DRS: Independent testing

[cricketrulez]

For those who care....

"The CEC stated it was "satisfied" with the improvements made over the
two key components of DRS technology. These include the new Hot Spot
cameras as well as the independent ball-tracking research conducted by
Dr Ed Rosten, a Cambridge University department of engineering expert
in computer-vision technology. Rosten, the ICC said, had "tested the
accuracy and reliability of ball-tracking in a recent Test series and
concluded that the results were 100% in agreement with the outcomes
produced from his assessments."

I believe this is the Ed Rosten they are referring to
http://www.edwardrosten.com/index.html

So i wrote to Dr. Ed Rosten. I'll post any reply I get back.

Dear Dr. Rosten,

I ran across your name in regards to the accuracy of the Hawkeye
system used in Cricket to predict the path of the ball.

http://www.espncricinfo.com/ci-icc/content/story/569949.html

"The CEC stated it was "satisfied" with the improvements made over the
two key components of DRS technology. These include the new Hot Spot
cameras as well as the independent ball-tracking research conducted by
Dr Ed Rosten, a Cambridge University department of engineering expert
in computer-vision technology. Rosten, the ICC said, had "tested the
accuracy and reliability of ball-tracking in a recent Test series and
concluded that the results were 100% in agreement with the outcomes
produced from his assessments."

1) Will any of the testing information be available to the public?

2) How were the tests conducted? In a cricket ground? using a bowling
machine? If players were used to bowl instead of a bowling machine,
were both spinners and fast bowlers used?

3) How exactly did you verify the accuracy of Hawkeye? did you feed
the system partial footage of deliveries that made it to the keeper?

Given that this system is being used for public consumption and your
research is being used to endorse its use, I'm hoping you will be able
to shed some light on this matter.

[alvey]

Why don't you expand your research and write to the BCCI and ask them why
they're such a mob of arseholes?



alvey

[Mike Holmans]

On Wed, 27 Jun 2012 06:18:17 +1000, alvey <al...@atattat.com> tapped
the keyboard and brought forth:

>On Tue, 26 Jun 2012 11:23:47 -0700 (PDT), cricketrulez wrote:
>
>> For those who care....
>>
>> "The CEC stated it was "satisfied" with the improvements made over the
>> two key components of DRS technology. These include the new Hot Spot
>> cameras as well as the independent ball-tracking research conducted by
>> Dr Ed Rosten, a Cambridge University department of engineering expert
>> in computer-vision technology. Rosten, the ICC said, had "tested the
>> accuracy and reliability of ball-tracking in a recent Test series and
>> concluded that the results were 100% in agreement with the outcomes
>> produced from his assessments."
>>
>> I believe this is the Ed Rosten they are referring to
>> http://www.edwardrosten.com/index.html
>>
>> So i wrote to Dr. Ed Rosten. I'll post any reply I get back.
>>
>> Dear Dr. Rosten,
>>

...

>>
>> Given that this system is being used for public consumption and your
>> research is being used to endorse its use, I'm hoping you will be able
>> to shed some light on this matter.
>
>Why don't you expand your research and write to the BCCI and ask them why
>they're such a mob of arseholes?

Credit where credit's due, though. Tony Greig, for whom I usually have
very little time, used his platform giving the MCC Spirit of Cricket
lecture to say precisely that the BCCI are a mob of arseholes, and in
particular that their stance on DRS is a disgrace. This may well be
the first time I have agreed with Greig in 25 years, if not longer.

Cheers,

Mike
--

[mike]

On Jun 26, 7:23 pm, cricketrulez <cricketrulez1...@gmail.com> wrote:
> For those who care....
>
> "The CEC stated it was "satisfied" with the improvements made over the
> two key components of DRS technology. These include the new Hot Spot
> cameras as well as the independent ball-tracking research conducted by
> Dr Ed Rosten, a Cambridge University department of engineering expert
> in computer-vision technology. Rosten, the ICC said, had "tested the
> accuracy and reliability of ball-tracking in a recent Test series and
> concluded that the results were 100% in agreement with the outcomes
> produced from his assessments."
>

> I believe this is the Ed Rosten they are referring tohttp://www.edwardrosten.com/index.html

>
> So i wrote to Dr. Ed Rosten. I'll post any reply I get back.
>
> Dear Dr. Rosten,
>
> I ran across your name in regards to the accuracy of the Hawkeye
> system used in Cricket to predict the path of the ball.
>
> http://www.espncricinfo.com/ci-icc/content/story/569949.html
>
> "The CEC stated it was "satisfied" with the improvements made over the
> two key components of DRS technology. These include the new Hot Spot
> cameras as well as the independent ball-tracking research conducted by
> Dr Ed Rosten, a Cambridge University department of engineering expert
> in computer-vision technology. Rosten, the ICC said, had "tested the
> accuracy and reliability of ball-tracking in a recent Test series and
> concluded that the results were 100% in agreement with the outcomes
> produced from his assessments."
>

I've nothing against DRS use, but isnt 100% a pretty big claim?
Anyone else find this a bit incredible?

> 2) How were the tests conducted? In a cricket ground? using a  bowling
> machine? If players were used to bowl instead of a bowling machine,
> were both spinners and fast bowlers used?
>

i too would love to know the answer to this

> 3) How exactly did you verify the accuracy of Hawkeye?

looking forward to this one too.

> Given that this system is being used for public consumption and your
> research is being used to endorse its use, I'm hoping you will be able
> to shed some light on this matter.

well done

mike

[Brian Lawrence]

On 26/06/2012 19:23, cricketrulez wrote:

> 2) How were the tests conducted? In a cricket ground? using a bowling
> machine? If players were used to bowl instead of a bowling machine,
> were both spinners and fast bowlers used?

A little searching reveals that he examined 14 sequences from the recent
South Africa vs. Australia series. His company's results matched those
of the system used in that series.

> 3) How exactly did you verify the accuracy of Hawkeye? did you feed
> the system partial footage of deliveries that made it to the keeper?
>
> Given that this system is being used for public consumption and your
> research is being used to endorse its use, I'm hoping you will be able
> to shed some light on this matter.
>

--

Brian W Lawrence
Wantage
Oxfordshire

[Unknown]

No $ky hooks attached,and Greig makes sense. That's not really fair to
say, but clearly Greig will play ball for the right pay-man.

max.it

[cricketrulez]

Brian,

thanks for info. Do you happen to have a link by any chance? I sure
hope that the testing involved more than 14 deliveries!

[jzfredricks]

On Wednesday, June 27, 2012 4:58:48 AM UTC+10, mike wrote:
> > 2) How were the tests conducted? In a cricket ground? using a  bowling
> > machine? If players were used to bowl instead of a bowling machine,
> > were both spinners and fast bowlers used?
> >
>
> i too would love to know the answer to this
>
> > 3) How exactly did you verify the accuracy of Hawkeye?
>
> looking forward to this one too.

Don't hold your breath. There's resistance to answering these questions.

[jzfredricks]

Frankly, it's all bullshit.

How on earth did he test "the accuracy and reliability of ball-tracking in a recent Test series and concluded that the results were 100% in agreement with the outcomes produced from his assessments."

I assume he didn't have his own version of xEye at the ground, and if he did, was it independently and thoroughly tested? No it wasn't.

If he didn't have his own xEye-like system, then all he really had to go off was TV replays. Surely he didn't use any xEye results to test xEye itself.

Are they trying to tell us that HD TV replays are enough to test xEye? If they, bullshit, bullshit, bullshit.

Furthermore, if the results were 100% ok, why on earth are they doing this;

"The ICC cricket committee's suggestion pertaining to the lbw regulation under DRS has also been approved. Under the LBW rule, the 'margin of uncertainty' regarding the point of impact with the batsman will be the same as that provided for the point of impact with the stumps."

xEye is so inaccurate that the margins of error that are being forced into the regulations are SO broad that it's getting to the point where it's almost pointless using the technology.

Get rid of it and use the super-imposed stump-to-stump line and TV replays.

[CaraMia]

On 27-06-2012 02:01 AM, Mike Holmans wrote:
> Credit where credit's due, though. Tony Greig, for whom I usually have
> very little time, used his platform giving the MCC Spirit of Cricket
> lecture to say precisely that the BCCI are a mob of arseholes, and in
> particular that their stance on DRS is a disgrace. This may well be
> the first time I have agreed with Greig in 25 years, if not longer.

About the speech, it was pretty hypocritical of Greig who has been
preoccupied with money his whole life to be criticize someone for being
preoccupied with money.

About DRS, my view is that make DRS mandatory, but not the tools - i.e.
it should be OK to just do DRS with just regular slow mo replays. Any
other tools should be agreed upon by both the teams playing.

[cricketrulez]

On Jun 26, 2:41 pm, Brian Lawrence <Brian_W_Lawre...@msn.com> wrote:

> A little searching reveals that he examined 14 sequences from the recent
> South Africa vs. Australia series. His company's results matched those
> of the system used in that series.

His company? meaning this wasn't done at Cambridge university?

[Brian Lawrence]

http://icc-cricket.yahoo.net/newsdetails.php?newsId=20417_1338564600

Your other question about where Rosten did his analysis - the same
report states that he is a FORMER Cambridge lecturer, so probably not
done AT the Uni.

It doesn't look - from that page - as if he examined more than 14
deliveries.

[cricketrulez]

Thanks Brian.

[Brian Lawrence]

On 27/06/2012 08:11, cricketrulez wrote:

>>> thanks for info. Do you happen to have a link by any chance? I sure
>>> hope that the testing involved more than 14 deliveries!
>>
>> http://icc-cricket.yahoo.net/newsdetails.php?newsId=20417_1338564600

>> It doesn't look - from that page - as if he examined more than 14
>> deliveries.

> Thanks Brian.

The SAf vs. Aus series was in November 2011, there were two Tests at
Cape Town & Johannesburg. I just did a quick scan of the Cricinfo
commentary, looking for "review". In the two Tests I found 13 deliveries
where a review was requested. The comms might not use the word review in
all cases, so it looks as though he examined the actual reviews from the
two matches.

[CaraMia]

On 27-06-2012 12:28 PM, Brian Lawrence wrote:
> It doesn't look - from that page - as if he examined more than 14
> deliveries.

Who selected those 14 deliveries?

[Brian Lawrence]

Cricket Archive actually log the reviews at the bottom of the scorecard
page. There were 24 in the two Tests:

http://www.cricketarchive.co.uk/Archive/Scorecards/341/341649.html
http://www.cricketarchive.co.uk/Archive/Scorecards/341/341650.html

I wondered if there were 14 where the umpire's decision was reversed,
but it looks like that only happened six times (Test 2 doesn't include
whether the review was upheld or struck down, but the Cricinfo comms
allows us to check, and the only T2 review that was upheld was by
Jacques Kallis in the 23rd over. He was given out by umpire Bowden,
reviewed and was given not out.)

The ones I didn't find in Cricinfo used the word 'referred' or
'referral' rather than review.

[decorr]

Apart from understanding "how" and "who" selected these 14 deliveries, shouldn't the precision/recall and the F1 score of the system to have a more balanced view?

[Brian Lawrence]

On 26/06/2012 19:58, mike wrote:

>> in computer-vision technology. Rosten, the ICC said, had "tested the
>> accuracy and reliability of ball-tracking in a recent Test series and
>> concluded that the results were 100% in agreement with the outcomes
>> produced from his assessments."
>>
>
> I've nothing against DRS use, but isnt 100% a pretty big claim?
> Anyone else find this a bit incredible?

From my elementary research, it's a perfectly reasonable claim. He
examined 14 deliveries where the DRS was invoked to confirm or deny
the umpire's decision. In his opinion, the decisions reached by the DRS
in those 14 cases were all correct. He's an 'expert' in the type of
technology in use, so his opinion is valuable. He's also claimed to be
independent - he's not associated with the ICC or Hawkeye, Hotspot, etc.

He wasn't asked to 'test' the technology for accuracy per se - in his
opinion it arrived at the right decision in all 14 instances. The 100%
accuracy should not be applied to any of the technology involved, he
was looking at the whole system (which includes the 'umpires call').

As Sir Viv keeps saying - in my opinion.

[jzfredricks]

On Wednesday, June 27, 2012 7:16:32 PM UTC+10, Brian Lawrence wrote:
> From my elementary research, it's a perfectly reasonable claim.

Yeah. Those 14 reviews might have been very easy for DRS (ie all pitching 20cm down leg) or very difficult. We don't know.

Surely we need the ACCURACY of xEye tested, not a set of referrals from one Test.

Whilst I consider this testing to be independent, I don't consider it to be thorough.

[Brian Lawrence]

On 27/06/2012 12:08, jzfredricks wrote:
> On Wednesday, June 27, 2012 7:16:32 PM UTC+10, Brian Lawrence wrote:
>> From my elementary research, it's a perfectly reasonable claim.
>
> Yeah. Those 14 reviews might have been very easy for DRS (ie all pitching 20cm down leg) or very difficult. We don't know.

Except that I expect they were close enough for the
fielders/bowler/batsman/umpires to disagree. We certainly don't know.

> Surely we need the ACCURACY of xEye tested, not a set of referrals from one Test.

I'm not sure I agree with that. I reckon that any high-tech system in
the 21st century is bound to be more accurate than a human umpire
deciding in real time. Given that assumption, which is pretty much
accepted throughout cricket, why not just use it. If it's only 80%
accurate so what? It's only going to be used in less than 1% of
deliveries in a match.

> Whilst I consider this testing to be independent, I don't consider it to be thorough.

Agreed, though it wasn't intended to be.

[jzfredricks]

On Wednesday, June 27, 2012 10:05:08 PM UTC+10, Brian Lawrence wrote:
> > Surely we need the ACCURACY of xEye tested, not a set of referrals from one Test.
>
> I'm not sure I agree with that. I reckon that any high-tech system in
> the 21st century is bound to be more accurate than a human umpire
> deciding in real time. Given that assumption, which is pretty much
> accepted throughout cricket, why not just use it. If it's only 80%
> accurate so what? It's only going to be used in less than 1% of
> deliveries in a match.

If we know how accurate it really is, we can write the regulations to suit. The first iteration of the DRS regulations let xEye overrule the onfield umpire more easily. Over time those regulations have changed, and given xEye less and less authority. This has happened because it keeps on returning results that LOOK dodgy.

I'm peeved it's taken this long, and we're STILL not there. Why couldn't they get it (the testing) done right in the first place?

With proper testing the skeptics, like myself, could simply be told to stfu.

[mcenley]

> I reckon that any high-tech system in
> the 21st century is bound to be more accurate than a human umpire
> deciding in real time. Given that assumption, which is pretty much
> accepted throughout cricket, why not just use it.

I would like to take up on this point. I am curious as to what makes you "generalize" that assumption. An system S1 may be more "accurate" than a human while a system S2 may not be more "accurate". It is perfectly plausible to have the aforementioned. I can countless examples across both categories in our daily life systems.

[Brian Lawrence]

On 27/06/2012 13:29, mcenley wrote:

>> I reckon that any high-tech system in
>> the 21st century is bound to be more accurate than a human umpire
>> deciding in real time. Given that assumption, which is pretty much
>> accepted throughout cricket, why not just use it.
>
> I would like to take up on this point. I am curious as to what makes you "generalize" that assumption. An system S1 may be more "accurate" than a human while a system S2 may not be more "accurate". It is perfectly plausible to have the aforementioned. I can countless examples across both categories in our daily life systems.

Point taken, but the DRS systems that have been tried so far are
certainly in the 'accurate' category. If there were countless different
systems produced by computer geeks or child prodigies, or whatever, many
would turn out to be inaccurate. The systems now in use have been
available for some considerable time, with a lot of success and have
been continually updated based on experience in real matches.

The only real issue with DRS is the accuracy of it's predictive
function, in particularly regarding lbw decisions, but again, using
computer analysis must give a more accurate prediction of ball
trajectory than the umpire's brain.

