Monday, 7 March 2011

More DRS rubbish

The latest twist in the DRS use of Hawkeye when the batsman comes down the pitch is hard to get a grip on, mainly because the reporting seems amazingly confused. There might be something we can get form the latest Cricinfo article, but it's hard to know exactly what, since it's riddled with nonsense.

Why do I say that? Let's start with last paragraph boilerplate that CI has been adding to all their article on this topic.
“The 2.5m clause was included in the DRS rules following the expert view that the ball-tracker technology, in this case Hawk Eye, lost its accuracy when the distance between the point of impact and the stumps was greater than 2.5m.”

The only expert view that seems to be published online is that of the proprietors themselves, and they firstly consider the 2.0m mark more relevant than 2.5m and more importantly don't think the loss of accuracy is great enough to make that much difference, saying the 2.5m rule is their for historical consistency, not technological reasons. Yes, the CI line follows the recent comments from the ICC, but they are speaking just as much nonsense (if accuracy were the issue, it should be relevant to “out” decisions as well), and journalists really should be holding them to account. If there was an altogether different expert view given at some point, is there any reason they can't publich it and point us to it?
The article also says that there has been a change to “rule 3.3” in the DRS code, and that the previous rule said a leg before could be reversed “only if the replay showed that the ball was hitting the middle stump dead centre.” Even if we assume that the writer meant “a not out leg before decision” rather than simply “a leg before”, there are a lot of problems with how the article describes this:
  1. The 2.5m rule (part of clause 3.3i)iii) can't by any stretch be related to a necessary condition for a not out decision to be reversed. The clause deals with when the umpire should be told definitively that the ball was hitting (or missing) the stumps. This is not the only factor.

  2. There is clearly no mention of a middle stump criterion of any sort in the DRS code - the decision is to simply said to be made using “normal cricketing principles” informed by the ball tracking data. Something like “only if the ball is hitting middle stump” has been popping up a lot in discussion of the Ian Bell decision, but it's not at all clear whether it comes from a less formal umpire's directive (for this tournament or more generally), an off-the-cuff press conference example of what might qualify as normal cricketing principles, or a player or commentatorsa (possibly hyperbolic) interpretation of normal cricketing principles applied down the pitch LBWs.

  3. We are told that umpire Erasmus asked whether the ball was hitting “any part of middle stump”. The box claiming to detail the new “law” tells us that it has to be the centre of the ball hitting any part of middle stump. If that's the new guideline, then how restricted was the meaning of “dead centre” in the reported old one? In any case, by my eyes the situation shown in the graphic accompanying the article doesn't meet that criterion anyway - if this is the Cusack referral, perhaps they really mean that any part of the ball is hitting any part of middle stump?

  4. The box with the claimed new law has also clearly misread the code. The wording which is said to have been replaced comes before the 2.5m exception, and is about the more general condition for reporting (and effectively determining) that the ball was hitting the stumps. (It also isn't anything like “hitting the middle stump dead centre” - a much bigger area than middle stump is described!)


So what can we read between all the errors? There is talk of an umpire's directive, described as changing the protocol. I'd guess that the umpires manager has given direction to the umpires that a trajectory hitting middle stump (in some sense) can be considered out (in the absence of other reasons for a not out). It's not a change to clause 3.3 – it is easily seen as a clarification of a "normal cricketing principle", set in stone as a kneejerk reaction to the predictable inconsistency in interpreting that phrase. It's possibly a change to a similar earlier directive, although that doesn't seem likely to me. It's not all that strange - I've certainly heard similar sentiments (“That far down, I'd only give it out if it were hitting middle.”) from umpires relying on their own sight, and especially if technological accuracy really is the issue, the logic transfers well.

So it looks to me that as well as spouting clueless press releases, the ICC has had made a small concession in the name of consistency in response to the media drama round this issue, and this itself hsa been beaten up. Then again, there's so much we can't know, and so much rubbish in the reporting, that I may well be wrong.

Tuesday, 1 March 2011

Stepping down the pitch to DRS

There's (unsurprisingly) a lot of talk about how the DRS dealt with the LBW referral for Ian Bell, where umpire Billy Bowden chose to stick with his not out decision after the tracking system "predicted" the ball was going to hit the stumps. I'm going to take my own stride down the pitch on this topic - you judge whether I'm out, stumped or LBW!

The way the DRS works, the decision is ultimately up to the umpire, no matter where the point of impact is, but according to the regulations, if the point of impact has been less than 2.5m, the third umpire would simply have told him it was in line, hitting the stumps and so on, and I can't imagine that being ignored. However, in the over 2.5 m situation, the umpire not only given more details concerning distances and wherhe the ball is projected to go, but is told to use "normal crticketing principles" in deciding whether to overrule his original decision.

Cricinfo says that the reason for the 2.5m distinction is because that's where the predictions become less reliable. Of course the reliability of the trajectory prediction does depend on factors including distance from the stumps, and it's easy to believe that this was a factor in the minds of some of the people involved in accepting the guidelines. However, while the Hawkeye reports of MCC testing don't in my opinion rigorously address this issue, they also give reasonably different criteria for their "extreme LBW" classification, hinting (along with the dependence on the original decision being "not out") that this isn't what the 2.5m is really about. In fact, Paul Hawkins says the main reason for it is to ensure that the traditional dispensation for batsmen coming down the pitch "continues to be the case", even when the benefit is no longer founded in quite as much doubt.

The relevant document form Hawkeye is found online, and was first brought to my attention by Kartikeya Date. He has used the traditional approach to LBWs as a reason to oppose the use of technology for LBWs, and while he argues against the DRS even in this form on several grounds, I gather that he thinks it is better to include this clause than not.

In some ways, the 2.5m rule seems odd, but it is one of the more logical of the current systems oddities. The basic intention is that if either "traditional cricketing principles" applied to the trajectory or the projection itself say not out, it's not out. Of course, it's less transparent than simply using the calculated uncertainties of the systems, or even deciding on mathematical factors to simulate the traditional approach. Bell, Watson (v Zimbabwe) and Paine (v England in Perth) have all been given not out in a >2.5m situation where the computer said the ball was hitting the stumps. One decision was upheld by the umpire, two were overturned. We can't know whether this is because different factors were involved in the original decisions, the umpires have different ideas of what "traditional cricketing principles" are, or simply are differently inclined to overturn there own decisions. The first possible reason is in line with the aims of the clause, the second arguably so and the third clearly an unwelcome human factor. Describing the process more explicitly might help (if such a thing can be agreed on), but applying it to "out" decisions as well misses the point.

Personally, assuming the demonstration of the tracking and projection accuracy was more satisfying than indicated by the Hawkeye document, I would like to put more emphasis on the predictions of the tracking systems, but not on the grounds of transparency or lack of human involvement. Actually, even without using the tracking systems in real time, I'm happy to see umpires let them inform future decisions to some extent, as has already reportedly happened. I would rather make LBW decisions as literally as possible than maintain traditional levels of doubt in the process, but that's not because of a technological argument, but because I think I'd like the change it brings.

I'm certainly not pretending it wouldn't be a change. Which leaves us with the current clause in the DRS, intended to avoid a drastic change to interpretation of the LBW law. It might not be perfect in that regard, given that it depends on how the umpires use it, but criticism needs to either have the same intention, or tell us why it doesn't matter.