Yesterday there was a brief bit of Twitter drama over a hockey analyst who deleted an archived post because it included a prediction about the surefire future failure of a particular goaltender, one that later proved to be false.
Before we get into this discussion in a soccer analytics context, anyone who lends their by-line to a published opinion about sports runs the risk of being wrong from time to time (or a lot if they’re not very good). There are a number of ways a writer can address their past mistakes. Here are four:
1. Ingore them. Simply pretend you never wrote the article in question that asserted Andre Villas-Boas career at Spurs would end in tears before the season’s finish as the club faces relegation. If you write for an old-timey newspaper hidden behind a pay-wall, this is much easier to do. If your work appears on-line, you’ll just have to tough it out for a while as commenters call into question your authority to speak on any sporting subject because, you know, you were wrong once.
2. Admit you made a mistake in your prediction, but reverse engineer several mitigating circumstances that determined the eventual outcome. Perhaps in your original article on AVB, you assumed the club wouldn’t have had the transfer budget it in August, or that a particular player would stay instead of leave. You’ll take the hit perhaps for not taking these factors into consideration in the first place, but you can at least salvage the illusion that you’re somehow never flat out wrong on occasion.
3. Delete previous instances in which you were wrong. In the digital age, why bother with accountability? Chances are no one will notice if you erase that pesky AVB prediction from a few months back. Your precious “expertise” remains intact, at least until someone catches you in the act, in which all of your credibility goes down the toilet, possibly forever.
4. Embrace your mistakes. Look to see where your prediction failed, and publicly discuss possible reasons why. Don’t be afraid to pose questions rather than pound out an entirely new set of answers. Use your prediction error as an opportunity to further refine your approach, to learn something new about the sport about which you thought you knew everything.
These options, you’ll note, aren’t simply available to all sports writers across the board. Some analysts, for example, write in such a way as to make option four impossible. This is the writer who stakes their entire career on being a capital ‘E’ expert. They work to convince their readers that they are where they are (and are paid what they’re paid) because their opinions on sports are invaluable, better than the myriad opinions of others. They’ll argue they either have a level of education or experience in the game that sets them apart.
The sportswriter-as-infallible-expert model worked well in the age of newspapers, a time when there was little accountability for the sports opinion writer beyond the angry glare of an editor or editors, in addition to tremendous pressure to shore up bona fides to the public to answer the implied question that dogs all staff writers in a competitive field: “Why is this guy and not someone else?”
The newspaper format didn’t help either. Daily single edition publications received through paid subscriptions aren’t exactly amenable to expressions of doubt or uncertainty, let alone regular follow-ups on a single topic, collaborative projects with several writers, or long-winded discussions within a frequently updated comments section.
In the age of papers, most readers would likely forget whatever a columnist predicted about something by the time the results came in to hold them accountable. There were no easily searchable newspaper archives, let alone a section in which a reader could gripe without having to write a letter in the hopes a kind editor would publish the dissent a day or days after the fact. While some writers established their writerly authoriteh with gorgeous prose, others shored up their credibility with the strength of their convictions.
The writing style of sports “expertism” has certainly carried over to the digital age, but today the writer can no longer post what they like and then hide behind the walls of their publication. Writers are now hounded continuously on comment pages or on Twitter by readers looking for ways to show them just how wrong they are. Often these confrontations devolve into name-calling or references to comparative numbers of ‘followers,’ but increasingly a fair number of sports columnists regularly engage in lively conversation with readers and other writers, which certainly isn’t a bad thing.
It’s about here where readers of this column are wondering where the bit about analytics comes in.
One the major criticisms of soccer analytics right now is the sense that analysts believe they know something the rest of the football-loving public doesn’t. Not only that, but this knowledge sets them high above the great unwashed football fan. In other words, they’re simply a newer brand of sports expert whose grasp of statistical science provides a shield from criticism.
Andi Thomas spoke to this a little in his lip-smackingly good review of Chris Anderson & David Sally’s new book, The Numbers Game:
From the ordinary fan point of view, the wider question is more-or-less moot: if you like systems, and analysis, and figuring out how things works, then this book will fit neatly into the burgeoning library of online and offline writings, and you’ll enjoy it. If you don’t, then you can safely ignore it without missing too much. As illuminating as much of this is, anybody nursing the idea that greater acceptance of analytics into the mainstream will put an end to people saying and believing incorrect things about football is being naive. Perhaps a few cliches will die a death, perhaps one or two columnists will set aside some established truths, but the broad sweep of footballing chitter-chat will retain its fundamental character, and be defined by plenty more than just the vitally-important-yet-terribly-reductive question of who is and isn’t any good.