[Brian Lawrence]

On 27/06/2012 13:13, jzfredricks wrote:
> On Wednesday, June 27, 2012 10:05:08 PM UTC+10, Brian Lawrence wrote:
>>> Surely we need the ACCURACY of xEye tested, not a set of referrals from one Test.
>>
>> I'm not sure I agree with that. I reckon that any high-tech system in
>> the 21st century is bound to be more accurate than a human umpire
>> deciding in real time. Given that assumption, which is pretty much
>> accepted throughout cricket, why not just use it. If it's only 80%
>> accurate so what? It's only going to be used in less than 1% of
>> deliveries in a match.
>
> If we know how accurate it really is, we can write the regulations to suit. The first iteration of the DRS regulations let xEye overrule the onfield umpire more easily. Over time those regulations have changed, and given xEye less and less authority. This has happened because it keeps on returning results that LOOK dodgy.

Unless I've misunderstood DRS, xEye doesn't have ANY authority. If a
decision gets referred it's the TV umpire who looks at what the various
replays, XEye, Hotspot, etc. indicate happened. IF the TV umpire sees
enough evidence to suggest that the on-field umpire made a mistake, he
advises his colleague of that and the decision is reversed. It's NOT
DRS that decides it's the umpires.

> I'm peeved it's taken this long, and we're STILL not there. Why couldn't they get it (the testing) done right in the first place?
>
> With proper testing the skeptics, like myself, could simply be told to stfu.

What makes you think that proper testing wasn't done when the systems
were developed? It would seem logical that you would test the software
by feeding it video of deliveries taken without a batsman in the way and
freezing the image at various points. The companies involved are
competing in a narrow field and won't make their data available publicly
in case competitors learn their secrets, but I'm sure they
have the data available to impress the ICC (and others).

[jzfredricks]

On Wednesday, June 27, 2012 11:46:47 PM UTC+10, Brian Lawrence wrote:
> Unless I've misunderstood DRS, xEye doesn't have ANY authority. If a
> decision gets referred it's the TV umpire who looks at what the various
> replays, XEye, Hotspot, etc. indicate happened. IF the TV umpire sees
> enough evidence to suggest that the on-field umpire made a mistake, he
> advises his colleague of that and the decision is reversed. It's NOT
> DRS that decides it's the umpires.

Yes, you've misunderstood (part of) the DRS.
xEye can provide evidence on 3 things - landing spot, impact spot, and position at stumps.
No other technology in use can provide evidence on those 3, therefore xEye's evidence on these 3 matters is authoritative and cannot be overruled. Obviously taking into account the "umpire's zone/margin of error".
Other technologies (TV replay, hotspot, snicko) can provide different types of evidence which will affect an LBW decision. All 3 of those, though, require a human umpire to analyse the evidence and make a decision.

> > I'm peeved it's taken this long, and we're STILL not there. Why couldn't they get it (the testing) done right in the first place?
> > With proper testing the skeptics, like myself, could simply be told to stfu.
>
> What makes you think that proper testing wasn't done when the systems
> were developed? It would seem logical that you would test the software
> by feeding it video of deliveries taken without a batsman in the way and
> freezing the image at various points. The companies involved are
> competing in a narrow field and won't make their data available publicly
> in case competitors learn their secrets, but I'm sure they
> have the data available to impress the ICC (and others).

Proper testing MAY have been done*. They might know perfectly well how accurate it is.

What I want tested is their PUBLISHED accuracy/margins of error. If they say the margin of error is 5mm, and we test it as such, then we can write proper DRS regulations.

We currently have a 35mm margin of error at some points where their documentation says 5mm. Why is that?

On the other hand, maths shows that two 5mm errors can result in a 100mm error at the stumps, but the regulations only assume up to 35mm error at the stumps. Why is that?

If TESTING proves it to be 5mm, then the regulations should reflect that.

*The MCC testing mentioned in HawkEye material certainly WASN'T proper testing.

[Mike Holmans]

On Wed, 27 Jun 2012 14:29:14 +0100, Brian Lawrence
<Brian_W_...@msn.com> tapped the keyboard and brought forth:

I don't think it "must" give a "more" accurate prediction than an
umpire's brain, although I think it extremely likely. If there is
substantial turn or unusual bounce and there is very little distance
between bounce and pad, I suspect that the video technology has a
significant chance of having a large enough margin of error for an
umpire to be at least as accurate.

That's not to say I don't agree with you whole-heartedly in general.
The jzfs of this world with their endless insistence that whatever
teasting is done, they still don't believe it's good enough, have long
since become as obsessive as the people who still believe that Obama
was born in Kenya.

Cheers,

Mike
--

[Brian Lawrence]

On 27/06/2012 15:18, jzfredricks wrote:
> On Wednesday, June 27, 2012 11:46:47 PM UTC+10, Brian Lawrence wrote:
>> Unless I've misunderstood DRS, xEye doesn't have ANY authority. If a
>> decision gets referred it's the TV umpire who looks at what the various
>> replays, XEye, Hotspot, etc. indicate happened. IF the TV umpire sees
>> enough evidence to suggest that the on-field umpire made a mistake, he
>> advises his colleague of that and the decision is reversed. It's NOT
>> DRS that decides it's the umpires.
>
> Yes, you've misunderstood (part of) the DRS.
> xEye can provide evidence on 3 things - landing spot, impact spot, and position at stumps.
> No other technology in use can provide evidence on those 3, therefore xEye's evidence on these 3 matters is authoritative and cannot be overruled. Obviously taking into account the "umpire's zone/margin of error".
> Other technologies (TV replay, hotspot, snicko) can provide different types of evidence which will affect an LBW decision. All 3 of those, though, require a human umpire to analyse the evidence and make a decision.

Surely TV replay provides evidence of the first two? OK, there's no
point in using
it because xEye does that anyway, but it's there and would be more
reliable than the umpire - you can review the landing & impact as often
as you want in replay, while the poor old ump gets to do it real time
with no slo-mo.

>>> I'm peeved it's taken this long, and we're STILL not there. Why couldn't they get it (the testing) done right in the first place?
>>> With proper testing the skeptics, like myself, could simply be told to stfu.
>>
>> What makes you think that proper testing wasn't done when the systems
>> were developed? It would seem logical that you would test the software
>> by feeding it video of deliveries taken without a batsman in the way and
>> freezing the image at various points. The companies involved are
>> competing in a narrow field and won't make their data available publicly
>> in case competitors learn their secrets, but I'm sure they
>> have the data available to impress the ICC (and others).
>
> Proper testing MAY have been done*. They might know perfectly well how accurate it is.
>
> What I want tested is their PUBLISHED accuracy/margins of error. If they say the margin of error is 5mm, and we test it as such, then we can write proper DRS regulations.
>
> We currently have a 35mm margin of error at some points where their documentation says 5mm. Why is that?

I know you have discussed this ad nauseum previously, but I haven't
taken much notice, so I don't know where 35mm comes from.

> On the other hand, maths shows that two 5mm errors can result in a 100mm error at the stumps, but the regulations only assume up to 35mm error at the stumps. Why is that?

I assume this means that the landing point might be 5mm in error and so
might the impact point, but again I'm unaware how that leads to 100m (or
35mm for that matter). You don't need to explain it again, I'm not
particularly bothered.

> If TESTING proves it to be 5mm, then the regulations should reflect that.

Surely the regulations don't need to take into account the actual margin
of error. In terms of pitching on the wicket (or outside off),
if the equipment measures it to pitch outside leg by more than the
margin of error then it's definitely not out. If it's measured as more
than the width of the ball minus the margin inside the leg stump line
then it's pitched on the wicket. With that regulation it matters not
whether the margin for error is 1mm, 5mm, 10mm or anything up to half
the diameter of the ball, you just apply whatever the margin is. You
don't need to write a regulation for 5mm and then rewrite it for 4mm or
6mm. You need to rule on whether the ball is inline or not when it's
within the margin of error, you either rule it on the wicket or outside,
but again the size of the margin can be flexible.

Without thinking it through any further I'd imagine the same can be
regulated for the other two points, albeit with different margins.

> *The MCC testing mentioned in HawkEye material certainly WASN'T proper testing.

Really though, does it matter? The bowler appeals for lbw, the umpire
decides it's not out. The fielding side decide to review it. Whatever
system is in use isn't going to be 100% accurate and the margin for
error might not be known with any certainty, but regardless of that,
it's more accurate than the umpire.

The umpires get it wrong sometimes, they are human after all, but the
technology is demonstrably less fallible, even if not foolproof. Instead
of bickering and whinging about it it ought to be used now, and any
problems arising dealt with as necessary.

[Brian Lawrence]

On 27/06/2012 16:34, Brian Lawrence wrote:

> Surely the regulations don't need to take into account the actual margin
> of error. In terms of pitching on the wicket (or outside off),
> if the equipment measures it to pitch outside leg by more than the
> margin of error then it's definitely not out. If it's measured as more
> than the width of the ball minus the margin inside the leg stump line
> then it's pitched on the wicket. With that regulation it matters not
> whether the margin for error is 1mm, 5mm, 10mm or anything up to half
> the diameter of the ball, you just apply whatever the margin is. You
> don't need to write a regulation for 5mm and then rewrite it for 4mm or
> 6mm. You need to rule on whether the ball is inline or not when it's
> within the margin of error, you either rule it on the wicket or outside,
> but again the size of the margin can be flexible.

I see that the ICC guidelines for third umpire state that the centre of
the ball should be between the wickets, so the above would need changing
to reflect that.

> Without thinking it through any further I'd imagine the same can be
> regulated for the other two points, albeit with different margins.

ICC guidelines regarding point of impact and whether the ball would have
hit the stumps also use centre of the ball.

[mike]

14 is quite a small sample to base conclusions on.
If they were paying him to properly test a systems accuracy
in a scientific way, i would have thought that the tests would
have been based on a much larger sample of decisions.

BTW, I'm quite happy to accept that DRS is more efficient
than the human eye. But since the main area of dispute
lies with the trajectory of the ball in LBW appeals, it
would surely have been sensible to concentrate on that.

mike

[Brian Lawrence]

:-)

Doing a bit more research I found the patent application for Hawkeye.


"http://worldwide.espacenet.com/publicationDetails/originalDocument;jsessionid=A1FD156FDAF83E6104D64FDDFE8AFCE5.espacenet_levelx_prod_2?CC=WO&NR=0141884A1&KC=A1&FT=D&ND=&date=20010614&DB=&locale=en_EP"

Also, regarding testing which cropped up earlier in the thread, Paul
Hawkins said that they 'filmed' thousands of balls from a bowling
machine stopping the video about two feet from the stumps and processed
that data. He also said that accuracy depended on the trajectory
distance post-bounce before the ball reached the pads - the longer the
better.

[jzfredricks]

On Thursday, June 28, 2012 1:34:36 AM UTC+10, Brian Lawrence wrote:
> On 27/06/2012 15:18, jzfredricks wrote:
> > On Wednesday, June 27, 2012 11:46:47 PM UTC+10, Brian Lawrence wrote:
> >> Unless I've misunderstood DRS, xEye doesn't have ANY authority. If a
> >> decision gets referred it's the TV umpire who looks at what the various
> >> replays, XEye, Hotspot, etc. indicate happened. IF the TV umpire sees
> >> enough evidence to suggest that the on-field umpire made a mistake, he
> >> advises his colleague of that and the decision is reversed. It's NOT
> >> DRS that decides it's the umpires.
> >
> > Yes, you've misunderstood (part of) the DRS.
> > xEye can provide evidence on 3 things - landing spot, impact spot, and position at stumps.
> > No other technology in use can provide evidence on those 3, therefore xEye's evidence on these 3 matters is authoritative and cannot be overruled. Obviously taking into account the "umpire's zone/margin of error".
> > Other technologies (TV replay, hotspot, snicko) can provide different types of evidence which will affect an LBW decision. All 3 of those, though, require a human umpire to analyse the evidence and make a decision.
>
> Surely TV replay provides evidence of the first two? OK, there's no
> point in using
> it because xEye does that anyway, but it's there and would be more
> reliable than the umpire - you can review the landing & impact as often
> as you want in replay, while the poor old ump gets to do it real time
> with no slo-mo.

Are you saying "Surely it does" or "surely it can"?
If the former - no it doesn't. TV replays aren't used as part of the current DRS system for LBW appeal landing spots. The times when xEye was optional and not available, LBWs were un-referrable.

If the latter - I agree. Dump xEye and use the super-imposed red line plus TV replays.

However, I WANT xEye. I think it has potential. I just think any technology that totally eliminates judgement by a human umpire MUST be thoroughly and independently tested.

> > Proper testing MAY have been done*. They might know perfectly well how accurate it is.
> >
> > What I want tested is their PUBLISHED accuracy/margins of error. If they say the margin of error is 5mm, and we test it as such, then we can write proper DRS regulations.
> >
> > We currently have a 35mm margin of error at some points where their documentation says 5mm. Why is that?
>
> I know you have discussed this ad nauseum previously, but I haven't
> taken much notice, so I don't know where 35mm comes from.

Roughly half a ball's width.

> > On the other hand, maths shows that two 5mm errors can result in a 100mm error at the stumps, but the regulations only assume up to 35mm error at the stumps. Why is that?
>
> I assume this means that the landing point might be 5mm in error and so
> might the impact point, but again I'm unaware how that leads to 100m (or
> 35mm for that matter). You don't need to explain it again, I'm not
> particularly bothered.

Pythagoras worked it out 2500 years ago. It's simple maths. 5mm error "north", then 5mm error "south", with a travel distance of 150mm (between pitching and impact) results in a REAL error of about 10cm in the prediction.

> > If TESTING proves it to be 5mm, then the regulations should reflect that.
> Surely the regulations don't need to take into account the actual margin
> of error. In terms of pitching on the wicket (or outside off),
> if the equipment measures it to pitch outside leg by more than the
> margin of error then it's definitely not out. If it's measured as more
> than the width of the ball minus the margin inside the leg stump line
> then it's pitched on the wicket. With that regulation it matters not
> whether the margin for error is 1mm, 5mm, 10mm or anything up to half
> the diameter of the ball, you just apply whatever the margin is. You
> don't need to write a regulation for 5mm and then rewrite it for 4mm or
> 6mm. You need to rule on whether the ball is inline or not when it's
> within the margin of error, you either rule it on the wicket or outside,
> but again the size of the margin can be flexible.

huh? This makes little sense to me. You say "regulations don't need to know the margin of error", then mention MoE about 30 times.

The current margin of error is, roughly speaking, 35mm. What if the REAL margin of error was 80mm? That means this technology will overrule correct umpiring decisions and give batsmen out when they are not.

If we know the real MoE, then the regulations can be written so that doesn't happen.

> > *The MCC testing mentioned in HawkEye material certainly WASN'T proper testing.
>
> Really though, does it matter? The bowler appeals for lbw, the umpire
> decides it's not out. The fielding side decide to review it. Whatever
> system is in use isn't going to be 100% accurate and the margin for
> error might not be known with any certainty, but regardless of that,
> it's more accurate than the umpire.
>
> The umpires get it wrong sometimes, they are human after all, but the
> technology is demonstrably less fallible, even if not foolproof. Instead
> of bickering and whinging about it it ought to be used now, and any
> problems arising dealt with as necessary.

"technology is demonstrably less fallible" how can you say that?

[Calvin]

On Thu, 28 Jun 2012 01:34:36 +1000, Brian Lawrence
<Brian_W_...@msn.com> wrote:

> Surely the regulations don't need to take into account the actual margin
> of error.

Yes they do, and this is precisely the problem with the current system.

The technology is just a tool. In most cases it is pretty accurate,
certainly more so than a human. But, and it is a big but, there are
identified instances (eg where the distance between pitch and impact with
the pad is short) where it is known to be far less accurate. It is
therefore crucial that the margin of error in any particular review is
acknowledged so that the third umpire can include that margin in his
decision making process.

The final decision must always be made by a human, so that person needs to
be able to make sensible judgments about the accuracy of all the data
presented. To put it another way, using common sense (rather than being
locked into set protocols) is the best decision making process.

> In terms of pitching on the wicket (or outside off),
> if the equipment measures it to pitch outside leg by more than the
> margin of error then it's definitely not out.

Agreed. So the margin of error must be known for this occur.

