CHENNAI: The 'Umpire's Call' has become a dirty phrase in cricket. Even if the projected path of the ball indicates that it's hitting a part of the stumps, the umpire continues to be king. Teams either hate it and captains snigger at it. How did we reach this stage? What can be done to try and address this? What about the other gremlins in the system? Is it even foolproof? What about the fabled 5mm line, the apocryphal reason why umpires were sort of given control over technology for marginal decisions? To help navigate through all these questions, The New Indian Express spoke to Ian Taylor, the head of Virtual Eye, the technology that was in play for the India-Australia series. Excerpts:
A bit of background on Virtual Eye itself?
Both Hawk-Eye and us, the whole system has been tested by MIT. They devised a system that they came up with, which we went through testing for over a week by bowling balls under really controlled conditions. They made sure the system was working and that's kind of what everybody has gone through. It is a calibration, there are high-speed cameras in special positions that are only set up to track the ball. We are pretty happy with the result and so was MIT and they did the same with Hawk-Eye.
You spoke about placing these cameras in special positions. Do these change positions depending on venues?
What happens is that we have an optimal position, no matter what. There are four cameras placed above so that they can look down. We have got four high-speed cameras that are at roughly 40 degrees in each corner looking straight down that pitch so the only thing they see is the pitch, so nobody gets in the way of tracking. That is why we have them up high. There are different options. We went for an option that had four cameras specifically and solely dedicated to tracking and those four cameras see the entire pitch. So now they see the ball from the time it leaves the hand to the hit, in four different places, the algorithm then triangulates that within half a second to create the 3D path the ball was travelling on.
When The New Indian Express spoke to Hawk-Eye a few years earlier, they had said something about a 5mm margin of error. Does Virtual Eye have a similar margin of error?
The accuracy was set not by us, it was set by MIT. There have some claims about 5mm. Back in the very first early days, tracking was designed for enhancing television. Out of the blue, the ICC thought well, why don’t we start using it for umpiring as well. We actually said to the ICC if you are going to do that, we have to go back again and we have to develop them at a whole other safety level. Now it’s important, now it will decide the result of the match. I am sure Hawk-Eye had to do it. ICC didn’t pay any money for it at all and to upgrade to a standard, a standard we were happy with... we spent $1mn and that’s what we were using. A few years down the track, the ICC decided that they needed to set a standard and measure the standard to make sure it was being met. And so they set a standard, I still don’t know what the standard was. MIT was given that standard. Just like any kind of verification, if you are gonna qualify, you have to meet the standard that has been set. What has always concerned me is technology is always technology. Cricket is the only sport where not only you have to track the ball, but predict where it’s going. We do all sorts of sports, that’s the only sport where we have to predict. So somewhere, there is always a chance something might happen in the condition that actually isn't ideal.
On a number of grounds... this isn’t complaining but to put in context... if you want 100% accuracy, then ICC would say 'the cameras go where the Hawk-Eye and Virtual Eye tell you they must go'. A lot of times you can’t put them there. They won’t let you put them there, they lock someone’s view out there or there's a corporate box or there's a sponsor's box. We had a problem I think in 2019. In one of the grounds, the camera position was in the wrong place, we asked to move it and couldn’t move. That’s one of the first instances we had where we lost the tracking right at the start. We met with everybody and talked about it and the camera was moved to the correct place and it’s been fine ever since. It’s really important that cameras are in the right place.
You speak about tracking 560 balls a day, what are the chances of Virtual Eye showing in excess of the accepted percentage per day? Three balls? Four balls?
Whatever we say, people say we are making it up, but we are really confident. We are tracking 560 balls every day, and you can play them back and you can play them forward. What we see immediately is how the pitch is playing. You might see an over where four balls are bowled, they hit around the same spot, the batsmen lets them go, and you actually see where they are going. Then 5th and 6th balls are bowled at the same spot, they are stopped, but you do your prediction and they are just like the four that went past.
That was one of the questions, there was one prediction that looked like it was going to hit the stumps but it went over... people only hear the comments and they go 'oh that must have been wrong'. Later in the day, we showed four balls from the bowler landing on the same pitch and in the same area and let go by the batsmen, all bouncing over the stumps, all of them.
I think you are talking about the Marnus Labuschagne incident at the MCG...
