. I also think, in regard to lbws that there is much to be said for giving a batsman out when he looks out (as in, is comprehensively beaten), provided it is "close enough" and vice versa.

    Maybe its a bowler thing, but there is nothing more central to cricket than a huge appeal, the pause and the raised finger. We should try and preserve that.

  12. The point of introducing UDRS is to eliminate obvious errors. But given the way it's set up, it introduces too many other changes. I agree that in certain situations the players may have good information (for example inside edges), and in a significant percentage of cases they do walk (such as when they're bowled or caught at cover), but should that be used? Shouldn't the information available to the third umpire be used first?

    On Technology - Im cautious about believing that technology takes us closer to the truth (especially technologies like Hawkeye). Yes this technology is probably more compelling than a video replay, but it is still just a newer, apparently more sophisticated reconstruction.

    The quick glance approach, as Russ suggests, shouldn't take too long, and what's more, there are unlikely to be too many reversals because Umpires don't make too many mistakes.

    In any event, we always see a replay of every single appeal before the next ball is bowled. The TV Umpire can see the same. In the vast majority of instances, he will find that the Umpire is either right, or it's marginal. Unless the umpire is obviously wrong, the game goes on. There doesn't need to be an active positive affirmation that the umpire is not obviously wrong every time.

  13. Kartikeya, if they implemented a "quick glance" solution you wouldn't need the third umpire because the central umpire would have all the available information before he made a decision.

    It would also be expandable as more technology became available. There is no reason, for instance, why the ICC couldn't mandate bats had a thin touch-strip on the edge, that communicated to the umpire that there was an edge.

    It is also ridiculously easy. Communicating the results of hawkeye's lbw assessment to an umpire with a smart phone is literally a day's work.

  14. True. I think some redundancy will be necessary. So the Third Umpire should still be around.

  15. UDRS is a must in order to eradicate howlers... Hawk Eye is pretty pathetic and should not be used for Leg before decisions except if the ball pitches outside leg stump decision. Rest like snicko & Hotspot are very helpful technologies.

  16. There is only one reason why Sachin Tendulkar and the Indian cricket board are not in favour of UDRS.

    India played 3 tests under UDRS with Sri Lanka and we lost the series I think 2-1. Sachin that year scored over 1000 runs but in the UDRS series made 95 in 6 innings at an average of 15+

    Among all the batsmen today in world cricket if you did some serious indepth research and looked at all the benefit of doubt umpiring decisions in favour of them wherein they were actually out, the handsdown winner would be Sachin "The Desi God" Tendulkar.

    He knows it. The Indian board knows it. Other than that test matches would be shorter and the entire set of batting records would just disappear.