> Really though, does it matter? The bowler appeals for lbw, the umpire
> decides it's not out. The fielding side decide to review it. Whatever
> system is in use isn't going to be 100% accurate and the margin for
> error might not be known with any certainty, but regardless of that,
> it's more accurate than the umpire.

Not always, and making assumptions that it is, is dangerous.

--
cheers,
calvin

[jzfredricks]

On Thursday, June 28, 2012 3:06:30 AM UTC+10, Brian Lawrence wrote:
> Also, regarding testing which cropped up earlier in the thread, Paul
> Hawkins said that they 'filmed' thousands of balls from a bowling
> machine stopping the video about two feet from the stumps and processed
> that data. He also said that accuracy depended on the trajectory
> distance post-bounce before the ball reached the pads - the longer the
> better.

Interesting. Link please? I'd like to see if there are more clues as to his methodology.

[mcenley]

> The only real issue with DRS is the accuracy of it's predictive
> function, in particularly regarding lbw decisions, but again, using
> computer analysis must give a more accurate prediction of ball
> trajectory than the umpire's brain.

Once again you keep on making this point regarding a "computer" is more accurate at predicting than a human brain. You can only make that statement if you understand both aspects completely - [1] A computer and [2] A human brain. We understand [1] very well but [2] we haven't cracked the code. So, currently at best we are comparing a "fully known" system to a "not so completely known" entity. The best example I can take to prove you, by your logic, is that a computer would find it easy to match human faces. However, one just need to speak to some Computer Vision experts (or particularly Biometric researchers) as to why there's no way that any computer analysis in this world can match the human brain cognition for this task.

Yes - the prediction is the problem and in the current scenario it is best to leave the prediction to the umpire. Also, let me add that it's the prerogative of automated systems to prove that they are better than humans and not vice versa.

[Brian Lawrence]

On 28/06/2012 07:40, mcenley wrote:
>> The only real issue with DRS is the accuracy of it's predictive
>> function, in particularly regarding lbw decisions, but again, using
>> computer analysis must give a more accurate prediction of ball
>> trajectory than the umpire's brain.
>
> Once again you keep on making this point regarding a "computer" is more accurate at predicting than a human brain. You can only make that statement if you understand both aspects completely - [1] A computer and [2] A human brain. We understand [1] very well but [2] we haven't cracked the code. So, currently at best we are comparing a "fully known" system to a "not so completely known" entity. The best example I can take to prove you, by your logic, is that a computer would find it easy to match human faces. However, one just need to speak to some Computer Vision experts (or particularly Biometric researchers) as to why there's no way that any computer analysis in this world can match the human brain cognition for this task.
>
> Yes - the prediction is the problem and in the current scenario it is best to leave the prediction to the umpire. Also, let me add that it's the prerogative of automated systems to prove that they are better than humans and not vice versa.

You are writing in more general terms than I did. I was writing of the
DRS systems in use in comparison to the umpire. The computer-based
system is measuring pitching point - is it on the wicket or outside off
stump, and the impact point with the batsman, is it within the bounds of
the stumps. It then projects the ball's measured trajectory to decide if
it would have hit the stumps. It does this using various algorithms and
mathematic principles which are clearly defined.

The umpire uses the evidence of his eyes, his recollection of what just
happened and his experience. He combines that to make an almost instant
decision. A decision which has often been proven wrong in the past. He
should only give the batsman out if he has no doubt - the batsman is
said to receive the benefit of the doubt. It's estimated that before
DRS at least 50% of lbw decisions were wrong. DRS can, and has, correct
some of those. It also can't be 'sure' in all cases, and if it appears
to be sure it cannot always be correct.

Leaving the prediction to the umpire means about half of lbw decisions
will be wrong, adding DRS to the umpire raises that to 80-90%.

[Brian Lawrence]

I started from Wikipedia:

http://en.wikipedia.org/wiki/Hawk-Eye

In the references at the bottom of the page, number 29 links to a
20-page paper about Hawk-Eye (pdf). The two authors searched for all
online references. The Hawkins' quote came, I think, from an interview
that was on Cricinfo, but it's quoted in their paper.

The 2nd ref. is the Hawk-Eye patent submission, which can be downloaded
(in pdf form).

Regarding accuracy, it might be worth adding that the ICC 3rd umpire
guidelines require them to notify the on-field umpires whenever the
distance between the pitch point and the impact point is less than 40cm
or the distance from impact point to the stumps is greater than 250cm.
They also must report on the actual distances measured - essentially the
on-field umpires will incorporate those distances into their decision.
Basically if the ball strikes the pad shortly after pitching, or hits
the pads a long way (8 feet in old money) from the stumps they will give
the predicted stump impact with caution, since it's not as reliable.

ICC pdf file is here:


"http://static.icc-cricket.yahoo.net/ugc/documents/DOC_D3AEB8DE36A8BD5FA8742FA636A83951_1306651186327_548.pdf"

[Brian Lawrence]

On 27/06/2012 23:30, jzfredricks wrote:
> On Thursday, June 28, 2012 1:34:36 AM UTC+10, Brian Lawrence wrote:
>> On 27/06/2012 15:18, jzfredricks wrote:
>>> On Wednesday, June 27, 2012 11:46:47 PM UTC+10, Brian Lawrence wrote:
>>>> Unless I've misunderstood DRS, xEye doesn't have ANY authority. If a
>>>> decision gets referred it's the TV umpire who looks at what the various
>>>> replays, XEye, Hotspot, etc. indicate happened. IF the TV umpire sees
>>>> enough evidence to suggest that the on-field umpire made a mistake, he
>>>> advises his colleague of that and the decision is reversed. It's NOT
>>>> DRS that decides it's the umpires.
>>>
>>> Yes, you've misunderstood (part of) the DRS.
>>> xEye can provide evidence on 3 things - landing spot, impact spot, and position at stumps.
>>> No other technology in use can provide evidence on those 3, therefore xEye's evidence on these 3 matters is authoritative and cannot be overruled. Obviously taking into account the "umpire's zone/margin of error".
>>> Other technologies (TV replay, hotspot, snicko) can provide different types of evidence which will affect an LBW decision. All 3 of those, though, require a human umpire to analyse the evidence and make a decision.
>>
>> Surely TV replay provides evidence of the first two? OK, there's no
>> point in using
>> it because xEye does that anyway, but it's there and would be more
>> reliable than the umpire - you can review the landing & impact as often
>> as you want in replay, while the poor old ump gets to do it real time
>> with no slo-mo.
>
> Are you saying "Surely it does" or "surely it can"?
> If the former - no it doesn't. TV replays aren't used as part of the current DRS system for LBW appeal landing spots. The times when xEye was optional and not available, LBWs were un-referrable.

I agree that ICC guidelines state that the 3rd ump WILL use the
technology to determine pitch point & impact point. However, the TV
replay(s) ARE part of the process. Suppose that DRS says the ball
pitched just inside leg stump, but TV replay clearly shows that it was
outside, the umpire is going to ignore his own eyes and trust the
technology. In reality he would probably indicate that the technology
appeared to be inconclusive.

> If the latter - I agree. Dump xEye and use the super-imposed red line plus TV replays.
>
> However, I WANT xEye. I think it has potential. I just think any technology that totally eliminates judgement by a human umpire MUST be thoroughly and independently tested.
>
>>> Proper testing MAY have been done*. They might know perfectly well how accurate it is.
>>>
>>> What I want tested is their PUBLISHED accuracy/margins of error. If they say the margin of error is 5mm, and we test it as such, then we can write proper DRS regulations.
>>>
>>> We currently have a 35mm margin of error at some points where their documentation says 5mm. Why is that?
>>
>> I know you have discussed this ad nauseum previously, but I haven't
>> taken much notice, so I don't know where 35mm comes from.
>
> Roughly half a ball's width.
>
>>> On the other hand, maths shows that two 5mm errors can result in a 100mm error at the stumps, but the regulations only assume up to 35mm error at the stumps. Why is that?
>>
>> I assume this means that the landing point might be 5mm in error and so
>> might the impact point, but again I'm unaware how that leads to 100m (or
>> 35mm for that matter). You don't need to explain it again, I'm not
>> particularly bothered.
>
> Pythagoras worked it out 2500 years ago. It's simple maths. 5mm error "north", then 5mm error "south", with a travel distance of 150mm (between pitching and impact) results in a REAL error of about 10cm in the prediction.

I suspect that the 5mm 'north' at pitch and 5mm 'south' at impact
possibility is very remote. Both points are arrived at by triangulation
using six fixed high-speed, HD cameras. Any error at pitch is almost
certainly going to be repeated in the same direction at the impact
point. It's not impossible for a north/south anomaly (or even one in
height), but I expect it to be remote, which is probably why the
Hawk-Eye error at the stumps is apparently 35mm.

You wrote that if a north/south error occurs with 150mm of travel
between pitch & impact it leads to a 100mm REAL error in the prediction.
Surely the predicted error at the stumps depends on
the distance between the impact point and the stumps as well?
If the system predicted where the ball would reach the boundary behind
the stumps the error arising from the 5mm north/south situation would be
a lot more than 10cm - it depends on the distance involved. The
predicted trajectories diverge more the further the ball 'travels'.

>>> If TESTING proves it to be 5mm, then the regulations should reflect that.
>> Surely the regulations don't need to take into account the actual margin
>> of error. In terms of pitching on the wicket (or outside off),
>> if the equipment measures it to pitch outside leg by more than the
>> margin of error then it's definitely not out. If it's measured as more
>> than the width of the ball minus the margin inside the leg stump line
>> then it's pitched on the wicket. With that regulation it matters not
>> whether the margin for error is 1mm, 5mm, 10mm or anything up to half
>> the diameter of the ball, you just apply whatever the margin is. You
>> don't need to write a regulation for 5mm and then rewrite it for 4mm or
>> 6mm. You need to rule on whether the ball is inline or not when it's
>> within the margin of error, you either rule it on the wicket or outside,
>> but again the size of the margin can be flexible.
>
> huh? This makes little sense to me. You say "regulations don't need to know the margin of error", then mention MoE about 30 times.

OK, the regulation (actually it's a guideline, but ..) doesn't
necessarily need to have the actual margin hard-coded into it.
The margin can be different in different scenarios, particularly with
regard to the distance between pitch point and impact point, and between
impact point and the stumps - and, in fact that is actually already in
the guidelines provided by the ICC.

> The current margin of error is, roughly speaking, 35mm. What if the REAL margin of error was 80mm? That means this technology will overrule correct umpiring decisions and give batsmen out when they are not.

If the margin was 80mm that would be true. If.

> If we know the real MoE, then the regulations can be written so that doesn't happen.
>
>>> *The MCC testing mentioned in HawkEye material certainly WASN'T proper testing.
>>
>> Really though, does it matter? The bowler appeals for lbw, the umpire
>> decides it's not out. The fielding side decide to review it. Whatever
>> system is in use isn't going to be 100% accurate and the margin for
>> error might not be known with any certainty, but regardless of that,
>> it's more accurate than the umpire.
>>
>> The umpires get it wrong sometimes, they are human after all, but the
>> technology is demonstrably less fallible, even if not foolproof. Instead
>> of bickering and whinging about it it ought to be used now, and any
>> problems arising dealt with as necessary.
>
> "technology is demonstrably less fallible" how can you say that?

According to the ICC more decisions are correct with DRS than without.
They have reviewed umpires decisions in some detail for a good many
years, their findings show the accuracy has improved. Their data could
be wrong or fabricated, but they make that claim - it's in their annual
reports (I should say that I haven't checked on that recently, but my
recollection is that some data is provided in those reports).

I'll pose a question or two.

How many lbw decisions have been referred to DRS?
How many were by the fielding side and how many by the batsman?
How many were overturned in each case (out or not out)?

How many of those overturned by DRS can be shown to have been incorrect?

[Brian Lawrence]

On 27/06/2012 23:32, Calvin wrote:
> On Thu, 28 Jun 2012 01:34:36 +1000, Brian Lawrence
> <Brian_W_...@msn.com> wrote:
>
>> Surely the regulations don't need to take into account the actual
>> margin of error.
>
> Yes they do, and this is precisely the problem with the current system.
>
> The technology is just a tool. In most cases it is pretty accurate,
> certainly more so than a human. But, and it is a big but, there are
> identified instances (eg where the distance between pitch and impact
> with the pad is short) where it is known to be far less accurate. It is
> therefore crucial that the margin of error in any particular review is
> acknowledged so that the third umpire can include that margin in his
> decision making process.

I have mentioned this elsewhere I think, but the ICC Guidelines for
Third Umpires require them to advise the on-field umpires whenever the
distance between pitching point and impact point is less than 40cm (16
inches) or the distance between impact point and the stumps is more than
250cm (just over 8 feet). In either case they MUST inform the on-field
umpires of the measured distances too, and the on-field umpire must bear
that information in mind when deciding whether to change his decision.
They are not told to ignore the system in those cases, but that they
should not rely on it's accuracy as much as in more 'normal'
circumstances.

> The final decision must always be made by a human, so that person needs
> to be able to make sensible judgments about the accuracy of all the data
> presented. To put it another way, using common sense (rather than being
> locked into set protocols) is the best decision making process.

The on-field umpire still makes the decision - the third umpire provides
him with information based on what he sees and hears from the technology
available to him. The technology doesn't decide, it's just a tool, the
third umpire doesn't decide, he uses the tool to glean information which
he passes on to the on-field umpire. He never says, "that wasn't out" or
the opposite, he just relates what he sees. He is allowed to inform the
on-field umpire if the batsman was out in a different way - if the
umpire gives a batsman out lbw and the batsman refers it and the 3rd
umpires finds that the ball hit the bat first and the ball was then
caught by a fielder, he can say 'it wasn't lbw, but he was caught'.

>> In terms of pitching on the wicket (or outside off),
>> if the equipment measures it to pitch outside leg by more than the
>> margin of error then it's definitely not out.
>
> Agreed. So the margin of error must be known for this occur.
>
>> Really though, does it matter? The bowler appeals for lbw, the umpire
>> decides it's not out. The fielding side decide to review it. Whatever
>> system is in use isn't going to be 100% accurate and the margin for
>> error might not be known with any certainty, but regardless of that,
>> it's more accurate than the umpire.
>
> Not always, and making assumptions that it is, is dangerous.

No one says it always is. However, if batsman or fielding side have any
doubt it can be referred for a closer look. Having looked, the evidence
of the technology, filtered by the third umpire, is relayed to the
on-field umpire who can, if he is convinced, change his decision. If the
umpire believes his original decision was correct he won't change the
decision, although he knows that the evidence appears to show that he
was wrong, and he may have difficulty justifying his stance. The ICC do
review decisions post-match, and an umpire who goes against the
technology might not be asked to umpire again.

It's no more dangerous than umpires getting half of lbw decisions wrong
without DRS. They are also more likely to be cautious without DRS -
evidence clearly shows that there are more lbws than there used to be,
simply because umps are more prepared to make a decision knowing that
DRS will probably confirm or deny it.

[Mike Holmans]

On Thu, 28 Jun 2012 09:38:37 +0100, Brian Lawrence

<Brian_W_...@msn.com> tapped the keyboard and brought forth:

>It's no more dangerous than umpires getting half of lbw decisions wrong
>without DRS. They are also more likely to be cautious without DRS -
>evidence clearly shows that there are more lbws than there used to be,
>simply because umps are more prepared to make a decision knowing that
>DRS will probably confirm or deny it.

I dispute this, and I think the point worth making because it's
gaining currency in an unhelpful way.

The "unhelpful" is that some batsmen say rude things about DRS because
it makes them change their technique, and they cite the umpires'
increasing willingness to give lbws as evidence.

The problem with that is that it's Hawkeye which has caused the
umpires to give more lbws, and that trend was noticeable some time
before any DRS was proposed. Monty Panesar's Test career was mostly in
the pre-DRS era, and he got loads of lbws that wouldn't have been
given 30 years ago because umpires have watched loads of Hawkeye
replays and seen that they can give lbw a lot more often than they
previously thought.

Not having a DRS won't change the fact that the umpires are a lot more
confident about lbw decisions than they used to be. You'd have to ban
Hawkeye from being used at all, even for the private consumption of
umpires wishing to review their own decisions in their hotel rooms,
and then wait for the present crop of Hawkeye-savvy umpires to retire.

