Re: Review system: what's the point?



"eusebius" <eusebius12@xxxxxxxxx> wrote in message news:2a6b4fd2-f748-4fac-b90e-7b47d340d014@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx
On Dec 9, 2:59 am, Nirvanam <viz.sha...@xxxxxxxxx> wrote:
On Dec 8, 8:34 am, "Andrew Dunford" <adunf...@xxxxxxxxxxx> wrote:





> "Nirvanam" <viz.sha...@xxxxxxxxx> wrote in message

>news:86cb9485-fd6e-4223-954d-7c31b3e7466a@xxxxxxxxxxxxxxxxxxxxxxxxxxxxxx

> > I haven't seen the incident myself, but I'd think that the review
> > system is designed to operate with a null hypothesis that "on field
> > umpire's decision is correct". Now, the only way the TV umpire can
> > reject the null hypothesis is if there is enough conclusive evidence
> > that the on-field umpire's decision was wrong. The key words here are
> > 'enough' and conclusive. Conclusive can be interpreted as at least > > 95%
> > people would agree with the conclusion...in this case "bat touching
> > ball". If it is conclusive, then the next step is whether that
> > particular angle is enough to reject/fail to reject the null
> > hypothesis. In this case there would probably be only 1 meaningful
> > angle whereas in a run out or grounded catches, we would need 2 to 3
> > angles concluding consistently. Also the algorithm for decision > > making
> > would follow the sequence of first determining conclusiveness of
> > evidence, and then determining enough-ness of evidence.

> > In other words, conclusion = necessary, enough = sufficient. First > > ask
> > if it the necessary thing is ticked, then ask if the sufficiency has
> > been ticked.. Then you have enough ground to reject the null
> > hypothesis / over-turn on-field umpire's decision

> > So, did Rauf have "conclusive" evidence = no. Since the "necessary"
> > aspect itself is not ticked, there is no point in going to the
> > sufficiency aspect at all. The null hypothesis stays. Therefore,
> > Rauf's decision would be a wrong one.

> > The claim one can make is that Rauf thought it was conclusive. The
> > important thing is to separate subjective opinion from objective > > data.
> > What Rauf "thought" is immaterial now because the decision has been
> > made. The question to be asked is: Would 95% of the people who saw > > the
> > same thing also "think" that the same evidence as "conclusive". As,
> > logical/rational methods and systems stand today, if it is less than
> > 95% it makes no sense to reject the null hypothesis. (No need to get
> > pedantic about taking the alpha-risk and reducing it to less than > > 95%)

> > So, was the evidence so conclusive that 95% people would've thought
> > the same way as Rauf did?

> Don't know why you think that a certain percentage of people agreeing > has
> any relevance - umpiring decisions are not made by polling the general
> public, who in any case are not trained to interpret what they see in > the
> way that an umpire is. The only thing the third umpire need concern > himself
> with is what 'evidence' he personally believes exists.

> Andrew

The point is not about certain percentage of people agreeing, it is
about 95% of the times a particular event happens. The people thing
comes in only because this particular case relies on people making the
call. In hypothesis testing what we normally do is to see whether a
particular thing when done 100 times, is there any evidence to suggest
that 95 of those 100 times the event will happen in a particular way.
For ex - suppose you feed some inputs to a financial model and it
throws up an estimate. You then see the actual, let's say the actual
and the estimate given by the model are within an acceptable range.
Now suppose you gave inputs to the financial model 100 different
times, then you see the difference between actual and estimate for
each instance. If this difference is acceptable 95% of the time then
the model is good enough, that'll be the conclusion.

Similarly, in the umpire case, we are asking if 100 different people/
umpires see this video, will 95 of them conclusively say it is "out"?
Or put in a different way: is the evidence so conclusive that 95 out
of 100 people would also infer, like you/Rauf/whoever, that the ball
touched the bat? If they won't then there is no point in changing the
status quo, in this case on-field umpire's decision. You may decide
what that level should be based on the criticality of the item. For
ex, if you are testing a CT scanner or some medical device, you would
probably want all 100 instances to give you correct output. If the
thing you are testing is the process of ball manufacturing. You might
be OK if 90 balls out of a sample of 100 balls have their diameter
within an acceptable range. If you are testing the manufacturing
process of candy bars. You might be OK with even just 50 of them
having a length difference which is less than 1mm of the standard
size, maybe 90% for length difference of less than 1 cm of standard
size. That is the "risk" you take in rejecting the status quo. The
risk you are taking in our umpire's case is 5% in other words there
would not be more than 5% of the people/umpires who were asked to make
a decision with the same evidence who will then decide that it was not
out. Clearly the evidence wasn't that good that you won't find even 5%
people disagreeing (or 95% agreeing). Hence there is not enough data/
evidence to suggest that the status-quo should be changed.

Generally, most tests set 95% as a thumb rule (but depending on
criticality of the thing u are testing you may want to reduce it or
increase it).- Hide quoted text -

- Show quoted text -

Whatever, but the point remains that Rauf's overrule was indeed
correct. There was substantial deviation, there was a substantial
noise. You could say, that this was as out to the naked eye as the
famous Symonds decision, which led to such admirable restraint from
nationalistic cricket fans, players and public.

The problem was, 'hotspot' is not set up to pick anything up that
cannot be seen from a frontal view from the bowler's end umpire's
perspective. If someone had've thought of the possibility of such a
situation and set it up so it could see a side on view, then there
would have been little or no dispute.

It is easier also to pass judgement on such cases after seeing them
onesself, extensive viewing in fact is advantageous.

As there appears to be no definition of what constitutes 'conclusive evidence' that the ball hit the bat in the case of an edged catch, we're not in a very strong position to determine the correctness or otherwise of Rauf's decision, in which case he is justified in overruling on what *he* saw.

Andrew

.

Relevant Pages

  • Re: Review system: whats the point?
    ... the only way the TV umpire can ... angles concluding consistently. ... evidence, and then determining enough-ness of evidence. ... of 100 people would also infer, like you/Rauf/whoever, that the ball ...
    (rec.sport.cricket)
  • Re: Review system: whats the point?
    ... the only way the TV umpire can ... angles concluding consistently. ... evidence, and then determining enough-ness of evidence. ... of 100 people would also infer, like you/Rauf/whoever, that the ball ...
    (rec.sport.cricket)
  • Re: Review system: whats the point?
    ... the only way the TV umpire can ... angles concluding consistently. ... evidence, and then determining enough-ness of evidence. ... said it was out conjectured based on the fact that the ball deviated ...
    (rec.sport.cricket)
  • Re: Review system: whats the point?
    ... the only way the TV umpire can ... angles concluding consistently. ... evidence, and then determining enough-ness of evidence. ... said it was out conjectured based on the fact that the ball deviated ...
    (rec.sport.cricket)
  • Re: Review system: whats the point?
    ... the only way the TV umpire can ... angles concluding consistently. ... evidence, and then determining enough-ness of evidence. ... said it was out conjectured based on the fact that the ball deviated ...
    (rec.sport.cricket)