Saturday, July 24, 2010

Reporting on Hawkeye - II

I have had another email exchange with Paul Hawkins of Hawk-Eye Innovations since my last post about Hawkeye, and finally, things seem to be moving in a direction which makes sense. Mr. Hawkins pointed out, the margin of error in Hawkeye (he cites the specific example of the Tendulkar LBW, which, without the benefit of Hawkeye, was Out in my opinion, would be 5mm) is more binary than one might imagine.

Mr. Hawkins wrote that Hawk-Eye Innovations had suggested that an "uncertainty-ellipse" be shown around the path of the ball as it appears on TV, but that this has been opposed by broadcasters. I quote the relevant part of his email in full:

We have proposed on a number of occasions the idea of presenting an “uncertainty ellipse” around the ball, but all broadcasters have rejected the idea for a number of reasons:
1) Almost all commentators are ex cricketers, and generally not that scientifically bright. They would have no ability to explain what an uncertainty ellipse is
2) Broadcasters prefer a “definitive” where was the ball going – some commentators are then good at interpreting that information saying something like: “Hawk-Eye shows it just clipping leg stump, so a good decision to give the benefit to the batsman”
3) If the uncertainty ellipse is supposed to reflect Hawk-Eye error, then the ellipse would be so small around the ball that you would not see it around the ball
The ICC, in it's response through Colin Gibson is maintaining a hands off attitude to the matter, in much the same way that they seem to operate when it comes to the actual details of getting the DRS technology to work. "What is reflected by the broadcaster is provided by Hawkeye or Virtual Eye (another company some broadcasters use)". This, as Paul Hawkins' email reveals, is not strictly true.

The upshot of it is, that the Decision Review System, and the use of technology to aid Umpiring decisions, is being worked out between the company or companies that make the technology, and the company the broadcasts the game. The agency that  is actually supposed to run cricket is only marginally in the picture.

The point about the uncertainty ellipse is exactly as I have long suspected it to be. It has been set aside in the interests of persuasiveness, a polite word for marketability. So the way Hawkeye and all Technology works in Cricket - all judgment of what is good technology and what is bad technology, falls in the hands of the single most unaccountable, untrained group in cricket - Commentators.

Do you still believe in the wisdom of the UDRS?


  1. What does the incompetence of commentators have to do with one's belief in UDRS? The informed people know there is a margin of error involved with the predictive portion of Hawkeye, which is why certain LBW referrals go back to the on-field umpire who upholds his original decision.

    You mentioned the Tendulkar LBW dismissal and that the margin or error for it would be 5 mm. It's less than the radius of a cricket ball, which is the margin of error used for UDRS. So if the referral system had been in place the original decision would have been upheld anyway.

  2. Do you still believe in the wisdom of the UDRS?
    What Mahek said. Yes. If the ICC forced Hawekeye and all broadcasters to show the uncertainty ellipse, then all LBW decisions made with the UDRS would be the same as they are now.

  3. Players being able to dispute decisions?

  4. What do you think the cumulative effect of seeing graphic after graphic showing some scientific uncertainty, rather than the pretty black magic certainty we see right now will be?

    Why do you think people trust Hawkeye so much? Because they know the mathematics of it inside out? Or because it's been drilled into them that Hawkeye can perfectly trace the path of a delivery from the bowlers hand to the point where it hits the batsman and beyond, due to a beautiful animated graphic that is basically a lie?

    This is not about what you or I think is correct, this is about how truth is socially produced.

  5. This is not about what you or I think is correct, this is about how truth is socially produced.
    I thought it was about improving LBW decision-making with the UDRS....

  6. No.... it was simply about representing hawkeye properly, which is currently not happening.

    In my view, Hawkeye is not necessary at all, neither is the whole spectacle of players asking for referrals.

  7. as far as "improving LBW" goes that is.

  8. Why did you ask if we "still believe in the wisdom of the UDRS" then?

    I agree that the uncertainties should be shown. I don't think that their absence is doing much damage to anything.

  9. No? If you think their absence is trivial, then why do you think they should be shown?

    Also, isn't it obvious by now, that the whole spectacle of getting teams to ask for reviews is a broadcasters dream? Isn't is also clear that the ICC is not in the drivers seat as was once assumed - given that they don't own the technology and don't make the contracts with the people who use the technology (broadcasters), and don't have the means to enforce anything.