Cheers,

Mike


--

[CaraMia]

On 28-06-2012 12:54 PM, Brian Lawrence wrote:
> Leaving the prediction to the umpire means about half of lbw decisions
> will be wrong, adding DRS to the umpire raises that to 80-90%.

These figures just don't seem right, unless you are specifically talking
about a certain Sri Lankan umpire.

I would think that most elite panel umpires would get at least 70-80% of
decisions right. The big mistakes could be rectified by having a very
simple DRS with a slow mo replay and an overlay of a patch between the
stumps at the two ends to check if the ball pitched in line or not. This
can be done at very little extra cost, unlike the highly expensive DRS
technology. As of today, even forgetting other things, I don't think
these things make economic sense. I would take it at 20% of the current
price and not any more. It's still at a early adopter price point.

Another thing is that DRS technology overturns decisions only when the
path is shown to hit a full stump and not just graze a stump. Which
means that even if DRS technology is far more accurate than a human
umpire, it still doesn't bring about the amount of improvement you say
it could.

[Brian Lawrence]

These are in their annual reports and they are not consistent and it's
sometimes unclear exactly what they refer to, but anyway.

All are supposedly the ICC's calculation of the umpires' performance in
making decisions expressed as a percentage of decisions that were judged
to be correct.

Test ODI
2003-04 91.7% 90.7%
2004-05 94.8% 93.4%
2005-06 94.4% 95.2%
2006-07 96%+
2007-08 94.3% 95.1%
2008-09 not recorded
2009-10 93.7% 94% Combined 97%+
2010-11 98.32%
2011-12 98%+

In 2010-11 it reports that before DRS the figure was 93.14% and after
DRS it was 98.32%.

The year runs from April to March.

Also in 2010-11 it reports that DRS was used in 49 games (Tests & ODIs),
umpires made 714 decisions, 183 of those were reviewed (25.6%), and of
those reviewed 37 were reversed (20.2%).

37 out of 714 is 5.2%. In a hypothetical Test where all 40 wickets fell
and all 40 were decided by the umpires that means that two would be
wrong. On average it means about one incorrect decision per Test (ODIs
would be lower because there are fewer wickets/decisions).

The ICC has an Umpire Manager who receives three reports after each
game, two from the captains and one from the referee - all three assess
the umpires performance. He also receives a recording of each game and
reviews all decisions made by the umpires and decides if the decision
was correct or not. That data is reflected above.

[CaraMia]

On 28-06-2012 06:51 PM, Brian Lawrence wrote:
> These are in their annual reports and they are not consistent and it's
> sometimes unclear exactly what they refer to, but anyway.
>
> All are supposedly the ICC's calculation of the umpires' performance in
> making decisions expressed as a percentage of decisions that were judged
> to be correct.
>
> Test ODI
> 2003-04 91.7% 90.7%
> 2004-05 94.8% 93.4%
> 2005-06 94.4% 95.2%
> 2006-07 96%+
> 2007-08 94.3% 95.1%
> 2008-09 not recorded
> 2009-10 93.7% 94% Combined 97%+
> 2010-11 98.32%
> 2011-12 98%+
>
> In 2010-11 it reports that before DRS the figure was 93.14% and after
> DRS it was 98.32%.
>

How exactly was the correctness of the decisions(DRS and manual)
determined?

[Brian Lawrence]

On 28/06/2012 13:43, Mike Holmans wrote:
> On Thu, 28 Jun 2012 09:38:37 +0100, Brian Lawrence
> <Brian_W_...@msn.com> tapped the keyboard and brought forth:
>
>
>> It's no more dangerous than umpires getting half of lbw decisions wrong
>> without DRS. They are also more likely to be cautious without DRS -
>> evidence clearly shows that there are more lbws than there used to be,
>> simply because umps are more prepared to make a decision knowing that
>> DRS will probably confirm or deny it.
>
> I dispute this, and I think the point worth making because it's
> gaining currency in an unhelpful way.
>
> The "unhelpful" is that some batsmen say rude things about DRS because
> it makes them change their technique, and they cite the umpires'
> increasing willingness to give lbws as evidence.
>
> The problem with that is that it's Hawkeye which has caused the
> umpires to give more lbws, and that trend was noticeable some time
> before any DRS was proposed. Monty Panesar's Test career was mostly in
> the pre-DRS era, and he got loads of lbws that wouldn't have been
> given 30 years ago because umpires have watched loads of Hawkeye
> replays and seen that they can give lbw a lot more often than they
> previously thought.

Monty took 137 wickets between March 2006 & July 2009 in matches without
DRS. 37 of those were lbws (27%). In that period he played in 2 Tests in
the WIndies with DRS available. He took 5 wkts, 2 were lbws.
In 2012 he has taken 16 wickets, 8 of which were lbws, so with DRS he
has taken 21 wkts, 10 lbws (47.6%). The DRS numbers are too small to be
significant.

> Not having a DRS won't change the fact that the umpires are a lot more
> confident about lbw decisions than they used to be. You'd have to ban
> Hawkeye from being used at all, even for the private consumption of
> umpires wishing to review their own decisions in their hotel rooms,
> and then wait for the present crop of Hawkeye-savvy umpires to retire.

Here are some more numbers - I calculated the percentage of lbws for 10
slow bowlers with the most Test wickets who played before DRS was
introduced.

Warne 138/708 31.5%
Kumble 156/619 25.2
Gibbs 21/309 6.8
Underwood 24/297 8.1
Bedi 16/266 6.0
Benaud 26/248 10.5
Chandra. 64/242 26.4
A Qadir 52/236 22.0
Grimmett 38/216 17.6
MacGill 33/208 15.9

Warne, Kumble & MacGill bowled after Hawk-Eye was introduced. On that
limited selection Panesar would be second behind Warne.

[Brian Lawrence]

On 28/06/2012 16:12, Brian Lawrence wrote:

> Here are some more numbers - I calculated the percentage of lbws for 10
> slow bowlers with the most Test wickets who played before DRS was
> introduced.
>
> Warne 138/708 31.5%
> Kumble 156/619 25.2
> Gibbs 21/309 6.8
> Underwood 24/297 8.1
> Bedi 16/266 6.0
> Benaud 26/248 10.5
> Chandra. 64/242 26.4
> A Qadir 52/236 22.0
> Grimmett 38/216 17.6
> MacGill 33/208 15.9
>
> Warne, Kumble & MacGill bowled after Hawk-Eye was introduced. On that
> limited selection Panesar would be second behind Warne.

I don't know how I got 31.5% for Warne - that should be 19.4%. All the
others are correct.

[Mike Holmans]

On Thu, 28 Jun 2012 19:26:19 +0530, CaraMia <ca...@mia.com> tapped the
keyboard and brought forth:

How is the correctness of any decision going to be reviewed?

There is no Absolute Great Architect who can be consulted to determine
the truth, nor can you use mechanical means other than those already
part of the DRS panoply.

All that they can do is examine the decisions as they were recorded
and come to a view as to whether the decision was correct or not.

Obviously those doing the reviewing look at Hawkeye. If you wish to
object that they don't know how accurate Hawkeye is, then you need to
explain why they aren't allowed to use their common sense and
experience of the game to judge whether what Hawkeye comes up with is
reasonable. I realise that the people who are busy trying to say that
the possibility of an error of an unspecified few mm here and there
invalidates the entire system specifically rule out the use of common
sense because it would completely destroy their argument, but for
those who think that common sense is perfectly normal in officiating
on cricket matches, this would seem perfectly sensible.

Cheers,

Mike
--

[cricketrulez]

It has been two days since I wrote to the bloke. I haven't heard any
thing back yet. I also tried ringing up the phone no. listed on his
web page. no answer or voice mail.

[Brian Lawrence]

These are the percentages of lbws for all slow bowlers with at least 100
Test wickets.

Span Wkt lbw %age
Saeed Ajmal (Pak) 2009-2012 114 37 32.46
MS Panesar (Eng) 2006-2012 142 39 27.46
GP Swann (Eng) 2008-2012 188 50 26.60
BS Chandrasekhar (Ind) 1964-1979 242 64 26.45
A Kumble (Ind) 1990-2008 619 156 25.20
Abdul Qadir (Pak) 1977-1990 236 52 22.03
Mushtaq Ahmed (Pak) 1990-2003 185 39 21.08
Mohammad Rafique (Ban) 2000-2008 100 21 21.00
DL Vettori (ICC/NZ) 1997-2012 359 74 20.61
SK Warne (Aus) 1992-2007 708 138 19.49
HMRKB Herath (SL) 1999-2012 144 28 19.44
M Muralitharan (ICC/SL) 1992-2010 800 150 18.75
S Ramadhin (WI) 1950-1961 158 29 18.35
WJ O'Reilly (Aus) 1932-1946 144 26 18.06
N Boje (SA) 2000-2006 100 18 18.00
MH Mankad (Ind) 1946-1959 162 29 17.90
CV Grimmett (Aus) 1925-1936 216 38 17.59
SP Gupte (Ind) 1952-1961 149 26 17.45
Saqlain Mushtaq (Pak) 1995-2004 208 35 16.83
JC Laker (Eng) 1948-1959 193 32 16.58
Danish Kaneria (Pak) 2000-2010 261 43 16.48
Harbhajan Singh (Ind) 1998-2011 406 65 16.01
SCG MacGill (Aus) 1998-2008 208 33 15.87
S Venkataraghavan (Ind) 1965-1983 156 24 15.38
FJ Titmus (Eng) 1955-1975 153 23 15.03
PR Adams (SA) 1995-2004 134 19 14.18
AF Giles (Eng) 1998-2006 143 20 13.99
JH Wardle (Eng) 1950-1957 102 14 13.73
HJ Tayfield (SA) 1949-1960 170 23 13.53
EAS Prasanna (Ind) 1962-1978 189 25 13.23
DR Doshi (Ind) 1979-1983 114 15 13.16
Iqbal Qasim (Pak) 1976-1988 171 22 12.87
H Verity (Eng) 1931-1939 144 18 12.50
GAR Lock (Eng) 1952-1968 174 21 12.07
RJ Shastri (Ind) 1981-1992 151 18 11.92
AL Valentine (WI) 1950-1962 139 16 11.51
H Trumble (Aus) 1890-1904 141 16 11.35
C Blythe (Eng) 1901-1910 100 11 11.00
JE Emburey (Eng) 1978-1993 147 16 10.88
PL Harris (SA) 2007-2011 103 11 10.68
R Benaud (Aus) 1952-1964 248 26 10.48
R Illingworth (Eng) 1958-1973 122 12 9.84
JG Bracewell (NZ) 1980-1990 102 9 8.82
W Rhodes (Eng) 1899-1930 127 11 8.66
IWG Johnson (Aus) 1946-1956 109 9 8.26
DL Underwood (Eng) 1966-1982 297 24 8.08
Intikhab Alam (Pak) 1959-1977 125 10 8.00
R Peel (Eng) 1884-1896 101 8 7.92
CL Hooper (WI) 1988-2002 114 9 7.89
NS Yadav (Ind) 1979-1987 102 8 7.84
PCR Tufnell (Eng) 1991-2001 121 9 7.44
DA Allen (Eng) 1960-1966 122 9 7.38
LR Gibbs (WI) 1958-1976 309 21 6.80
AA Mallett (Aus) 1968-1980 132 8 6.06
BS Bedi (Ind) 1966-1979 266 16 6.02
J Briggs (Eng) 1886-1899 118 6 5.08
PH Edmonds (Eng) 1975-1987 125 4 3.20

DRS was first trialed in 2008. Hawk-Eye was first seen in the UK on
Channel 4 in 2001.

[John Hall]

In article <a53431...@mid.individual.net>,

Brian Lawrence <Brian_W_...@msn.com> writes:
>All are supposedly the ICC's calculation of the umpires'
>performance in making decisions expressed as a percentage of
>decisions that were judged to be correct.
>
> Test ODI
>2003-04 91.7% 90.7%
>2004-05 94.8% 93.4%
>2005-06 94.4% 95.2%
>2006-07 96%+
>2007-08 94.3% 95.1%
>2008-09 not recorded
>2009-10 93.7% 94% Combined 97%+
>2010-11 98.32%
>2011-12 98%+

How do you combine 93.7% and 94% and get 97%+? Or are the first two
columns supposed to be for matches without DRS and the third column for
matches with DRS?
--
John Hall
Johnson: "Well, we had a good talk."
Boswell: "Yes, Sir, you tossed and gored several persons."
Dr Samuel Johnson (1709-84); James Boswell (1740-95)

[Brian Lawrence]

On 28/06/2012 19:02, John Hall wrote:
> In article <a53431...@mid.individual.net>,
> Brian Lawrence <Brian_W_...@msn.com> writes:
>> All are supposedly the ICC's calculation of the umpires'
>> performance in making decisions expressed as a percentage of
>> decisions that were judged to be correct.
>>
>> Test ODI
>> 2003-04 91.7% 90.7%
>> 2004-05 94.8% 93.4%
>> 2005-06 94.4% 95.2%
>> 2006-07 96%+
>> 2007-08 94.3% 95.1%
>> 2008-09 not recorded
>> 2009-10 93.7% 94% Combined 97%+
>> 2010-11 98.32%
>> 2011-12 98%+
>
> How do you combine 93.7% and 94% and get 97%+? Or are the first two
> columns supposed to be for matches without DRS and the third column for
> matches with DRS?

Don't know. These are exact quotes:

"The improvement can be seen in their correct decision
percentages (CD%) in ODIs being enhanced from 91.9 per
cent in the previous year to 94 per cent in the same period
of April to March 2009-2010. In Tests the CD% also showed
marginal improvement from 93.1 per cent to 93.7 per cent."

"Overall, the DRS proved to be very effective as
the correct decision percentage moved to a new high of better
than 97 per cent."

That does fill in the 2008-09 figures.

Maybe 97% IS only DRS matches, in which case I misread it first time.

[CaraMia]

Exactly. And that makes the whole report fucking meaningless IMO. The
fact that the report is meaningless doesn't mean that the DRS tools are
not accurate, it just means that the report is stupid.

> All that they can do is examine the decisions as they were recorded
> and come to a view as to whether the decision was correct or not.
>
> Obviously those doing the reviewing look at Hawkeye. If you wish to
> object that they don't know how accurate Hawkeye is, then you need to
> explain why they aren't allowed to use their common sense and
> experience of the game to judge whether what Hawkeye comes up with is
> reasonable.

I think these people who used their judgement and common sense and
experience should be appointed as third umpires. And they can use their
judgement, common sense and experience to judge using slow mo replays
and an overlay to check if it pitched in line without other expensive
tools. This will surely improve the accuracy of decisions. And any other
marginal increase which the tools provide over a non-tooled 3rd umpire
is just not worth the money.

> I realise that the people who are busy trying to say that
> the possibility of an error of an unspecified few mm here and there
> invalidates the entire system specifically rule out the use of common
> sense because it would completely destroy their argument, but for
> those who think that common sense is perfectly normal in officiating
> on cricket matches, this would seem perfectly sensible.

I don't care about unspecified mms of error. I just don't think it's
value for money at current cost. Let DRS be compulsory but let the tools
or the lack of the them be decided by the 2 countries playing the match.
Imposing an additional cost on a Bangladesh-WI test match is a sure fire
way of killing Test Cricket.

