Ashes 2013
HotSpot, Snicko and Ball Tracking
October 9th
In the
interregnum between the two Ashes series that have not been too many hot topics
worthy of blogging. However, today an issue has come up that really is
interesting as it is one in which the average fan has generally little knowledge of
the underlying science and technology.
Today, Channel
9, the broadcaster in Australia, announced that it was removing HotSpot from
the technology available to umpires for use in the Decision Review System. The
stated reason is cost, but it is not an unexpected decision because HotSpot
became controversial after a series of decisions in the summer series. Some of
the disputed decisions were due to deficient use of the system, in some HotSpot
may have proved unable to detect fine edges, in others there is a suggestion
that batsmen may have deliberately set out to neutralise HotSpot. Yesterday
Kevin Pietersen was awarded substantial damages for libel after an
advertisement suggested that he had used silicone tape on the edge of his bat
to supress the HotSpot signal (these damages were donated to charity).
Some fans
are expressing indignation that HotSpot, originally sold as “based on science”
should be shown to be so unreliable. What is the problem with what appeared to
be proven technology and can we actually make it “more accurate” (whatever that
means)?
HotSpot
works on a sound physical principal that is very simple to understand and,
there lies the rub. It is just a thermal imaging camera that detects small
temperature variations on the surface of the bat. When the ball touches the
edge of the bat friction heats both bat and ball slightly. HotSpot detects this
warmth as a bright spot on the image. It is totally infallible, in theory.
Snicko uses
the same principal. An impact between bat and ball sets both vibrating,
producing sound waves. When the ball takes the edge of the bat, friction
produces a vibration that causes a sound that a sensitive stump microphone can
detect. An impact between two hard surfaces gives a characteristic, sharp
signal, quite unlike the muffled sound when ball hits pad, or the whooshing of
a ball whistling past the edge without contact. Again, totally infallible, in theory.
Snicko’s
main problem and one that was seen more than once in the Ashes series is that
it is extremely difficult to synchronise sound and vision, leading to clear
edges being detected when the ball was half a metre past the bat and the bat
was nowhere near pad or pitch. In the movie business this synchronisation is
done with the famous clapper board that gives a visual image and a clear sound
that allow the two to be synchronised with no fuss and bother: Snicko relies on
a technician laboriously synchronising two, totally independent feeds, without
the clear clapperboard signal to aid him.
Anyone with
minimal notions of science – and cricketers, in the main, tend to be a bright
bunch, with a high proportion of university-educated players, at least in the
United Kingdom – can see how both Snicko and HotSpot can be largely neutralised with a
bit of thought. In fact, for at least two seasons it has been well known on the
County circuit how to do it. Both systems rely on friction so, if you reduce
friction, you reduce the tell-tale signal of a ball taking the edge of the bat.
Rumours were
certainly circulating a while back that a good dose of Vaseline on the outside
edge of the bat would protect the batsman from very fine edges showing up. Not
only would it stop the friction of ball on bat warming the edge of the bat and
giving a HotSpot signal, it would also reduce the Snicko signal. However, it
does not need to be Vaseline: a well-oiled bat would also be effective (who
would think twice about a batsman applying a generous dose of linseed oil on
the bat before going out to the middle?) Or, this summer, silicone tape has
been the method that it is suggested is being used. It should also said that
there is nothing in the laws that stop a batsman applying tape – only the maximum
thickness of the covering of the bat and the fact that it should not damage the
ball are covered – and batsmen have taped bats innocently as long as cricket
has been played; where do you draw the line?
However,
HotSpot has two other vulnerabilities that Snicko does not have. As it relies
on a warming of the bat, its sensitivity it impacted when conditions are
hotter. That is not something that a batsman can influence. The other, most
certainly is. HotSpot depends on a localised heat signature and that can be
neutralised easily and quite legally. Here is a hint: Dennis Lillee’s famous
aluminium bat would have almost immune to HotSpot. If something in the edge of
the bat conducts heat away quickly, you can either neutralise the heat
signature or, at very least, make it disappear so fast that it would be hard to
detect. I am amazed that no batsman, or bat manufacturer, has thought of adding
fine copper wires to the covering on the outside edge of the bat. Copper is so
highly heat conducting that I would take a reasonable bet that you could make a bat
virtually proof against HotSpot detection of edges with it with minimal
difficulty and at almost no cost.
The normal
reaction to suspicions that a batsman has spoofed a bat against HotSpot is a
call for the makers to design a HotSpot that is better and less fallible. This
is completely the wrong way to approach it. You can make some improvements by
using a larger detectors, with better resolution and higher frame rates and
lower noise, you can adjust the bias on a hot day so that a faint signal is
better detected, but I seriously doubt that you can improve the sensitivity
much: detector technology has improved a long way from the CCDs that I used to
use that detected less than a third of incident light and the infrared
detectors that had painfully low sensitivity. The detectors that HotSpot uses
are direct descendants of the military detectors used to warn of ICBM launches
and are pretty much state-of-the-art.