    Who then is in a position to affect things? Home boards and Broadcasters? Of course, this is all passed off as being in the interests of "Cricket".

    Thats why I asked the question - how much power does cricket want to hand over to broadcasters?

  10. If you think their absence is trivial, then why do you think they should be shown?
    I think they should be shown because that would be a more accurate representation of the projected path of the ball. Nevertheless, with errors almost always less than a ball radius, and often a lot less, I don't let it bother me.

    (On a vaguely related point, I'm also not bothered by averages being quoted to two decimal places, even though that's a ridiculous level of precision for a number that roughly represents the 'true talent' of a batsman or bowler.)

    The ICC isn't completely powerless. I don't know what their financials are like, but presumably if they really wanted to, they could budget some money for Hotspot cameras, rather than relying on host broadcasters to supply them. If they wanted to, the ICC could also discard the UDRS entirely. The Hawkeye representation is a minor issue; on the actual treatment of errors, the ICC has looked at the numbers and decided that if Hawkeye says that the ball hits the stumps by less than a ball radius, then it's an on-field call.

    So yes, there are various interests at play, but in the end we've muddled our way to a system that improves decision making.

    On another aspect of broadcaster influence, the broadcasters are obviously a big part of the push for day-night Tests (which I am opposed to). But home boards are obviously going to push for this as well, because then they'll get more money for the TV rights. And if you get all the full member boards supporting day-night Tests, then what is the ICC going to do? It basically is the full member boards.

    I don't see how you can take much power away from the broadcasters.

  11. umm... haven't you answered your question yourself david? don't support udrs.

    what has hawkeye improved about lbw decisions?

  12. I don't see why something that is in the interest of broadcasters is necessarily bad for the game. Sure sometimes there is conflict but I don't think that's the case here.

    The reason they don't show the uncertainty ellipse is quite intuitive. There is a margin of error in pretty much everything in real life. Hell you have population figures for towns, cities and countries. Do you think they're 100% accurate? The ICC has already made the margin of error for a decision to be reversed a lot bigger than the margin of error for Hawkeye, if people fail to understand that after watching countless Hawkeye pictures it's their problem. The ICC has never claimed Hawkeye has no margin of error, which is why the decision isn't reversed everytime Hawkeye shows something contrary to what the umpire has decided.

    I don't have a problem with players referring decisions. If it were upto the umpires they would either send every decision upstairs or be bullheaded and not go upstairs at all. Atleast in the current system you get to see how often the umpires get it right, and it also shows up certain players for just being stupid (Remember Harbhajan's referrals from the Lanka series in 2008). The umpires have already shown how they can sabotage this system when it was tried in county cricket. The third umpire always upheld the original decision because the umpires had decided they wouldn't make each other look bad.

  13. Forgot to mention this.

    The ICC doesn't have control over it because it has decided that the host board has to pay for the technology if the UDRS has to be in effect. If UDRS is not in effect for a series, it is upto the broadcaster to foot the bill for the tools used in UDRS. In this case, it is just something the broadcaster feels it should invest in to improve the viewing experience in much the same way as broadcasters in the past have used cameras at both ends, stump cam, stump mics, etc.

    The official line from the ICC is that since its funds are meant for global cricket, using them on Hawkeye, Hotspot, etc. would take some money out of what is spent in countries which don't play test cricket and hence don't use UDRS. I think there is a way around it if there is enough political will. Have the ICC, host board and the broadcaster spend a certain percentage each. Also, use it to sell branding space so that the cost can be recouped.

  14. umm... haven't you answered your question yourself david? don't support udrs.

    what has hawkeye improved about lbw decisions?

    Where in what any of I wrote was something that suggests I should be against the UDRS?

    Hawkeye gets LBW decisions correct more often than the on-field umpires, as shown by the fact that some referrals get overturned. Some of these referrals would have been overturned by a third umpire shown two conventional replays (front-on view with the purple line showing the line of the stumps; a side-on view showing height). But even if you think that the latter method would be just as accurate as Hawkeye, I would still prefer the computer because, being an automatic system, it won't get accused by enraged fans of being biased against a particular batsman/team.