[cricketrulez]

On Jun 26, 11:23 am, cricketrulez <cricketrulez1...@gmail.com> wrote:
> For those who care....
>
> "The CEC stated it was "satisfied" with the improvements made over the
> two key components of DRS technology. These include the new Hot Spot
> cameras as well as the independent ball-tracking research conducted by
> Dr Ed Rosten, a Cambridge University department of engineering expert
> in computer-vision technology. Rosten, the ICC said, had "tested the
> accuracy and reliability of ball-tracking in a recent Test series and
> concluded that the results were 100% in agreement with the outcomes
> produced from his assessments."
>
> I believe this is the Ed Rosten they are referring tohttp://www.edwardrosten.com/index.html
>
> So i wrote to Dr. Ed Rosten. I'll post any reply I get back.
>
> Dear Dr. Rosten,
>
> I ran across your name in regards to the accuracy of the Hawkeye
> system used in Cricket to predict the path of the ball.
>
> http://www.espncricinfo.com/ci-icc/content/story/569949.html
>
> "The CEC stated it was "satisfied" with the improvements made over the
> two key components of DRS technology. These include the new Hot Spot
> cameras as well as the independent ball-tracking research conducted by
> Dr Ed Rosten, a Cambridge University department of engineering expert
> in computer-vision technology. Rosten, the ICC said, had "tested the
> accuracy and reliability of ball-tracking in a recent Test series and
> concluded that the results were 100% in agreement with the outcomes
> produced from his assessments."
>
> 1) Will any of the testing information be available to the public?

>
> 2) How were the tests conducted? In a cricket ground? using a  bowling
> machine? If players were used to bowl instead of a bowling machine,
> were both spinners and fast bowlers used?
>

> 3) How exactly did you verify the accuracy of Hawkeye? did you feed
> the system partial footage of deliveries that made it to the keeper?
>
> Given that this system is being used for public consumption and your
> research is being used to endorse its use, I'm hoping you will be able
> to shed some light on this matter.

Here is the reply from Dr. Rosten.

I'd be happy to talk about the ball tracking once I get the go ahead
from the ICC. In the mean time, I've been asked to direct all
enquiries to Colin Gibson at the ICC (colin.gibson@icc-cricket). If
you would like, I can contact you when I can talk about the work.

Regards,

-Ed

[jzfredricks]

On Thursday, June 28, 2012 5:24:26 PM UTC+10, Brian Lawrence wrote:
> DRS at least 50% of lbw decisions were wrong. DRS can, and has, correct
> some of those. It also can't be 'sure' in all cases, and if it appears
> to be sure it cannot always be correct.
>
> Leaving the prediction to the umpire means about half of lbw decisions
> will be wrong, adding DRS to the umpire raises that to 80-90%.

These numbers are wonderful, but they are meaningless IF xEye's margin of error isn't tested and proven.

I see 3 types of affirmed decisions;
a) Either the original errors were picked up on TV replay (no ball etc)

b) other elements in DRS helped correct the error (ie HotSpot or TV replay for indside edges)

c) Referrals that are *pure* xEye, ie Where did it pitch, Where did it strike, and Where would be at the stumps?

I don't know how many fall into each category, but with category C there is no way for the ICC to say "they were correct n% of the time" because they simply have no way to test xEye.
After the live event, we can only provide 2 types of data to the testers; TV replays and xEye data.
If TV replays are enough, by themselves, to work out if a decision was correct.. then why are we using xEye in the first place? Speed?
And surely you can't use xEye data to validate xEye data!

[jzfredricks]

On Thursday, June 28, 2012 6:18:06 PM UTC+10, Brian Lawrence wrote:
> I agree that ICC guidelines state that the 3rd ump WILL use the
> technology to determine pitch point & impact point. However, the TV
> replay(s) ARE part of the process. Suppose that DRS says the ball
> pitched just inside leg stump, but TV replay clearly shows that it was
> outside

Sorry, I didn't realise xEye was that inaccurate. If it is, we shouldn't be using it.

You say DRS, but you mean xEye/balltracking.

[jzfredricks]

On Thursday, June 28, 2012 6:18:06 PM UTC+10, Brian Lawrence wrote:
> I suspect that the 5mm 'north' at pitch and 5mm 'south' at impact
> possibility is very remote. Both points are arrived at by triangulation
> using six fixed high-speed, HD cameras. Any error at pitch is almost
> certainly going to be repeated in the same direction at the impact
> point. It's not impossible for a north/south anomaly (or even one in
> height), but I expect it to be remote, which is probably why the
> Hawk-Eye error at the stumps is apparently 35mm.

Suspect? Suspect? How can you say that? We can't base cricket regulations on some random bloke's gut feeling (forgive me for being flippant).

You and I have NO idea how often the north/south errors occur, and it's dangerous to pretend we do.

No, the current MoE isn't "35mm", it's "half the ball" (which is roughly 35mm). A purely arbitrary figure made up because it looks and sounds good.

The 100mm MoE I described if based on the published MoE of 5mm. If the REAL MoE (in the initial 2 readings) is more than 5mm, then that 100m will blow out significantly.

On other thing not yet mentioned in this thread - xEye staff *manually select* the point of impact in some (how many? no idea) referrals. Has THAT been tested?

> You wrote that if a north/south error occurs with 150mm of travel
> between pitch & impact it leads to a 100mm REAL error in the prediction.
> Surely the predicted error at the stumps depends on
> the distance between the impact point and the stumps as well?

Yeah, sorry. My figures were based on "worst case", so it was at a distance of 250cm. Best case is it's 100% accurate, obviously.

> If the system predicted where the ball would reach the boundary behind
> the stumps the error arising from the 5mm north/south situation would be
> a lot more than 10cm - it depends on the distance involved. The
> predicted trajectories diverge more the further the ball 'travels'.

Yeah. I used to say "as the distance between pitching and impact approaches 0, the error at the stumps approaches infinity". This confused the average reader, so I just changed it to the "100mm error" speech.

> > huh? This makes little sense to me. You say "regulations don't need to know the margin of error", then mention MoE about 30 times.
>
> OK, the regulation (actually it's a guideline, but ..) doesn't
> necessarily need to have the actual margin hard-coded into it.
> The margin can be different in different scenarios, particularly with
> regard to the distance between pitch point and impact point, and between
> impact point and the stumps - and, in fact that is actually already in
> the guidelines provided by the ICC.

Yeah, but you still need to know the tested/proven MoE!
But I do agree, the MoE for pitching/impact should NOT be 35mm, as I SUSPECT that the real MoE for these reading LESS than 35mm.
My main beef is the MoE at the stumps, which gets compounded by any north+south errors previously described.

This is the ideal scenario for me;
Pitching/impact MoE is tested/proven to be 5mm.
DRS regulation for pitching/impact allow for a 6mm (or 10mm if you want to be sure)(but if proven, I'd be happy with 6mm) MoE
The MoE at the stumps is based on a WORST case scenario for that particular delivery, ie 6mm/10mm (depending on which we go with) error north at pitching, 6mm/10mm error south at impact, then work out angles based on distances from stump.
We NEED this calculated MoE at stumps as science tells us we do!

> > The current margin of error is, roughly speaking, 35mm. What if the REAL margin of error was 80mm? That means this technology will overrule correct umpiring decisions and give batsmen out when they are not.
>
> If the margin was 80mm that would be true. If.

If? If? So how do we rule out the if? We do thorough and public testing. That's how.

> According to the ICC more decisions are correct with DRS than without.

We're discussing one element of DRS, xEye.

> I'll pose a question or two.
>

> How many of those overturned by DRS can be shown to have been incorrect?

Without proper testing of xEye, or a replacement technology, we'll never be able to answer that.

[jzfredricks]

On Thursday, June 28, 2012 6:38:37 PM UTC+10, Brian Lawrence wrote:
> It's no more dangerous than umpires getting half of lbw decisions wrong
> without DRS.

I'd like to see how many LBW referrals are overturned based on HotSpot evidence. It detects a lot of inside edges or not.

[jzfredricks]

On Thursday, June 28, 2012 10:43:18 PM UTC+10, Mike Holmans wrote:
> The "unhelpful" is that some batsmen say rude things about DRS because
> it makes them change their technique, and they cite the umpires'
> increasing willingness to give lbws as evidence.

Another issue that might be upsetting to players is we're asking them to be umpires, and we're punishing them when they fail at it.

[jzfredricks]

On Thursday, June 28, 2012 11:18:56 PM UTC+10, CaraMia wrote:
> Another thing is that DRS technology overturns decisions only when the
> path is shown to hit a full stump and not just graze a stump. Which
> means that even if DRS technology is far more accurate than a human
> umpire, it still doesn't bring about the amount of improvement you say
> it could.

All technology is flawed, and no technology will ever be perfect.
The big margins of error currently in the DRS regulations reflect that.
Do they reflect it enough, though?

[jzfredricks]

On Friday, June 29, 2012 1:40:28 AM UTC+10, Mike Holmans wrote:
> I realise that the people who are busy trying to say that
> the possibility of an error of an unspecified few mm here and there
> invalidates the entire system specifically rule out the use of common
> sense because it would completely destroy their argument,

You're just being obtuse. I suspect deliberately so.

All technology is flawed, and will have errors. I'm just saying that until we understand just how error prone the technology is we shouldn't use it. Furthermore, once we DO know a technology's limits, the DRS regulations should take them into account.

I think that's a very sensible position to take, especially when testing should be fairly easy to do.

[alvey]

On Friday, June 29, 2012 11:30:06 AM UTC+10, jzfredricks wrote:
> On Friday, June 29, 2012 1:40:28 AM UTC+10, Mike Holmans wrote:
> > I realise that the people who are busy trying to say that
> > the possibility of an error of an unspecified few mm here and there
> > invalidates the entire system specifically rule out the use of common
> > sense because it would completely destroy their argument,
>
> You're just being obtuse. I suspect deliberately so.

Can you prove that?

> All technology is flawed, and will have errors.

Pointless, melodramatic generalisation.

> I'm just saying that until we understand just how error prone the technology > is we shouldn't use it. Furthermore, once we DO know a technology's limits, the > DRS regulations should take them into account.

Absolutely! And until this DRS technology is PROVEN BY THOROUGH, PUBLIC TESTING, several times, and my pet hamster likes it, and it's available in more colour options, and accessories, "we" should revert to using that THOROUGLY PROVEN infallible cluster of machines, the EUP. What's the odd accusation of corruption hey! Even if it is PROVEN.

> I think that's a very sensible position to take,

Well, you would say that wouldn't you?

> especially when testing should be fairly easy to do.

It is. But it's achieving that 120% accuracy that the looney fringe are demanding that's PROVING a real bitch.



alvey
wondering what the digital equivalent of a cracked record is.

[jzfredricks]

On Friday, June 29, 2012 1:57:36 PM UTC+10, alvey wrote:
> On Friday, June 29, 2012 11:30:06 AM UTC+10, jzfredricks wrote:
> > On Friday, June 29, 2012 1:40:28 AM UTC+10, Mike Holmans wrote:
> > > I realise that the people who are busy trying to say that
> > > the possibility of an error of an unspecified few mm here and there
> > > invalidates the entire system specifically rule out the use of common
> > > sense because it would completely destroy their argument,
> >
> > You're just being obtuse. I suspect deliberately so.

> It is. But it's achieving that 120% accuracy

Perhaps Mike was talking about you and your ilk. If so, I owe him an apology.

[alvey]

The vet said that my ilk was in perfect health.

> If so, I owe him an apology.

What did yours say about you?




alvey

[jzfredricks]

On Friday, June 29, 2012 1:57:36 PM UTC+10, alvey wrote:

> Absolutely! And until this DRS technology is PROVEN BY THOROUGH, PUBLIC TESTING, several times,

0 != several times

[cricketrulez]

I have a bad feeling that this "testing" was actually validating one
prediction model using another and did not involve comparing the
predictive element of HE to a real event.

[jzfredricks]

On Friday, June 29, 2012 2:17:08 PM UTC+10, alvey wrote:
> The vet said that my ilk was in perfect health.

It's no wonder with all the loving you give it

[jzfredricks]

On Friday, June 29, 2012 2:35:44 PM UTC+10, cricketrulez wrote:
> I have a bad feeling that this "testing" was actually validating one
> prediction model using another and did not involve comparing the
> predictive element of HE to a real event.

Strictly speaking the predictive element is EASY.
The true magic is gathering *accurate* initial readings!

[alvey]

DRS > EUP

[Brian Lawrence]

On 29/06/2012 01:50, jzfredricks wrote:

> On Thursday, June 28, 2012 6:18:06 PM UTC+10, Brian Lawrence wrote:
>> I suspect that the 5mm 'north' at pitch and 5mm 'south' at impact
>> possibility is very remote. Both points are arrived at by triangulation
>> using six fixed high-speed, HD cameras. Any error at pitch is almost
>> certainly going to be repeated in the same direction at the impact
>> point. It's not impossible for a north/south anomaly (or even one in
>> height), but I expect it to be remote, which is probably why the
>> Hawk-Eye error at the stumps is apparently 35mm.
>
> Suspect? Suspect? How can you say that? We can't base cricket regulations on some random bloke's gut feeling (forgive me for being flippant).

Sorry, I didn't realise my ramblings had any importance whatsoever.

> You and I have NO idea how often the north/south errors occur, and it's dangerous to pretend we do.

Dangerous, that word again. Emotive, but meaningless in this context.

> No, the current MoE isn't "35mm", it's "half the ball" (which is roughly 35mm). A purely arbitrary figure made up because it looks and sounds good.
>
> The 100mm MoE I described if based on the published MoE of 5mm. If the REAL MoE (in the initial 2 readings) is more than 5mm, then that 100m will blow out significantly.
>
> On other thing not yet mentioned in this thread - xEye staff *manually select* the point of impact in some (how many? no idea) referrals. Has THAT been tested?

Yes, I've heard that, but had no idea what it meant. On reflection I
think it might be that the exact pitching point isn't necessarily
captured in one frame, but is somewhere between frames. Do you pick the
previous frame or the next one, can you estimate whether it's nearer to
one or the other? Are the cameras now 230fps? How far does an object
travel in one 230th of a second at say 50mph? 30cm? 40cm? Hm.

One problem with this testing would be that the errors that are present
in establishing the pitch point and impact point will obviously be
variable. The hardware (cameras mainly) need to be installed at the
venue before each game - equipping every International ground with
permanent fixed installations would be much more expensive.

The installation process will probably introduce slight errors, but
there will be a calibration process which should reduce those. However,
these errors are specific to that ground and that installation process,
by that installation team, on that date in history, with whatever
ambient weather conditions are present. The MoE produced by testing at
Lord's in June will be different from Melbourne in December, or even
from Lord's in July.

The differences are probably very slight, and they can 'tune' the system
to make it better. Then again, if it's tuned before the match in sunny,
hot weather and is then used in dark overcast maybe windy conditions the
margins will most likely have changed.

The xEye companies (two?) will have a lot of data, they may make that
available to boards and/or the ICC, but probably not. They just need the
'executive summary' that says about 5mm.

Not sure how you could test the accuracy of the pitching point though,
cameras fixed in position a few feet from the pitch perhaps. Lines
painted on the pitch, a grid pattern.

[Mike Holmans]

On Thu, 28 Jun 2012 18:30:06 -0700 (PDT), jzfredricks
<jzfre...@gmail.com> tapped the keyboard and brought forth:

>On Friday, June 29, 2012 1:40:28 AM UTC+10, Mike Holmans wrote:
>> I realise that the people who are busy trying to say that
>> the possibility of an error of an unspecified few mm here and there
>> invalidates the entire system specifically rule out the use of common
>> sense because it would completely destroy their argument,
>
>You're just being obtuse. I suspect deliberately so.

Have you tested that statement? In public, with all the other idiotic
qualifiers you always want to add to the testing?

As to being obtuse deliberately, are you saying your 179-degree
obtuseness is accidental?

But, are you saying that applying common sense is sensible? If so, why
are you continuing to fail to use any?

>All technology is flawed, and will have errors. I'm just saying that until we understand just how error prone the technology is we shouldn't use it. Furthermore, once we DO know a technology's limits, the DRS regulations should take them into account.
>
>I think that's a very sensible position to take, especially when testing should be fairly easy to do.

The trouble is that testing has been done, but you won't accept it
becuase you weren't there.

As I have said, on this issue, you have long since departed this
planet and are out there with the cretins who won't believe the proof
that Obama was born in the USA.

Cheers,

Mike
--

[Brian Lawrence]

"http://stats.espncricinfo.com/ci/engine/records/team/match_results.html?class=1;id=2011;type=year"


"http://www.cricketarchive.co.uk/Archive/Seasons/Lists/2010-11_t_Match_List.html"

Tests only, I wouldn't bother with ODIs.

