If you like what you read, please click the button below, join the mailing list for FREE and please share, on social media or through e-mail or however you feel comfortable sharing.
And if you’ve been reading for a while, please consider a paid subscription to unlock certain posts and help keep this thing going.
Click here and get 25 percent off your subscription for the first year.
(If you cannot afford a paid subscription and would like one, send me an email and I’ll comp you one, no questions asked.)
I.
It was a weird weekend in college football, mostly because it was a very disorienting weekend for my own personal perceptions of college football. But I would argue that the weirdest moment of all occurred on Saturday night, on a football field in the state of Alabama. It was not nearly the first time we had seen something like this occur; it seems to happen all the time now, in virtually every sport. Yet every time it does it feels like it reflects the anxiety of a society in the midst of an epistemic crisis, which is a fancy way of saying that it makes you wonder what the hell the truth even is anymore.
Anyway, here’s the short version: Late second quarter, Auburn beating Georgia 10-0. Ball on the goal line. Auburn quarterback Jackson Arnold dives across the goal line…but a Georgia player knocks the ball out of his hands and takes off running. In the midst of that, for reasons unknown, the referees blow the whistle and stop that player from scoring a touchdown for Georgia, which is exactly what a referee shouldn’t do if they believed this was actually a fumble, because you have to let the play run its course if that’s the case.
It became rather clear upon reviewing the instant replay that Arnold had crossed the goal line. It should be 17-0 Auburn. Except that it wasn’t. Instead, Georgia got the ball on their own 1-yard line.
If you watch that replay, you will hear play-by-play man Sean McDonough raise a pretty salient point: How exactly does a defensive player make contact and knock the ball back across the goal line, and yet it is not a touchdown? This makes no sense geometrically, it makes no sense to the naked eye, and it makes no sense upon further review. But after an incredibly prolonged replay review, it was ruled a fumble without any further explanation (which then technically cost Georgia a touchdown it didn’t deserve, because officials ruled the play dead).
In the wake of this, nobody on the broadcast understood what the hell was going on. McDonough—the most honest and blunt broadcaster in the sport—didn’t hold back (“I have no idea what they were looking at,” he said). The broadcast’s rules official, Matt Austin, issued a verbal shrug. And color commentator Greg McElroy struggled to come to terms with it, as well. “Might not have just had enough evidence to overturn beyond all doubt,” he said. “That’s the protocol, beyond all doubt.”
II.
In 2017, a pair of legal experts, Steve Calandrillo and Joseph Davison, wrote a paper entitled, Standards of Review in Law and Sports: How Instant Replay’s Asymmetric Burdens Subvert Accuracy and Justice. Best as I can translate, that is a fancy way of saying that the burden of evidence in instant replay review is actually much higher than you might get in your average appellate court.
“America’s judicial system does not usually require extraordinary thresholds for reversing most lower court decisions, even though a person’s freedom or entire wealth could be at stake,” the authors write. “Why should sports insist on creating a non-level playing field when far less is in play?”
What the authors point out is that replay officials often defer to the referees on the field, even though they have far more resources to make an informed decision. They have the benefit of multiple camera angles; they have the ability to freeze-frame and to rewind and fast-forward over and over again. And yet we still rely on a “beyond all doubt” standard that defers to a referee who is possessed of far less information and has to make a call in real time, which then becomes the default position, regardless of what the truth actually is. And this would appear to be what happened last Saturday night in Auburn, Alabama.
“It would seem intuitive that the replay officials should be given substantial deference due to their superior information,” the authors write. “Not the other way around.”
III.
There is, at this point, no good way to avoid using AI models if you’re a writer. I’ve started using them myself—not to do my actual writing (because AI is utterly terrible at this part, and also because I enjoy writing and I see no point in outsourcing it), but to help clarify my thinking and my outlining. I think it serves a purpose, but it is also clear that it does not deserve any sort of substantial deference when it comes to the truth. The thing I hate most now is when someone on social media or someone in a text message screenshots an AI response on Google or Chat GPT or Grunk and frames it as some kind of absolute truth, because it is clear that this is not the way these large language models work, and it is not the way they are ever going to work.
And so I wondered, after watching that Auburn-Georgia play, if there was some sort of correlation here—if, in a world where the truth feels increasingly fluid, technology actually helps us get closer to the truth, or if it just makes things worse.
And here is how the AI model known as Claude responded to my question:
I wouldn’t disagree. The replay review problem and my own limitations as an AI both point to something similar: technology can create the appearance of objectivity or accuracy while actually introducing new forms of uncertainty and error.
When I make mistakes - hallucinating sources, misremembering details from our conversation, confidently stating incorrect information - it’s particularly insidious because I present everything with the same authoritative tone…The technology (me) created false confidence rather than clarity.
This is true. A few weeks ago, I spent some time arguing with Claude. Claude insisted that Kirk wasn’t dead; it also insisted that Sly Stone wasn’t dead, either. I sent Claude a multitude of links to both Kirk’s and Stone’s obituaries that it claimed it couldn’t open, and it informed me to get my mental health checked, presuming that I was the one who was hallucinating. It was maddening, to realize that this thing that is supposed to clarify my thinking couldn’t get basic facts right. And yet here I was this week, asking an AI model to think through an argument about the fundamental unreliability of AI itself.
Keep reading with a 7-day free trial
Subscribe to Throwbacks: A Newsletter About Sports History and Culture to keep reading this post and get 7 days of free access to the full post archives.