We have looked at how the science of decision-aid technologies works and have ignored the mythology. Measurement is always subject to error, and decision-aid technologies are not always accurate; they cannot and should not be an infallible replacement for human judgment. There is no exact measurement on the sports field. Accuracy must not become a fetish.
Philosophical thinking can help in the understanding of scientific and technological practice, in this case replacing an obsession with accuracy with a desire for justice for the sports fan. Taking a philosophical look at match-officiating shows how the problem of justice has arisen out of the introduction of TV but can also show how it can be resolved by that same TV technology: broadcast TV created the problem of a mismatch between TV viewers’ epistemological privilege and match officials’ ontological authority, and broadcast TV can resolve it. Long delays associated with TV occur only when the replays are being used to try to solve an accuracy problem rather than a justice problem. The solution is to restore match officials’ ontological authority and accept RINOWN, the Right If Not Wrong principle. Match officials should continue to create what exists in the way of balls, strikes, runs, outs, and not-outs, just as they have always done in the past, unless there is some technology that is already revealing their mistakes to a wide audience. In that case, third officials can use the same technology to put the matter right. But under RINOWN, match officials’ decisions are overturned only if the mistake is quickly and glaringly obvious both to the third official and TV viewers. When no mistake is obvious we revert to the presumptive justice that has supported sports throughout history and still supports it at every contest, from a lower-league fixture to a kick-about in the street.
It is true that the decision over whether, for example, a “try” in rugby union has been scored can be interminable when it is supported by TV replays. For those who do not know the game, the decision has to be over whether downward pressure has been applied to a ball that is in contact with the ground beyond the try line. It is like American football in that the ball has to reach the end zone but in rugby it must not just cut the plane of the zone but also be firmly grounded; opposition players may legitimately try to hold the ball up or interpose their hand or bodies between the ball and the turf, and there is often a heap of scrambling bodies around the try line with the ball buried somewhere among them. Officials can take “forever” examining a series of slow-motion replays from a variety of angles, trying to find the optimum view for making what will sometimes be an impossible decision. Such delays are allowed either because rugby union isn’t concerned with keeping the game flowing or because rugby union officials do not understand that what they seek should be not accuracy but justice. If they understood this and wanted to speed things up, they would return ontological authority to the on-field referee and adopt RINOWN. They would tell on-field referees that in televised games, just like every other game of rugby that has ever been played, their duty is to make a “try/no-try” decision themselves and only refer it to the off-field referee if they have doubts. If the decision is referred, TV replays, visible to the crowd and home viewers, could rapidly show if the on-field referee was clearly wrong, and only in those circumstances would the on-field decision be reversed. It would take seconds, and every TV viewer would see that justice was not being violated. Ever since rugby was invented, referees have been deciding whether a try was fairly scored when the ball is under a heap of struggling bodies without any technological aids. Doubtless many of their decisions were, in some absolute sense, “wrong.” But nobody was in a position to make a better decision than the referee (except maybe a player or two at the bottom of the heap, but they can’t be allowed to make those kinds of decisions). So, there was no perceived injustice.
Cricket, though it is not a hurried sport, is our model sport as far as the RINOWN principle is concerned, at least when it is a matter of leg before wicket (lbw) decisions. In the lbw case, the umpire makes a decision and then, if there is a challenge, it is up to the third umpire to overturn it. If there is any doubt, the original decision stands. Cricket is a game played as a series of mini-contests with minor intervals between them, and the sometimes drawn-out decisions of the third umpire in the case of a catch, as he or she along with the crowd examines Snicko and Hot Spot, are enjoyable, and all can watch justice unfolding. In the case of run-outs and stumpings, the decision-making process could be improved if the cricket authorities wanted to speed things up. As it is, umpires tend to call immediately for a replay, whereas they could be asked to make a decision first and then only refer it if they have serious doubt. That way the much speedier RINOWN principle could be applied in the case of those decisions too.
In international tennis, there is no waste of time. If a player challenges the umpire’s call, the track estimator is called in to adjudicate just as though the umpire had made no judgment in the first place: the entire decision—the entire ontological authority—is delegated to the technology (and whoever is sitting in the private control room), and the decision is made quickly with a nice graphic presentation that the crowd can cheer. Everybody is happy—except one or two of the tennis players, who are pretty sure the technology makes mistakes, and, of course, the authors of this book and those who agree with them. We are not happy because we think that presenting virtual reality as real reality is not healthy; we think citizens should understand the world we live in. Once more, we’ve discussed a range of simple solutions. (1) Remove track estimators and bring back TV replays with umpires allowed to view them after they have made a call; once more, if the call was not obviously wrong the original decision should stand. (2) Keep track estimators, but use them, along with the RINOWN principle, to overrule the umpire only when the umpire’s call is substantially at variance with the track estimator’s call. If, according to the track estimator, the ball strikes within, say, 5 mm of the line, the umpire’s call should stand. Or, if the crowd has become addicted to the drama, if the ball lands within, say, 5 mm of the line, the track estimator reconstruction should be replaced with a graphic of a spinning coin driven by a random-number generator which will land “heads” or “tails”—“in” or “out.” If someone ever investigates the statistics properly, the coin could be appropriately biased.