[jzfredricks]

On Friday, June 29, 2012 5:54:00 PM UTC+10, Brian Lawrence wrote:
> > On other thing not yet mentioned in this thread - xEye staff *manually select* the point of impact in some (how many? no idea) referrals. Has THAT been tested?
>
> Yes, I've heard that, but had no idea what it meant. On reflection I
> think it might be that the exact pitching point isn't necessarily
> captured in one frame, but is somewhere between frames. Do you pick the
> previous frame or the next one, can you estimate whether it's nearer to
> one or the other? Are the cameras now 230fps? How far does an object
> travel in one 230th of a second at say 50mph? 30cm? 40cm? Hm.

https://groups.google.com/d/msg/rec.sport.cricket/f4F-wpJZjcQ/xFKt1clNiq8J

[jzfredricks]

On Friday, June 29, 2012 6:02:27 PM UTC+10, Mike Holmans wrote:
> >I think that's a very sensible position to take, especially when testing should be fairly easy to do.
>
> The trouble is that testing has been done, but you won't accept it
> becuase you weren't there.

The only 3rd party testing I've seen mention of was the MCC's and their methodology was comical.
Plus ofc this round of testing, which to my knowledge isn't complete.

[jzfredricks]

On Friday, June 29, 2012 5:54:00 PM UTC+10, Brian Lawrence wrote:
> Not sure how you could test the accuracy of the pitching point though

If we can't test its accuracy, why are we using it?

Here's another of my post regarding accuracy;
https://groups.google.com/d/msg/rec.sport.cricket/f4F-wpJZjcQ/xFKt1clNiq8J

[mcenley]

Brian, It's still befuddles to me to understand that you think the decisions by algorithms are "correct" an humans are "wrong". Please note, that algorithms were made to mimic human decisions not the other way around.

On Thursday, June 28, 2012 3:24:26 PM UTC+8, Brian Lawrence wrote:
> On 28/06/2012 07:40, mcenley wrote:
> >> The only real issue with DRS is the accuracy of it's predictive
> >> function, in particularly regarding lbw decisions, but again, using
> >> computer analysis must give a more accurate prediction of ball
> >> trajectory than the umpire's brain.
> >
> > Once again you keep on making this point regarding a "computer" is more accurate at predicting than a human brain. You can only make that statement if you understand both aspects completely - [1] A computer and [2] A human brain. We understand [1] very well but [2] we haven't cracked the code. So, currently at best we are comparing a "fully known" system to a "not so completely known" entity. The best example I can take to prove you, by your logic, is that a computer would find it easy to match human faces. However, one just need to speak to some Computer Vision experts (or particularly Biometric researchers) as to why there's no way that any computer analysis in this world can match the human brain cognition for this task.
> >
> > Yes - the prediction is the problem and in the current scenario it is best to leave the prediction to the umpire. Also, let me add that it's the prerogative of automated systems to prove that they are better than humans and not vice versa.
>
> You are writing in more general terms than I did. I was writing of the
> DRS systems in use in comparison to the umpire. The computer-based
> system is measuring pitching point - is it on the wicket or outside off
> stump, and the impact point with the batsman, is it within the bounds of
> the stumps. It then projects the ball's measured trajectory to decide if
> it would have hit the stumps. It does this using various algorithms and
> mathematic principles which are clearly defined.
>
> The umpire uses the evidence of his eyes, his recollection of what just
> happened and his experience. He combines that to make an almost instant
> decision. A decision which has often been proven wrong in the past. He
> should only give the batsman out if he has no doubt - the batsman is
> said to receive the benefit of the doubt. It's estimated that before

> DRS at least 50% of lbw decisions were wrong. DRS can, and has, correct
> some of those. It also can't be 'sure' in all cases, and if it appears
> to be sure it cannot always be correct.
>
> Leaving the prediction to the umpire means about half of lbw decisions
> will be wrong, adding DRS to the umpire raises that to 80-90%.
>

[Brian Lawrence]

On 29/06/2012 10:23, jzfredricks wrote:
> On Friday, June 29, 2012 5:54:00 PM UTC+10, Brian Lawrence wrote:
>> Not sure how you could test the accuracy of the pitching point though
>
> If we can't test its accuracy, why are we using it?

Well, we've been using fallible human umpires (in Tests) for more than
130 years. Technology can double check their decisions, whuy wouldn't we
be using it?

> Here's another of my post regarding accuracy;
> https://groups.google.com/d/msg/rec.sport.cricket/f4F-wpJZjcQ/xFKt1clNiq8J

[Brian Lawrence]

On 29/06/2012 10:24, mcenley wrote:
> Brian, It's still befuddles to me to understand that you think the decisions by algorithms are "correct" an humans are "wrong". Please note, that algorithms w
were made to mimic human decisions not the other way around.

Again, you seem to be writing in more general terms and attributing what
I may have written as being generic rather than specific.

Refresh my memory, where did I write that algorithms made any decisions?
Or, that humans are wrong? If a human makes a decision it may be wrong
or it may be right, or it may not be possible to decide. Humans are like
that.

I am not aware of any 'algorithms' involved in the DRS that mimic human
decisions, or in fact that make any decisions. The DRS gathers data and
presents it to humans who then decide (or not, sometimes they can't decide).

[Brian Lawrence]

On 29/06/2012 10:15, jzfredricks wrote:
> On Friday, June 29, 2012 5:54:00 PM UTC+10, Brian Lawrence wrote:
>>> On other thing not yet mentioned in this thread - xEye staff *manually select* the point of impact in some (how many? no idea) referrals. Has THAT been tested?
>>
>> Yes, I've heard that, but had no idea what it meant. On reflection I
>> think it might be that the exact pitching point isn't necessarily
>> captured in one frame, but is somewhere between frames. Do you pick the
>> previous frame or the next one, can you estimate whether it's nearer to
>> one or the other? Are the cameras now 230fps? How far does an object
>> travel in one 230th of a second at say 50mph? 30cm? 40cm? Hm.
>
> https://groups.google.com/d/msg/rec.sport.cricket/f4F-wpJZjcQ/xFKt1clNiq8J

Thanks. I wonder exactly what sort of frame they are selecting. I guess
we all think of video frames, but they have 4/5/6 cameras - are they all
time synched so that the nth frame from each camera was taken at the
same instant. Logically it might be better if all were timed
differently, so camera 1 takes a frame followed by cameras 2, 3, 4, 5,
6, and then back to 1. If all were at 100 fps, then each second would
have 600 frames of data.

I think it's more likely the frame they are choosing is from the
animation movie that the third umpire sees (and the tv audience).
If that is so, it is probably at a much lower frame rate - I don't
know what that would be, 24fps used to standard for movie film, but
I imagine it would be a bit higher these days. I seem to recall HD tv is
around 60?

[Mike Holmans]

On Fri, 29 Jun 2012 12:24:29 +0100, Brian Lawrence
<Brian_W_...@msn.com> tapped the keyboard and brought forth:

>On 29/06/2012 10:23, jzfredricks wrote:
>> On Friday, June 29, 2012 5:54:00 PM UTC+10, Brian Lawrence wrote:
>>> Not sure how you could test the accuracy of the pitching point though
>>
>> If we can't test its accuracy, why are we using it?
>
>Well, we've been using fallible human umpires (in Tests) for more than
>130 years. Technology can double check their decisions, whuy wouldn't we
>be using it?

Because it is better to accept hundreds of wrong decisions by human
umpires than to allow the remote possibility that an umpire might
guess right on a hairline decision which the technology could get
wrong if it were at the limit of its tolerances, and that the
discussion between the third umpire and the standing umpire could then
result in the on-field umpire deciding that there was convincing
evidence that he was wrong when in fact he wasn't (even though no-one
would have the means to show that the on-field umpire was in fact
right the first time). Better to make sure that a once-in-forty-years
event cannot happen than to correct hundreds of clear errors made in
the meantime.

When one is confronted by fanaticism of that order, there is little
point in arguing the toss.

Cheers,

Mike

--

[alvey]

Jealous.

[jzfredricks]

On Friday, June 29, 2012 11:22:20 PM UTC+10, Brian Lawrence wrote:
> Thanks. I wonder exactly what sort of frame they are selecting. I guess
> we all think of video frames, but they have 4/5/6 cameras - are they all
> time synched so that the nth frame from each camera was taken at the
> same instant.

No idea how they're setup. I *image* they are simply turned on and let run, and *not* tuned/syncronised like you've described.

But as Phil. points out in my other thread, there's a second issue - exposure time. A (relatively) slow exposure time means there is motion blur.

There are so many questions. Public (or independent) and thorough testing will answer them.

[Bharat Rao]

On Friday, June 29, 2012 4:50:05 PM UTC-4, Mike Holmans wrote:

> Because it is better to accept hundreds of wrong decisions by human
> umpires than to allow the remote possibility that an umpire might
> guess right on a hairline decision which the technology could get
> wrong if it were at the limit of its tolerances, and that the
> discussion between the third umpire and the standing umpire could then
> result in the on-field umpire deciding that there was convincing
> evidence that he was wrong when in fact he wasn't (even though no-one
> would have the means to show that the on-field umpire was in fact
> right the first time). Better to make sure that a once-in-forty-years
> event cannot happen than to correct hundreds of clear errors made in
> the meantime.
>
> When one is confronted by fanaticism of that order, there is little
> point in arguing the toss.
>

I've stayed out of this one, but I'll bite. The problem is that the technology is not used to eliminate howlers. When a difference of 1mm in where the ball pitched / or was projected to hit the stumps, leads to either an overturning of a decision or losing an appeal, there is nothing "one in forty years" about it. This happens every match.

I have no problem with using DRS. I have huge problems with the slavish equivalence of objecting to the CURRENT use of DRS to being anti-technology. I'm one of the most pro-technology folks around -- it is patently obvious to me that DRS is flawed -- not the technology per se, but the decision making around the technology.

a) DRS makes knife-edge decisions (at the 1-2mm projection level) that have huge impact (reverse / preserve a decision or lose an appeal), which introduces a huge element of randomness in whether you should use an appeal or not.
b) DRS is now a tactical tool in the captain's arsenal. People talk about wasting reviews -- I have a fundamental problem with this. If the goal is to improve decisions, then it should not be a TACTICAL tool -- about when to use DRS reviews and not (and when the 1-2mm difference will make a huge impact).
c) The error projections are clearly naive. It is obvious to anyone schooled in even middle-school physics that the levels of error in estimation will depend upon the speed of ball, degree of deviation from the pitch, distance traveled between impact on ground and impact on pad. Yet the errors are uniformly applied.
d) The rules of what constitutes an over-rule on a catch are unclear. Is the absence of Hot Spot sufficient to reverse a decision? Is it clear that faint nicks won't appear?
e) The decision-making is not uniform by any means. At least every 2 or 3 Tests there is a review and even after the slow-mo & projections the commentators and viewers are not sure if it will be reversed or upheld. And I'm not even talking about the variation in decision-making across 3rd umpires.

The technology has flaws -- that is not the problem. It is the implementation of the decision-making around DRS that is the problem. And that is my objection to it. Most of all, about the fact that it is now a tactical tool.

People say "how else can you use it?" Thats not the point. There are far better ways of using DRS (even with the current flaws).

A few quick pointers:
a) reversal only upon clear error (with definitions of what constitutes a clear error)
b) clear differences in "margin of error" in reversing an out versus an not-out
c) no batting reviews. All 10 outs automatically reviewed. 3rd umpire has 2 minutes (before other umpire takes guard)
d) fielding reviews to lose their appeal only if clearly wrong
e) systematic review of 3rd umpires, with fines if they fail to make the correct rule / over-rule in the time allocated. Clear reports showing uniform decision making.

All this put in place today, and then rigorous testing about the margin of error in the technology to refine the "margins of error."

Oh and please, can we have umpires make a decision on on-field close-to-the-ground catches and overturn them only if the camera shows it was wrong. Right now every close catch is not out if the umpire takes it up.

Except for run outs and stumped (where the technology is pretty much 98% given the non-moving crease and the relatively slow moving batsman) always make an on-field decision. (And yes, runouts aren't 100% either...)

Bharat
--

[Mike Holmans]

On Sat, 30 Jun 2012 02:56:05 -0700 (PDT), Bharat Rao
<rao.b...@gmail.com> tapped the keyboard and brought forth:

>On Friday, June 29, 2012 4:50:05 PM UTC-4, Mike Holmans wrote:
>
>> Because it is better to accept hundreds of wrong decisions by human
>> umpires than to allow the remote possibility that an umpire might
>> guess right on a hairline decision which the technology could get
>> wrong if it were at the limit of its tolerances, and that the
>> discussion between the third umpire and the standing umpire could then
>> result in the on-field umpire deciding that there was convincing
>> evidence that he was wrong when in fact he wasn't (even though no-one
>> would have the means to show that the on-field umpire was in fact
>> right the first time). Better to make sure that a once-in-forty-years
>> event cannot happen than to correct hundreds of clear errors made in
>> the meantime.

> When a difference of 1mm in where the ball pitched / or was projected to hit the stumps, leads to either an overturning of a decision or losing an appeal, there is nothing "one in forty years" about it. This happens every match.

But it doesn't.

As to where accuracy of where it pitched, even the BCCI don't have a
problem with that for whether it pitched/hit the pad inside/outside
the line of the stumps. That's the bit of Hawkeye which tennis uses
and no-one gets their knickers in a twist about it.

A difference of 1 mm in where it was projected to hit the stumps will
result in "umpire's call", so the decision would not be overruled.

Constructive suggestions for modifying the protocols such as yours
don't bother me. I might not agree with them all, but they aren't
silly. I may bother to debate your suggestions, but not in this post.

What bothers me is jzf (and others) drivelling on about how the
testing has been completely inadequate, and that *therefore* we should
not use it *at all*.

The system has clearly been tested by its makers. It has been tested
again under the auspices of the ICC. Don't start wittering on about 14
trials being insufficient to test a computer program: we are talking
here about user acceptance testing, not initial unit testing by the
programmer. If you run 14 tests and the result of every single one is
exactly in line with the expected result, no-one is going to blame a
manager for signing off the program as being fit for production use.

Of course the system is not accurate enough for purposes such as
finding a Higgs boson or landing a probe on the north pole of Mars.
The question is whether it is fit for use in a production environment,
here on Earth, for helping umpires make decisions in cricket matches.

This is surely your field: when you market medical diagnostic software
as an aid for doctors, given all the testing you've already put into
it and the caveats you issue regarding its use, and if a Medical
Research Authority (I obviously don't know the precise name of the
relevant body in any country) had also tested it and pronounced it fit
for purpose, how impressed would you be with fanatical campaigners who
said that the testing was inadequate and the technology should be
banned?

Cheers,

Mike


--

[jzfredricks]

On Saturday, June 30, 2012 7:56:05 PM UTC+10, Bharat Rao wrote:
> People say "how else can you use it?" Thats not the point. There are far better ways of using DRS (even with the current flaws).
>
> A few quick pointers:
> a) reversal only upon clear error (with definitions of what constitutes a clear error)

I thought that was the regulation already. Ultimately the 3rd ump is human, and humans interpret things differently. I've seen some bizarre DRS decisions. I don't have a big problem with this, because overall DRS makes things so much better.

> b) clear differences in "margin of error" in reversing an out versus an not-out

Again, I think this is already the case (for xEye at least)

For on-field "Not Out" calls, there is a big margin of error applied. The *center* of the ball (roughly 35mm radius) must be hitting the stumps. However, the "stumps" also have a margin of error built in - the zone starts at the *lower* edge of the bails, and the *middle* of the leg/off stumps.
If we assume the stumps are 40mm diam, and the ball 70mm diam, it means this - 55mm of the ball could be INSIDE the graphical representation of the stumps, and the Not Out decision would be upheld.
Yep, the viewer could see 55mm of ball hitting the stumps, but it would still be not out.

For on-field "Out" calls, however, there is less of a buffer. xEye just has to predict that no part of the ball is touching any part of the stump (and there is no "buffer" on the stump area in these cases).

Same goes for pitching and impact location. For Not Outs, there's a 35mm ("center of ball") buffer in favour of the original not out call.