    (One pseudo-example: the Tendulkar shoulder-before-wicket dismissal. Some people looked at the side-on replay and thought that it would have clipped the bails. Others thought it would have gone over the top. I'd rather have Hawkeye make that decision on referral than, say, Steve Bucknor watching the TV screen.)

  15. If the uncertainty ellipse is supposed to reflect Hawk-Eye error, then the ellipse would be so small around the ball that you would not see it around the ball

    This is the crux.

    A visual solution could be to make the ball an uncertainity ellipse with a hollow core.

    I however tend to agree with David Barry (on an empirical basis without any facts ot figures to back me up) that as of now, Hawk Eye appears to be more accurate than umpires.

    Kudos for the initiative Kartikeya.

    Illuminates much the dark world of tracking and predictive science for us. I hope someday finely-tuned technology does come in.

  16. @David - you didn't, i'm saying you should :) Become a technophobe my friend... or we'll have day-night Tests in colored clothing with 5 innings of 20 overs each before you know it, played over five evenings! :)

    @Mahek - Because for broadcasters, cricket is just one more television software. How much ground should the game concede to TV? TV needs Cricket more than Cricket needs TV, simply because there's only one Cricket and many TV Channels. Thats why broadcasters should not be dictating to TV what the rules should be.

  17. I think they've struck a balance in this case. The common man really doesn't get concepts like the uncertainty ellipse. Why make the experience that much more difficult to comprehend for him? As of now, he's told that Hawkeye will be used for LBW decisions but certain decisions will go back to the on-field umpire because there is a margin for error. That's a subtle way of saying there is an uncertainty ellipse around the ball. Now if people still don't get it you can't do much else.

  18. But then there is still a claim of accuracy in the whole thing that is unfounded. As is the claim of certainty. It's like selling a medicine without reporting a side effect, leading people to believe that it doesn't exist.

    The company that makes the program would like to put it there.

    Besides, how hard is it to understand that the ball could hit the stumps anywhere within this ellipse? What are commentators for?

  19. Gee Kartikeya, I thought you knew they were there to call LKM a visionary and to extol every Citi Moment of Success. I mean do you really think they're there to give us an insight into the game? :)

  20. I think I've said before that I'd like to see uncertainty indicated, but I'm not sure what the best way to do this is. To some extent I agree with Mahek in that if any watcher doesn't realise there is some uncertainty in the Hawkeye tracking/prediction then they have problems in much broader contexts - the issues are hardly unique! I'd like to see something out of interest and a desire to encourage people to think like this more broadly - not to talk down Hawkeye. If anything, if we used Hawkeye's figures on this, we'd get more power to Hawkeye, not less.

    Apart from that, an ellipse does seem to be the most obvious answer. However, it would seem to simply make the ball bigger. I'd guess that a lot of viewers would instinctively pick up the fact that the ball might have hit something close to the predicted path, but not so much the case when it is shown to hit and might have missed. A hollow core might be explicable to some. but you still need to rely on some common sense.

    Apart from that, there is the third of the reasons given by Hawkins. The fact that a case has been overstated by other interested parties (the broadcasters) doesn't mean that we shouldn't give a fair hearing to those making a more basic case.

  21. @Mahek: For about 95% of the time that i have been following cricket, they weren't doing that.

  22. Sure, but if that 5% has been the most recent then it's a sign of where things are going.

    I had the pleasure of watching Sky's coverage of cricket series when I was in America. Sure some of the discussions were rubbish, but when it was much better than what people back home are subjected to. In fact for series in which the local channel just borrows the Sky feed it fills up the breaks with some highlights section or a non-cricket programme. I'd rather watch a discussion comprising Michael Atherton, Michael Holding and Nasser Hussain. They're surprisingly good when not talking about England, except for Holding who's good almost all the time.

  23. Yes, but I'm interested in how think should be.

  24. LOL @ "If the uncertainty ellipse is supposed to reflect Hawk-Eye error, then the ellipse would be so small around the ball that you would not see it around the ball" -- thanks for the entertainment, Mr Hawkins =)