Paul Hawkins, representing Hawk-Eye and writing in 2015, has claimed that the tennis error is 2.6 mm.1 This is an improvement on past claims, and such an improvement is just what we would expect as the technology gets better. Perhaps the International Tennis Federation has made more measurements and perhaps they are prepared to make them public. But an average error of 2.2 mm still allows for much larger errors, and we have no idea how these errors are distributed nor how they play out for different kinds of ball impact. To move away from Hawk-Eye in particular, it still seems to us that when a track estimator shows a ball landing close to the line in tennis—say, within the 5 mm range of error allowed by the International Tennis Federation (or maybe, nowadays a millimeter less)—and when the public sees an exact reproduction of a line, a ball, the ball’s path, and a reconstructed impact footprint with perfectly defined edges, then the public is seeing virtual reality, not real reality. Probably no one knows exactly how closely that virtual reality corresponds to real reality; unless we are missing something—and we are always ready to learn more—a ball shown as a millimeter or so out could in fact be a millimeter in, and vice versa. Viewers are seeing computer-generated graphics, of the kind they see when they watch a movie like Lord of the Rings, they are watching what is known in the field of artificial intelligence as a “microworld” (see appendix 4), but they are being allowed to think that the tennis equivalent of Orcs and Elves are real.
Another way of looking at it is akin to those TV series that come under the general title of “Crime Scene Investigation (CSI)” in which the potency of forensic science is mythologized in the way that the accuracy of decision-aid technology is mythologized. In CSI, careful and painstaking investigation will always produce definitive evidence that points without doubt to guilt or innocence. But neither real pioneering science nor real, routine, forensic science are like that—they are full of pitfalls, uncertainties, and, above all, contentious statistical inferences. To make proper use of these devices, if we really want them to make measurements, all this uncertainty needs to be presented. We say “proper” use of these devices, but we have argued that the attempt to reach ever more demanding standards of accuracy is misplaced; the better use for track estimators is in the aid of the construction of archival databases—a new and rather wonderful and rich variant of the history of sports.
The danger of this technology is false transparency: track estimators can present a kind of show-trial justice. And this is not only bad in itself but bad for society. The increasing speed and capacity of computers is making it harder and harder for us to distinguish computer-generated graphics from live video footage; computers, their technology driven by the ever more demanding computer-games industry, are reaching the point where they can generate a lifelike sequence of hue and intensity for every single pixel on a video screen. Once computers can do that in real time, then reality and artificial reality, as represented on the TV screen, will be indistinguishable to the naked eye. Once upon a time dictatorial regimes doctored photographs in the darkroom, removing a figure here, adding one there, so as to produce what looked like realistic evidence to back up their reconstructions of history; nowadays it can happen, or it soon will be able to happen, with video running at full speed in real time. No one can want track estimators to be the thin end of the wedge for the creation of a new kind of gullible public.
But why reach out for devices as complicated as track estimators when, once one understands that the problem is justice, not accuracy, and understands the principle of Right If Not Wrong, TV replays can do most of what we need. The general argument we have tried to mount is that we should always use the lowest level of technology. Consider the cup on the golf green: this could be replaced by a circle to guide the player’s aim, while multiple cameras set up around the green reconstruct the path of the ball so that a computer model can determine whether the ball would have fallen into a cup had there been one. Likewise, the bails in cricket could be removed and track estimators could be used to reconstruct path of a ball that brushed or came near to brushing the stumps and reach a conclusion as to whether it had touched or missed. And so on. But why would anyone want to do this? It is obviously much less expensive and much less trouble to work with lower levels of intermediation, and it is also more transparent because there are no unknown error-distributions and everyone can see what is going on. In general, once we eliminate the fetish of accuracy we need very good arguments to move to a higher level of technology.
The philosophical concept of justice versus accuracy and the ideas of ontological authority and epistemological privilege are the spine of the book. They are what enable us to speed up the game with the Right If Not Wrong principle. Things have changed for match officials with the introduction of television. Epistemological privilege increasingly resides with the television-viewing public who tend to have a better view of a disputed decision than the officials judging in real time. We have shown how to put things right.