> c) no batting reviews. All 10 outs automatically reviewed. 3rd umpire has 2 minutes (before other umpire takes guard)

I've no problem with that, although I don't think umpires should be batting :)

> d) fielding reviews to lose their appeal only if clearly wrong

I've suggested this myself, but I can't think of a way to define "clearly wrong". I think my best, last suggestion was "worked out on 2 TV replays" or something like that.

> e) systematic review of 3rd umpires, with fines if they fail to make the correct rule / over-rule in the time allocated. Clear reports showing uniform decision making.

No thanks. Reviews, YES YES YES. Please. Fines? No thanks.

[jzfredricks]

Sorry Brian, I pasted the same link twice!
Here's the 2nd one;
https://groups.google.com/d/msg/rec.sport.cricket/yXIgl6dx-2g/OEAJlnm-MssJ

[jzfredricks]

On Saturday, June 30, 2012 10:37:48 PM UTC+10, Mike Holmans wrote:
> The system has clearly been tested by its makers.

I have multiple beefs, one being if we assume xEye's own marketing brochure as being true, and the real MoE *is* 5mm, then the current DRS regulations don't allow for a big enough "umpire's buffer" in some, rarer, scenarios.

Ball pitching 250cm from stumps and a travel distance of 40cm before impact (inside 400mm and they clasify it "extreme") will result in an "At Stump" error of 67mm. Yep, the prediction will be out by 67mm.

A travel distance of 20cm and that 67mm blows out to ~80mm. 10cm travel distance == prediction error of 100mm. The DRS regulations simply don't take that into account.

[Phil.]

On Jun 28, 8:50 pm, jzfredricks <jzfredri...@gmail.com> wrote:
> On Thursday, June 28, 2012 6:18:06 PM UTC+10, Brian Lawrence wrote:
I first wrote about a system like this on here about 14 years ago,
since then the technology has improved substantially.
http://groups.google.com/group/rec.sport.cricket/msg/70c332ac6e2e7e5c

> On other thing not yet mentioned in this thread - xEye staff *manually select* the point of impact in some (how many? no idea) referrals. Has THAT been tested?

I'm not sure how they'd do it but how I'd do it would be to inspect
the series of frames an click on the one where the ball reverses its
vertical motion, the software would select the 5 frames (arbitrary
choice) before and after, fit quadratic to each set of data and
interpolate the x-y position of reversal (this is trivial to do). The
multiple cameras give you data for the three dimensional position in
space of the ball at specified times (which is why the cameras are
synched). The time separation is determined by the frame rate and the
precision of the position by the shutter time (a millisec at most,
that's the resolution in automatic timing systems in track with which
I've worked), the x-y-z position resolution will be defined by the
imaging system pixel size (much better than I described 14 years
ago!) This gets around the issue of the ball not bouncing at the
exact time the frame is exposed which by definition it won't.

>
> > You wrote that if a north/south error occurs with 150mm of travel
> > between pitch & impact it leads to a 100mm REAL error in the prediction.
> > Surely the predicted error at the stumps depends on
> > the distance between the impact point and the stumps as well?
>
> Yeah, sorry. My figures were based on "worst case", so it was at a distance of 250cm. Best case is it's 100% accurate, obviously.
>
> > If the system predicted where the ball would reach the boundary behind
> > the stumps the error arising from the 5mm north/south situation would be
> > a lot more than 10cm - it depends on the distance involved. The
> > predicted trajectories diverge more the further the ball 'travels'.
>
> Yeah. I used to say "as the distance between pitching and impact approaches 0, the error at the stumps approaches infinity". This confused the average reader, so I just changed it to the "100mm error" speech.

It really doesn't because the of the Laws of Cricket! If the distance
is 0 you don't need to know what the ball has done between pitching
and impact because the Laws say that in that case the umpire must only
consider the trajectory of the ball before impact and there's plenty
of data for that. You need about 3 ball positions between pitching
and impact to determine a subsequent trajectory (the more the better),
still far better than a human ump! So apply that provision of the
Laws if you have insufficient data to tell (which is the reason for
that provision for human umps)


Phil.

[Mike Holmans]

On Sat, 30 Jun 2012 02:56:05 -0700 (PDT), Bharat Rao

<rao.b...@gmail.com> tapped the keyboard and brought forth:

>I have no problem with using DRS. I have huge problems with the slavish equivalence of objecting to the CURRENT use of DRS to being anti-technology. I'm one of the most pro-technology folks around -- it is patently obvious to me that DRS is flawed -- not the technology per se, but the decision making around the technology.
>
>a) DRS makes knife-edge decisions (at the 1-2mm projection level) that have huge impact (reverse / preserve a decision or lose an appeal), which introduces a huge element of randomness in whether you should use an appeal or not.
>b) DRS is now a tactical tool in the captain's arsenal. People talk about wasting reviews -- I have a fundamental problem with this. If the goal is to improve decisions, then it should not be a TACTICAL tool -- about when to use DRS reviews and not (and when the 1-2mm difference will make a huge impact).
>c) The error projections are clearly naive. It is obvious to anyone schooled in even middle-school physics that the levels of error in estimation will depend upon the speed of ball, degree of deviation from the pitch, distance traveled between impact on ground and impact on pad. Yet the errors are uniformly applied.
>d) The rules of what constitutes an over-rule on a catch are unclear. Is the absence of Hot Spot sufficient to reverse a decision? Is it clear that faint nicks won't appear?
>e) The decision-making is not uniform by any means. At least every 2 or 3 Tests there is a review and even after the slow-mo & projections the commentators and viewers are not sure if it will be reversed or upheld. And I'm not even talking about the variation in decision-making across 3rd umpires.
>
>The technology has flaws -- that is not the problem. It is the implementation of the decision-making around DRS that is the problem. And that is my objection to it. Most of all, about the fact that it is now a tactical tool.
>
>People say "how else can you use it?" Thats not the point. There are far better ways of using DRS (even with the current flaws).

Since you're being sensible, this is worth discussing.

The first point I'd make is that the DRS protocols currently in place
are not:

- the latest Republican proposal to "improve" the Affordable Care Act
- the latest Pakistani proposal to settle the Kashmir question
- an alimony/property-split claim in an acrimonious divorce
- part of a nefarious conspiarcy to reassert Anglo-Australian
dominance of world cricket

but they are:

- an honest attempt to allow umpires to correct mistakes if one side
or the other thinks the umpire has got it wrong on first look, without
unduly undermining the umpire's authority or holding up what is
already a fairly stately game
- put together by people who know about umpiring with copious input
from the manufacturers of the system, including the results of the
company's internal tests

That the details of those tests and exactly what tolerances they imply
are not available to the general public does not concern me in the
slightest. I am quite happy to accept that the technicians and umpires
involved in working out what the tolerances should be are attempting
to come up with protocols which are equally fair to the bowling and
batting sides.

This is, after all, how the real world generally works: I am not a
medical expert, and I wouldn't understand the data you would present
to me about the testing you have carried out on medical diagnostic
software, yet I would be happy for my doctor to use it on me if it
were appropriate because I don't believe that you would have released
it to the world without being confident in it or that the relevant NHS
authority would have permitted its use without being pretty sure of it
themselves.

That said, I'd make some points on your details.

Don't get confused by commentators' spiel.

Imagine you are a defence lawyer with a limited number of challenges
you can make about prospective jurors. You are presented with a
potential juror wearing a green tie, and there is at least anecdotal
evidence that people who wear green ties are moderately likely to be
biased against defendants accused of property crime (as your defendant
is). Is it worth challenging this one if your defendant is from an
ethnic minority and your case is being heard in Mississippi, and there
are another twenty candidates to vet?

What the limits on challenges to the umpire's original decisions do is
say to captains: "Don't waste everyone's time just because Stuart
Broad thinks it could be out: be sure."

OK, so if it's the last five overs of an ODI innings or your last man
is at the crease and you've still got a challenge left, you might as
well take a punt on something that looks pretty hopeless because it's
remotely possible that a mistake has been made, but I think it's
grossly overstating it to call this a "weapon in a captain's arsenal".

When a commentator says, as they often do, "Clarke has wasted his
review", I can see how you might be tempted to call it "tactical". But
if the default commentator response were "Clarke has wasted everyone's
time with an obviously hopeless case, so it's a good thing they aren't
going to allow him any more", would you still be calling it
"tactical"?

Possibly you would, on the grounds that England captains seem to show
very good judgement and quite often have their appeals upheld (in the
England case, largely because Cook and Strauss have learned to ignore
Broad) while the Indian captain is a nitwit who has no idea at all,
but that's a bit like saying that it's an unfair advantage if one
captain is better at placing fields than his opposite number.

It's only a "weapon" if a captain wants to abuse the system. As things
stand with the limits on reviews, the likelihood is that attempting to
abuse the system will backfire, which is a reasonable deterrent to
abuse.

The game can usually accommodate a certain amount of abuse, but if it
gets out of hand, something has to be done. Allowing runners was a
sensible thing to do when people only asked for runners when they were
seriously injured, but the end result of people like Runnertunga
asking for a runner because he was bit tired has been the abolition of
the facility, which is a bit of a pity.

In calling the current margins of error applied "naive", you are
alleging that both the manufacturers and the umpires involved in
setting the margins are idiots, and that the considerations you
mention would come as a total surprise to them.

Now consider what the actual problem being dealt with is. Ignore the
2D graphic which is shown on a TV screen: it is a simplified
presentation of the underlying data.

I don't know exactly how Hawkeye does its calculations, but it is
pretty reasonable to guess that it does something like this:

1 Plot the 3D track of the point at the centre of the ball as
observed, and then project it further using your knowledge of physics.

2 Applying your known margins of error, plot a circle C on the plane
through the centre of the stumps showing where you predict the centre
of the ball to be and the limits of where it might be. (I presume that
it would have to be an ellipsoid in reality, but for the purposes of
this argument it's not going to be a lot different to a circle with
radius n with the centre being at the predicted point.)

Now, as you say, n will vary as ball speed v and
distance-yet-to-travel d do. For some values of d and v, Hawkeye's
maximum error m will be very small, and for others it will be a lot
higher, but there is going to be a maximum.

Hawkeye and the umpires on the protocol panel are aware of what m is
where d < 2.5 m and 40mph<v<100mph. Where d is < 2.5m, it is umpire's
call unless the centre of circle C is > 70mm from the borders of the
rectangle described by the edges of the stumps and bails. So if n <
70mm, it's umpire's call.

If the maximum value of m is 150mm, then a 70mm value for n is liable
to error. If the maximum value of m is 50mm, then a 70mm value for n
is more than sufficient to stop an on-field umpire's decision being
wrongly overturned after review.

Given that there has been tweaking of the various limits involved
during the current implementation, I see no reason to suppose that m
is now likely to be 150mm, and plenty of reason to suppose that m is
substantially < 70mm, because I think that the people who are setting
the limits aren't stupid: in particular, I don't think it likely that
the umpires on the protocol committee are going to go along with a
system which has every chance of wrongly undermining them.

The protests about "I don't know what the precise data involved in the
testing were, so I think the technology should be banned forthwith"
seem to me to be predicated on the idea that someone is malevolently
trying to force a useless technology on to the game or to introduce a
mechanism designed to allow people to manipulate results. I can easily
see that it is conceivable that either *could* happen, but I'm as
interested in that as I am in theories that the US government has been
secretly communing with beings from a planet orbiting Betelgeuse for
the last 50 years - which obviously could be true but are most
unlikely.

Nothing that anyone involved with the introduction of DRS has said has
given me any cause to believe that they are doing anything other than
attempting to make sure that the incidence of people protesting with a
fair amount of justice that star batsmen have been sawn off/given
loads of unwarranted lives is reduced to the bare practical minimum
consistent with allowing games to proceed in a reasonably timely way.
And I don't really see what kind of agenda you would need to be trying
to do something else.

>A few quick pointers:
>a) reversal only upon clear error (with definitions of what constitutes a clear error)

We'll always differ on this. You always love to come down with precise
definitions; my opinion is that the more precise you make your
definitions, the higher the likelihood that a hard case will come
along which your original draft hadn't anticipated and you end up
having to go along with a clear injustice.

Recent commentary shows that commentators have access to the
conversations between on-field and third umpire. From what they say,
the on-field umpire is asking whether the TV umpire can see any
evidence which suggests that the original decision was wrong, and the
third umpire relays what evidence he can see, and the on-field umpire
then decides whether to change his decision.

You want to lay down precise criteria for an umpire to say "OK, I'm
convinced I was wrong." How you do that without turning an umpire into
a clotheshorse is not clear to me, so you'll have to explain how you
legislate reasonableness in a quantitative fashion.

>b) clear differences in "margin of error" in reversing an out versus an not-out

Why?

>c) no batting reviews. All 10 outs automatically reviewed. 3rd umpire has 2 minutes (before other umpire takes guard)

You want to review a bowled decision when middle stump has gone
cartwheeling over the wicketkeeper's head? Why?

>d) fielding reviews to lose their appeal only if clearly wrong

This amounts to "You don't lose a review if it's umpire's call", which
certainly has a powerful logic to it. It would undoubtedly lead to
more reviews, however, and serves to weaken the presumption that
captains ought only to review when they're sure rather than when
they're merely hopeful, which would then increase the incidence of
commentators saying "he's wasted a review" - the form of words to
which you have a principled objection. I'm not sure why you propose
something which would worsen a problem you currently have.

>e) systematic review of 3rd umpires, with fines if they fail to make the correct rule / over-rule in the time allocated. Clear reports showing uniform decision making.

What does "uniform decision making" mean? That commentators and
viewers are unsure which way a decision will go, which is what you
used in favour of it currently being non-uniform, is a pretty vague
objection.

That Ian Botham, whose acquaintance with the Laws and the DRS
protocols is only slight, is puzzled by the evidence presented whereas
if an ex-first-class umpire like David Lloyd were commentating he'd
have no doubt on what the decision should be is not an argument about
umpires. You are arguing for uniformity of interpretation amongst
commentators, which is patently ridiculous.

>All this put in place today, and then rigorous testing about the margin of error in the technology to refine the "margins of error."

And your evidence that the manufacturers have not done this and that
the protocol committee haven't already been doing this is? Presumably
you have access to the transcripts of every meeting that has taken
place on the subject to back this up. if not, what is it? Is it simply
that those discussions have not been made public - in which case, what
do you think you will gain from them? Why do you think that the people
so far involved in setting the protocols and amending them in the
light of experience have been irresponsible or folowing an unhealthy
agenda (and what agenda might that be?)?

>Oh and please, can we have umpires make a decision on on-field close-to-the-ground catches and overturn them only if the camera shows it was wrong.

I'm with you in spirit on this, but I'm a little disappointed in the
inexactitude of your phraseology there.

Define "shows it was wrong" with your customary rigour, so that the
hard-and-fast criterion which you set allows for no possibility of
error in working out whether the decision is "wrong". And then make
that criterion practically implementable. Compare and contrast your
answer with what you came up with when trying to lay down criteria for
overturning lbws.

>Right now every close catch is not out if the umpire takes it up.

I'd certainly want to prevent terrible decisions like the one where
Sanath Jayasuriya was given out when England appealed to an obviously
incompetent umpire when the ball bounced at least a metre in front of
the fielder. So you'd obviously include close-to-the-ground catches in
the sort of thing people can appeal for (assuming we're allowing
batting appeals because we don't want to waste our time on automatic
reviews of stumps cartwheeling). And once you're allowing appeals,
some will be absolutely clear-cut and some will be marginal.

If we're going to avoid automatic not-outs, then we have to devise an
"umpire's call" region. You might like to consider whether setting
whatever boundary we want to set on that would be dependent on the
rate of growth of outfield grass during the course of a day, since I
can certainly see that affecting the clarity of the evidence in a
given case.

But for there to be an appeal, we're requiring the on-field umpire to
make a decision that the sides can appeal. Which leads to the
eminently sensible-sounding.....

>Except for run outs and stumped ... always make an on-field decision.

Great.

For whatever reason, it happens that at the precise moment a catch is
taken close to the ground, both umpires had their vision obscured by
hulking great fielders/bowlers.

What decision does the on-field umpire make?