What would
make HotSpot more sensitive is to increase
the friction from the edge of the bat. Make a stronger signal to detect.
A more
sensible approach to HotSpot is thus for the laws to regulate the covering of
the bat to make an edge more easy to detect. As most batsmen are happy to use a
bat with a protective cover, the simplest way to make HotSpot more reliable is
ensure that the covering is standard, at least on the edge of the bat and that
it is designed to maximise the HotSpot signature, by increasing friction and ensuring
that that the covering is not a good conductor or radiator of heat, so that the
signature stays present for long enough to detect clearly. This is not to say
that we should start making bats with sandpaper on the edges that will damage
the ball. The idea should be simply to ensure that HotSpot gets a fair chance.
Any additional covering, lubricants, etc. should be specifically prohibited by
the laws. A side effect of giving a better HotSpot signal is to give a better
signal for Snicko and a louder audible sound for the umpire. By having the same
high-signal covering on both edges of the bat, the chances that an LBW-saving
inside edge into the pads will be detected are also maximised.
Strangely,
the one element of DRS that a street-wise batsman cannot influence is the one
that is normally the most controversial of all. Hawk eye and its ball-tracking
competitors are proof against spoofing. The one problem that ball tracking
technology has is its application. The point of impact, be it with the pitch or
with the pad, is not an issue: when the ball hits the pitch its velocity ceases
to be downwards and changes to upwards; when it hits the pad flush on, it
ceases to be towards the stumps and changes to away from the stumps; if it hits
the pad a glancing blow the ball’s trajectory shows a clear inflection – all these
are meat and drink to ball-tracking technology which should be 100% reliable at
detecting these velocity changes if the product has been properly designed. The
problem is the predictive element and here, at least in part, the sceptics are
right.
Let’s take a
practical example. The batsman is on the front foot and is hit low on the shin by
an over-pitched ball. The point of impact may be two metres in front of the
stumps, but the ball may only have travelled 20cm after bouncing. Hawk Eye has
to predict where the ball would have gone from just 20cm of flight.
Let us suppose
that we are working at 100 frames per second and the bowler is bowling fast
medium. After being slowed by air resistance and impact with the pitch the ball
is probably travelling at 80-90km/h maximum, say 25m/s, when it hits the pad: we will
have just two frames to give us the trajectory of the ball and to extrapolate
its flight ten times as far. You can calculate the trajectory with Deep Blue
and it will still not give you a reliable answer because you simply do not have
enough flight after pitching to know with any confidence where the ball will go.
Now suppose
the opposite case: the batsman is on the back foot, 50cm from the stumps and is
hit by a ball that has pitched 5m down the wicket – the ball tracking has only
to make a tiny extrapolation and if it says that the ball will hit the stumps,
you can be as certain as anything on this Earth that it would have hit the
stumps.
The problem
with Hawk Eye is not that the technology is inherently unreliable, it is that
it is, on occasion, being asked to do things that it not capable of doing
because it does not have enough information to answer with any accuracy (think
of the computer on the Star Ship Enterprise, when it is being asked a question
for which there is no answer on available information – it will just reply
robotically “insufficient data” and not even Mr. Spock will argue with it).
If you feed
Hawk Eye a sensible set of rules, it will give sensible answers: if the batsman
is well down the pitch and the ball hits the pad just after bouncing, it cannot
give a sensible extrapolation and should tell the Third Umpire that essential
fact… NOT OUT! It is just a matter of programming sensible parameters so that
if an extrapolation is unreliable, the Third Umpire and the TV viewer are given
this information and do not ask the impossible of the system.
One thing that
could be programmed into Hawk Eye and that would make it so much easier for
officials and fans to understand is to do something that all scientists get
drummed into them at school: present the uncertainty in your result! Any
trajectory calculated by Hawk Eye will have an error that can also be
calculated which, in turn, allows you to calculate a probability that the ball
would have hit the stumps. If Hawk Eye tells you that it is 99% certain that
the ball would have hit, that should be enough for anyone. If, in contrast,
Hawk Eye says that it was just a 50-50 proposition, the batsman should, most
certainly, receive the benefit of the very considerable doubt.
As it is,
many Umpire’s Call decisions allow the batsman to stay at the wicket when the
probability that the ball would have hit was almost certainly well over 90%,
which hardly seems fair to the bowler. I particularly recall one case in the
recent Ashes where the centre of the ball would have hit under half a centimetre
outside the centre of the stump, but the call was “not out” on Umpire’s Call.
As the radius of a stump is a minimum of 1.75cm and the radius of the ball is
3.5cm, the ball has to be missing the centre of the stump by a full 5.25cm to
avoid hitting, which seemed a bit unlikely in that particular case. If the
umpire is uncertain, why not allow him to ask for assistance from Hawk Eye and
then just pass a message: “Aleem, 95% chance that that ball was hitting” and
let the umpire come to his own conclusion? Let the technology help the umpire,
not undermine him, but do it in a way that even the least
technologically-minded can see is above board.