I haven’t yet read the book, so I can’t speak to Thomas’ specific criticisms. But I think his point about tempering some of the more evangelical fire of analytics-thumpers (like yours truly) is a good one, if perhaps a little misguided.
I’m not sure that most analysts, amateur or otherwise, care as much about receiving widespread public ‘acceptance’ from those who otherwise don’t give a toss as they do in countering a few angry critics who claim, repeatedly and with no convincing evidence, that soccer analytics is simply a waste of time.
There are of course several hills on which soccer analysts would readily die, particularly those that involve warning against extrapolating general claims from single, ninety minute games. Thomas of course right in asserting that these dissenting voices on conventional football punditry aren’t limited to the PDO set. There is an entire cottage industry of angry young men and women who continue to shake their fist at lazy cliches and from-on-high footballing opinions shored up with grandiose self-regard (Joe Kinnear) and abusive shouting (Joe Kinnear). They don’t need a spreadsheet to know that Lawro’s predictions are wrong all the time.
The problem is this set has been ranting and railing against the BBC and Lawro and all the rest for decades now, stretching all the way back to the first issue of When Saturday Comes and even before that, to no avail. Football discourse has been locked in thesis/anti-thesis with no end in sight.
Perhaps everyone involved prefers it this way. Just as the fan will use his own eyes and ears to give his or her opinion on the general worth of Charlie Adam to anyone who will listen because it’s fun to talk about football, so too might the iconoclast football writer enjoy using a torrent of sarcasm in alternative publications/websites to demolish the tired cliches of Motson, Lineker and the rest because it’s fun to write angry.
I think the analytics community (at least the one I’m aware of) isn’t really in on the joke. They take football’s questions seriously (in so far as they think they’re worth exploring), and believe there may be answers beyond the subjective ravings of some guy in the paper. And, as with anyone looking to earnestly find an answer, they tend to begin by admitting their uncertainty, in first finding the right questions before rushing headlong into an answer.
In fact, good analytics trades in doubt. Few analysts of any worth that I know of would go to the wall for their predictive models; many of the best analysts view errors as a godsend, a means to make adjustments, learn more, collaborate more, improve. Errors are part of a process, rather than dents in credibility.
Yesterday I read Malcolm Gladwell’s review of Jeremy Alderman’s book on the extraordinary economist Albert O. Hirschman. Gladwell quotes a passage from the book on Hirschman’s mentor and friend Eugenio Colorni that nearly got me out of my bed, fist pumping to the ceiling.
Colorni believed that doubt was creative because it allowed for alternative ways to see the world, and seeing alternatives could steer people out of intractable circles and self-feeding despondency. Doubt, in fact, could motivate: freedom from ideological constraints opened up political strategies, and accepting the limits of what one could know liberated agents from their dependence on the belief that one had to know everything before acting, that conviction was a precondition for action.
Doubt, uncertainty, awareness of the limits of knowledge and acceptance failure are not the dead ends that the sports experts once stridently believed them to be, but often the seeds necessary to grow beyond the simple binary of thesis/anti-thesis and into synthesis.
To those within and without, the current analytics community has often resembled a classic internet circle-jerk. Yet a closer look shows a lot of consensus on some issues with rough dissent on others, almost always expressed publicly and without reservation. “Your sample size is much too small to reach that conclusion.” “There isn’t enough evidence to make that kind of claim.” “The shots on target data collected here isn’t very reliable,” etc. etc. In my limited experience, these criticisms lead to amendments, follow-up posts, corrections, and no doubt some hurt pride. But everyone knows that the public nature of football analytics writing prevents anyone from trudging out into the public realm making absolute claims without the benefit of solid evidence. An analyst is only as good as their process, not the stridency of their claims that they’re right.
Some writers perceive the skepticism of the analytics community over whether there are such things as good finishers or scoring streaks as a dig or an insult, a smug assertion of some greater knowledge of the game bestowed by a specialized science. And indeed, many analytics writers no doubt take great pleasure from exploding the “scoreboard journalism” that has dominated soccer writing for so long. But I don’t know of any who would claim that knowledge of analytics is a precursor to enjoyment of football (as daft a claim as anyone could make). Analytics writers are just as harsh on each other or themselves as they are on the self-proclaimed experts. They just don’t think that accepting lazy-half truths about soccer just because it’s a game and all in good fun is an adequate reason not to peel back its layers, to learn what makes it tick, to find new ways to enjoy the beautiful simplicity of the simplest game.