Does your answer to that question encourage skullduggery-minded teams
to instruct fielders to make efforts to get in the umpires' way?

If your answer is that they should have recourse to video evidence if
available, then how do you lay down a law as to when they shouldn't
because they are supposed to be making an on-field decision? Do you
have to have video evidence of whether the umpire was unsighted to
determine whether or not he can ask for the video evidence of the
catch?

OK, let's assume that you've managed to come up with a workable
criterion.

Given that the broadcasters will show endless slo-mo replays and use
their HD magnifying camera if possible whether those are part of your
protocol or not, how do you prevent loads of people being convinced
that their team has suffered a grave injustice because those replays
end up showing that the ball did [not] bounce?

I have no idea how to answer those questions in a satisfactory manner.
If you do, I shall be very grateful.

Cheers,

Mike
--

[jzfredricks]

On Sunday, July 1, 2012 3:40:57 AM UTC+10, Phil. wrote:
> > Yeah. I used to say "as the distance between pitching and impact approaches 0, the error at the stumps approaches infinity". This confused the average reader, so I just changed it to the "100mm error" speech.
>
> It really doesn't because the of the Laws of Cricket! If the distance
> is 0 you don't need to know what the ball has done between pitching
> and impact because the Laws say that in that case the umpire must only
> consider the trajectory of the ball before impact and there's plenty
> of data for that.

Do you REALLY want to get into an argument over the definition of "approaches"?If so, I'll go grab something more powerful than this cup of tea.

[jzfredricks]

On Sunday, July 1, 2012 4:59:35 AM UTC+10, Mike Holmans wrote:
> >b) clear differences in "margin of error" in reversing an out versus an not-out
>
> Why?

Because contrary to what you've said before, the Laws contain a Benefit of Doubt, in the batsman's favour, for some types of decisions (mainly catches and LBWs, 2 of more often debated decisions)

> >c) no batting reviews. All 10 outs automatically reviewed. 3rd umpire has 2 minutes (before other umpire takes guard)
>
> You want to review a bowled decision when middle stump has gone
> cartwheeling over the wicketkeeper's head? Why?

There might have been a no ball. The rest of the decision shouldn't take more than 3 seconds to review, so why not do it?

> >d) fielding reviews to lose their appeal only if clearly wrong
>
> This amounts to "You don't lose a review if it's umpire's call", which
> certainly has a powerful logic to it. It would undoubtedly lead to
> more reviews, however, and serves to weaken the presumption that
> captains ought only to review when they're sure rather than when
> they're merely hopeful, which would then increase the incidence of
> commentators saying "he's wasted a review" - the form of words to
> which you have a principled objection. I'm not sure why you propose
> something which would worsen a problem you currently have.

But do you agree or disagree with D?

> >All this put in place today, and then rigorous testing about the margin of error in the technology to refine the "margins of error."
>
> And your evidence that the manufacturers have not done this and that
> the protocol committee haven't already been doing this is? Presumably
> you have access to the transcripts of every meeting that has taken
> place on the subject to back this up. if not, what is it? Is it simply
> that those discussions have not been made public - in which case, what
> do you think you will gain from them? Why do you think that the people
> so far involved in setting the protocols and amending them in the
> light of experience have been irresponsible or folowing an unhealthy
> agenda (and what agenda might that be?)?

Why make massive assumptions?

> Given that the broadcasters will show endless slo-mo replays and use
> their HD magnifying camera if possible whether those are part of your
> protocol or not, how do you prevent loads of people being convinced
> that their team has suffered a grave injustice because those replays
> end up showing that the ball did [not] bounce?

If in doubt, give it not out. As per the Laws.

[Andrew B]

On 30/06/2012 23:40, jzfredricks wrote:
> On Sunday, July 1, 2012 4:59:35 AM UTC+10, Mike Holmans wrote:
>>> b) clear differences in "margin of error" in reversing an out versus an not-out
>>
>> Why?
>
> Because contrary to what you've said before, the Laws contain a Benefit of Doubt, in the batsman's favour, for some types of decisions (mainly catches and LBWs, 2 of more often debated decisions)

Where?

[jzfredricks]

On Sunday, July 1, 2012 8:27:27 PM UTC+10, Andrew B. wrote:
> > Because contrary to what you've said before, the Laws contain a Benefit of Doubt, in the batsman's favour, for some types of decisions (mainly catches and LBWs, 2 of more often debated decisions)
>
> Where?

It's implied, strongly IMO.
We had a good discussion on this a few months back, I'll try to find the link.
In the Law 36 (LBW) there are about 10 sections where the umpire MUST decide if X happened. If there is doubt, then he hasn't decided that X happened, and he must give it not out.
IIRC, it's the opposite for Run Outs and stumpings. BoD goes *against* the batsman.

I believe that discussion contained the following classic from Mike; "there is no BoD in the Laws" followed by "Yes, of course there is BoD - I meant there is no *blanket* BoD". Ahhh, good times.

[Mike Holmans]

On Sat, 30 Jun 2012 15:40:49 -0700 (PDT), jzfredricks
<jzfre...@gmail.com> tapped the keyboard and brought forth:

>On Sunday, July 1, 2012 4:59:35 AM UTC+10, Mike Holmans wrote:
>> >b) clear differences in "margin of error" in reversing an out versus an not-out
>>
>> Why?
>
>Because contrary to what you've said before, the Laws contain a Benefit of Doubt, in the batsman's favour, for some types of decisions (mainly catches and LBWs, 2 of more often debated decisions)
>

No, they don't. Unless you're looking at a different
code to me, that is. The Laws of Cricket do not contain the phrase
"Benefit of Doubt".

>> >c) no batting reviews. All 10 outs automatically reviewed. 3rd umpire has 2 minutes (before other umpire takes guard)
>>
>> You want to review a bowled decision when middle stump has gone
>> cartwheeling over the wicketkeeper's head? Why?
>
>There might have been a no ball. The rest of the decision shouldn't take more than 3 seconds to review, so why not do it?

Because the game of cricket looks silly enough without inviting
further ridicule.

>
>> >d) fielding reviews to lose their appeal only if clearly wrong
>>
>> This amounts to "You don't lose a review if it's umpire's call", which
>> certainly has a powerful logic to it. It would undoubtedly lead to
>> more reviews, however, and serves to weaken the presumption that
>> captains ought only to review when they're sure rather than when
>> they're merely hopeful, which would then increase the incidence of
>> commentators saying "he's wasted a review" - the form of words to
>> which you have a principled objection. I'm not sure why you propose
>> something which would worsen a problem you currently have.
>
>But do you agree or disagree with D?

No.

>> >All this put in place today, and then rigorous testing about the margin of error in the technology to refine the "margins of error."
>>
>> And your evidence that the manufacturers have not done this and that
>> the protocol committee haven't already been doing this is? Presumably
>> you have access to the transcripts of every meeting that has taken
>> place on the subject to back this up. if not, what is it? Is it simply
>> that those discussions have not been made public - in which case, what
>> do you think you will gain from them? Why do you think that the people
>> so far involved in setting the protocols and amending them in the
>> light of experience have been irresponsible or folowing an unhealthy
>> agenda (and what agenda might that be?)?
>
>Why make massive assumptions?

Assuming that people involved in setting up the protocols would act in
a sensible way is not a "massive" assumption. It's the assumption
which normal human beings make every day when they choose to follow
manufacturers' instructions for use of their product.

You, presumably, would buy a dishwasher and assume that the
instructions for use were written by people who knew nothing about it
or how it works, and refuse to use it until you had been present at
public testing of the dishwasher in all possible circumstances,
including bizarre ones of your own devising.

>> Given that the broadcasters will show endless slo-mo replays and use
>> their HD magnifying camera if possible whether those are part of your
>> protocol or not, how do you prevent loads of people being convinced
>> that their team has suffered a grave injustice because those replays
>> end up showing that the ball did [not] bounce?
>
>If in doubt, give it not out. As per the Laws.

As before, you will no doubt be able to tell us which Laws of what
country or sport you are referring to, because it ain't the Laws of
Cricket.

And, as one has come to expect, your response is completely inadequate
to the point being made. DRS came about because it had become painful
to see the number of decisions which the broadcasters' technology were
showing to be horribly wrong. I'm asking Bharat how he would deal with
that happening again. And, as I said later, I really don't know how to
answer that question.

Now, why don't you piss off and let the grownups have an adult
discussion?

Cheers,

Mike

--

[jzfredricks]

On Sunday, July 1, 2012 9:59:02 PM UTC+10, Mike Holmans wrote:
> >Because contrary to what you've said before, the Laws contain a Benefit of Doubt, in the batsman's favour, for some types of decisions (mainly catches and LBWs, 2 of more often debated decisions)
> >
> No, they don't. Unless you're looking at a different
> code to me, that is. The Laws of Cricket do not contain the phrase
> "Benefit of Doubt".

Putting words in my mouth again? I never said the Laws, with respect to LBW and catches, "contained the phrase".

> >There might have been a no ball. The rest of the decision shouldn't take more than 3 seconds to review, so why not do it?
>
> Because the game of cricket looks silly enough without inviting
> further ridicule.

Um. Fair enough.

> >Why make massive assumptions?
>
> Assuming that people involved in setting up the protocols would act in
> a sensible way is not a "massive" assumption. It's the assumption
> which normal human beings make every day when they choose to follow
> manufacturers' instructions for use of their product.
>
> You, presumably, would buy a dishwasher and assume that the
> instructions for use were written by people who knew nothing about it
> or how it works, and refuse to use it until you had been present at
> public testing of the dishwasher in all possible circumstances,
> including bizarre ones of your own devising.

Actually, I assume that either regulations or market forces will sort things out.
Make a crap product, and you'll either get sued out of existance or go bankrupt.
That's if there's competition. If you're protected by those in power, then I guess it's a lot easier to get away with crapness.

> >If in doubt, give it not out. As per the Laws.
>
> As before, you will no doubt be able to tell us which Laws of what
> country or sport you are referring to, because it ain't the Laws of
> Cricket.
>
> And, as one has come to expect, your response is completely inadequate
> to the point being made. DRS came about because it had become painful
> to see the number of decisions which the broadcasters' technology were
> showing to be horribly wrong.

Funny enough, you're quoting ME now.

[Mike Holmans]

On Sun, 1 Jul 2012 05:09:49 -0700 (PDT), jzfredricks

<jzfre...@gmail.com> tapped the keyboard and brought forth:

>On Sunday, July 1, 2012 9:59:02 PM UTC+10, Mike Holmans wrote:

>> >Why make massive assumptions?
>>
>> Assuming that people involved in setting up the protocols would act in
>> a sensible way is not a "massive" assumption. It's the assumption
>> which normal human beings make every day when they choose to follow
>> manufacturers' instructions for use of their product.

>If you're protected by those in power, then I guess it's a lot easier to get away with crapness.

What on earth does this mean in this context?

You have a protocol which has been drafted with the participation of
umpires with the specific aim of providing a protocol which will
assist them in making the right decisions and not make them look
ridiculous. The operation of this protocol is seen by thousands or
even millions of people every time there's an international match.
Apart from some senior Indian players and Jacques Kallis, the players
(the ones the most affected) don't think there's any crapness. The
umpires are reported to think it's good. And the vast majority of
commentators and spectators think it's good.

Who is protecting whom?

Cheers,

Mike



--

[jzfredricks]

On Sunday, July 1, 2012 10:32:15 PM UTC+10, Mike Holmans wrote:
> >If you're protected by those in power, then I guess it's a lot easier to get away with crapness.
>
> What on earth does this mean in this context?
>
> You have a protocol which has been drafted with the participation of
> umpires with the specific aim of providing a protocol which will
> assist them in making the right decisions and not make them look
> ridiculous. The operation of this protocol is seen by thousands or
> even millions of people every time there's an international match.
> Apart from some senior Indian players and Jacques Kallis, the players
> (the ones the most affected) don't think there's any crapness. The
> umpires are reported to think it's good. And the vast majority of
> commentators and spectators think it's good.
>
> Who is protecting whom?

Until the ICC conducts thorough, public testing of xEye, then I feel they're protecting the xEye technology. At the very least they're shielding it from informed scrutiny.

[jzfredricks]

On Sunday, July 1, 2012 9:59:02 PM UTC+10, Mike Holmans wrote:
> >Because contrary to what you've said before, the Laws contain a Benefit of Doubt, in the batsman's favour, for some types of decisions (mainly catches and LBWs, 2 of more often debated decisions)
> >
> No, they don't. Unless you're looking at a different
> code to me, that is. The Laws of Cricket do not contain the phrase
> "Benefit of Doubt".

Let me pull my "Make Mike run away and hide" trick again. Free of charge!

Mike, if an umpire said to you;

"If there's an LBW appeal, and half of me thinks it's hitting the stumps, and the other half thinks it's missing. It so close I simply cannot make my mind up! How should I rule? What sections of the Law should I use to come to this decision?"

What would you advise him?

[Mike Holmans]

On Sat, 30 Jun 2012 02:56:05 -0700 (PDT), Bharat Rao

<rao.b...@gmail.com> tapped the keyboard and brought forth:

>d) The rules of what constitutes an over-rule on a catch are unclear. Is the absence of Hot Spot sufficient to reverse a decision? Is it clear that faint nicks won't appear?

Hot Spot 1.0 certainly did not pick up some faint nicks, particularly
on hot days. We could see balls changing their direction or see that
their revolution in the air changed, or hear sounds which had no
obvious other source even when HotSpot showed nothing. And the
manufacturer even came on to rsc and admitted that there were some
problems with it.

Hot Spot 2.0, which was used in Australia over the winter and is in
use in England this summer is said to be better, not least by the
manufacturer himself here on rsc. From my observation, it certainly
shows a clearer image with greater contrast, and I have yet to see an
incident where there is other evidence which suggests that its reading
was wrong. I expect that as time goes on, people will become a great
deal more confident in its accuracy.

Without a lot more experience of Hot Spot 2.0, though, "clear rules"
are inappropriate. The early signs are certainly encouraging, but
nobody yet knows enough to lay down a law on it. So the status quo is
that the on-field umpire and third umpire have a discussion about what
evidence is available and the on-field umpire then decides whether
there is enough evidence to convince him to change his decision.
Which, in my view, is as it should be.

I am very much in favour of clear definitions of what is out in the
Laws of Cricket. I am not in favour of precise rules as to *how* an
umpire comes to a decision on whether or not the conditions specified
in the Laws have been satisfied, mostly because I do not believe that
you can draft a set of rules which takes into account everything that
may happen on a cricket field.

It's easy enough to draft one which covers everything you can think
of, but history shows that there is always a new situation which
no-one had previously imagined, and it usually turns out that
application of the precise rules that they drafted leads to a silly
result. I'd much rather leave the decisions around the margins to the
good sense of experienced umpires.

Cheers,

Mike
--

[Mike Holmans]

On Sun, 1 Jul 2012 05:50:45 -0700 (PDT), jzfredricks

<jzfre...@gmail.com> tapped the keyboard and brought forth:

Give up umpiring. You're incompetent.

Cheers,

Mike

--

[jzfredricks]

On Sunday, July 1, 2012 11:04:34 PM UTC+10, Mike Holmans wrote:
> Give up umpiring. You're incompetent.

Any chance of a serious answer, with respect to the ruling itself?

[Mike Holmans]

On Sun, 1 Jul 2012 06:07:28 -0700 (PDT), jzfredricks

<jzfre...@gmail.com> tapped the keyboard and brought forth:

>On Sunday, July 1, 2012 11:04:34 PM UTC+10, Mike Holmans wrote:
>> Give up umpiring. You're incompetent.
>
>Any chance of a serious answer, with respect to the ruling itself?

Your job is to decide whether or not it hit the stumps. Make a
decision and rule accordingly. Or give up umpiring.

Cheers,

Mike
--

[jzfredricks]

On Sunday, July 1, 2012 11:24:44 PM UTC+10, Mike Holmans wrote:
> Your job is to decide whether or not it hit the stumps. Make a
> decision and rule accordingly. Or give up umpiring.

So if there is an element of doubt, and you can't decide that it was hitting, you must give it not out?