Yeah, yeah. People with their eyes watching a television set, with a camera not working at 250 frames a second and have an opinion straight away that it’s wrong... we are happy with that as well. Our door is always open to umpires, players... to anybody coming down to have a look. When we play it back, with all the evidence... watch the speed of the ball coming off the pitch. When you come down and have a look, you have all the information to make your mind up whether or not it is wrong. We are not here trying to defend the technology. What we are saying is if you have problems at the ground, commentators, skippers, players, umpires, come here, our doors are open sit here like Tim Paine did. Because Paine really went after us (2019 New Zealand series), because he reckoned we were wrong. It took us five weeks to get him to come and visit us but when he did, he spent an hour and half there and the next day he spent the whole day with his fast bowlers. I don’t think anybody knows what a million dollar worth of technology looks like down there tracking a ball.
Speaking specifically about this series, what exactly happened with the Steve Smith incident where replays showed a fourth stump...
We have these four cameras that track the ball at 230 frames a second. So, we are tracking these balls and then when we track the ball, we come to the replay. The track that we get from those four cameras is played back over the broadcast camera at the end... the only time you see is when there is LBW. So if you imagine that the ball tracking is in a 3D space, in order to bring the ball track over the real camera, we have to calibrate those two cameras on the end, they are calibrated to match the virtual world in which we are tracking the ball. What happened on this particular day, which is really unusual, those cameras are always aligned. In this particular instance, the ball got tracked, the track was perfect, and the end on camera, the television camera, sometimes it jumps and loses focus. When it does that, you need to recalibrate and everything is aligned again. In this particular instance, the ball tracked perfectly, played it back and all that worked, just before they were about to play the DRS, the camera lost focus and it went out of alignment and our guys tried to realign it, they thought they had realigned it and played it back and it wasn’t... that’s why you saw four stumps, one of those was those computer-generated stumps, because two things were not aligned. What we did know was the ball had missed those stumps anyway. In the end, the decision was right.
But what we should be talking is the whole other thing which we always argued, that we shouldn't be relying on technology. We have got this argument going on that why do you have a margin of error on the ball, why don’t you just trust the technology.
The reason for DRS, right from the outset was to make sure there were no howlers. What we have always argued is, technology is really good, it's there but these decisions have always been made by umpires. I am a big fan of umpires. That’s their job, that’s what they do. Why don’t we have specially trained umpires that sit in the truck with our people. They never go out on to the field, these guys are trained to sit there. If you sit on the truck you can see the ball tracking, you can see the snicko, you can see the hotspot, you are seeing everything at the same time.
So imagine if we have specially trained umpire, he is watching this stuff and now he sees something he goes, they gave him/her out, but actually technology is showing me something else. So he puts a finger up and says there’s going to be a DRS and so players don’t do it. The umpire rather. Now he has a look and the only question he has to ask himself is... have I seen anything in here that would change my mind if I was out on the field, that’s how you decide what a howler is. You don’t sit there and measure mm of this and mm that. Because if you do that, it’s not a howler. If it takes you longer than a minute, then its not a howler.
But the umpires should be specifically trained in the technology... dump this 5mm accuracy. We cannot say that because when you are making a prediction, let's go back to prediction, I would back ball tracking 100% through the air, where it bounced and where it hit. A prediction is a prediction, so no one ever knows because it didn’t happen. Everyone has this 5mm number. If all the conditions are perfect, then probably less than 1mm. But predictions can only be as good as the data that’s coming into us. So if for any reason... there are many things that can happen, like a cloud cuts across the sky at a certain minute and takes one camera out... there are a whole lot of things that happen. The most important one is that the prediction is made from an algorithm from the information from the 4 or 6 cameras have got after the ball has bounced. So if the ball bounces up really full and you have only got 3 positions to make your prediction with, the prediction will be different to the one where you had 15 positions. The kind of thing that they say it’s always 5mm accurate is not correct.
You mentioned a thing about filing reports on incidents to both the ICC and MIT. How many of these reports like these have you had to file in the last 4-5 years?
Probably five in five years... most of the times, we don’t even get asked (by the ICC and the umpires). If we see something where there was a mistake, or where we hear people talking about it, like that particular one where cameras got out of alignment, we saw that and we knew the camera was out of alignment. We wrote a report immediately and sent it to umpires...
Can I just say we will always take responsibility for mistakes we have made. I am not trying to cover up anything... again, if you are ever around us, you would be more than welcome to come and look at the system. You will be surprised at the amount of technology involved, but I believe that we should have a specialist umpire and that umpire should be able to make the call.