The offside algorithm: what the World Cup taught me about our AI future

Another one-off piece based on a conversation with Richard Russell; regular programming will be restarting soon.

Watching the World Cup, I keep seeing more than football. VAR, optical trackers, microchipped balls and semi-automated offside calls turn the pitch into a miniature of our AI future where every movement is captured, judged, archived.

That panopticon changes the game.

The old dark arts (dives, shirt pulls, little fouls behind the referee’s back) get harder, but we humans will not stop gaming the system, we will just learn to game the algorithm instead.

But deeper problems are also becoming visible; for example, the offside rule written to stop unfair goal-hanging becomes a sterile tripwire under video surveillance. A toe over the line might as well be ten yards. The letter of the rule is beginning to replace its spirit.

My 2c: machines don’t have to replace us to change us. They reshape our incentives, our behavior, our sense of fairness. It goes well beyond football. In workplaces, schools and markets, people adapt to whatever the machine measures...

More below the fold…

Like almost everyone else on Earth, I’ve spent the last few weeks glued to the World Cup. It’s quite the spectacle, and a gripping one at that. And because these days my brain is wired to think about AI, I kept watching the games through a second lens. Somewhere between the late equalizers and the VAR controversies, it occurred to me that the tournament is a real-time metaphor for what AI is doing to society.

Look closely at what happens on the pitch now, with VAR and semi-automated offside technology, and the World Cup is no longer just a sporting event. It’s a perfect, hermetically sealed example of a surveillance state backed by AI. I don’t mean “surveillance state” in the Orwellian or Zuboffian sense. I mean it descriptively. A World Cup match is a 90-plus-minute event in which every physical interaction is recorded, digitized, and handed to machine learning systems for assessment. The sensor density is hard to overstate. Beyond the broadcast cameras, optical tracking cameras mounted around the stadium log the coordinates of every player’s limbs fifty times a second, and the match ball itself carries an inertial measurement unit that transmits data 500 times per second to the video operation room.

Every scrap of human movement gets fed into a computational maw. Was that a foul or a clean tackle? An unforced error, or a clever and intentional tactical move?

The data doesn’t stop with the referee, for FIFA uses it for immediate judging and archives it for history. Team data scientists run it through proprietary models for tactical analysis. Broadcasters get it, and eventually so do third-party analysts and betting markets who want to understand the minute developments of the game mathematically. Sitting on my couch, watching a referee pause the game to listen to a voice in his earpiece, I realized that AI in the World Cup is a small part of the computationalization of human life.

The pitch as a digital twin

The footage isn’t merely analyzed by an algorithm after the fact. The full-spectrum documentation happens in real time and is stored indefinitely, which means we’re building a parallel, digital twin of the physical game.

A faithful digital representation of the event has become part of what we think the event itself is. If a human referee sees a goal but the algorithm reconstructs a 3D wireframe showing the striker’s shoulder three millimeters offside, the reconstruction wins. The digital twin becomes the undisputed ground truth.

The simulation is the reality!

The VAR controversies are early symptoms of how computation will transform this game, and I think they preview something bigger. What’s being beta-tested inside a choreographed, expensive, valuable global sporting event will reach our daily lives within a few years. The World Cup is a sandbox for thinking about how human life will unfold in the age of AI.

What does that mean for us?

The death of the dark arts

The first is that omnipresent AI makes certain long-established human behaviors obsolete.

If you know the history of the sport, you remember Maradona in the 1980s. Facing brutal defending, he started diving: play-acting, tossing himself to the turf, pretending to be catastrophically hurt to draw a foul. At the time it was an ‘innovation’ born of necessity, but today every professional player is (presumably?) trained to dive convincingly. Faking a foul has become a core competency, a dark art that exploits the human limitations of the referee.

What happens when the referee is no longer limited? Footage can be recalled from twelve angles in slow motion, and it tells you ruthlessly whether a collision was a foul or a piece of theater. You can’t hide a dive from the optical trackers.

So at some point the incentives change. If a machine will catch your dive and hand you a yellow card for simulation, you stop diving. The same goes for defenders. The shirt pulls and ankle taps away from the ball, the sneaky stuff that used to happen behind the referee’s back, are all visible for automatic judgment now. The panopticon doesn’t have a blind spot.

The incentives for how to play change, and with them the incentives for how to game the system. Humans are endlessly creative; close one loophole and we find another. If a striker can’t dive to draw a penalty, maybe he learns to tangle his legs with a defender in a way that registers mathematically as a foul, even when both players know it was an accident.

That’s the obvious first step of our AI future: people won’t stop gaming systems, they’ll switch from tricking humans to tricking algorithms. We already write resumes to get past AI screening software and title YouTube videos to please the recommendation engine. The meta-game is well underway.

Algorithmic pedantry and the spirit of the law

Then come the marginal calls, which raise a thornier question.

This tournament has seen goals disallowed because a player’s toe, a fraction of an inch of a boot, crossed a digitally drawn three-dimensional line. No human referee could catch a millimeter margin while sprinting down a grass field, and no referee was ever meant to. That kind of hyper-accuracy changes what an offside is.

The offside rule existed to stop forwards from goal hanging: loitering near the opponent’s goal, waiting for a long ball to drop on their head. The rule was about intent, about keeping the game dynamic and preventing blatant cherry-picking. Enforced with algorithmic strictness, it loses that context and becomes a strict geometric boundary. The letter of the law replaces its spirit.

So we have to ask, in football and in society: is that what we want? Systems so strict, so lacking in nuance, that an offside stops being about unfair advantage and becomes a sterile invisible tripwire? Because if that’s the new rule, defenders will adapt. They’ll step forward a fraction of an inch as the ball is kicked, deliberately dragging the striker across the algorithmic line. They’ll learn to entice opponents into mathematical violations. The game becomes a legalistic exercise in millimeter-perfect trap-setting rather than a fluid contest of athleticism.

Even in a situation as controlled and choreographed as a football match, computational technology changes the character of human behavior, and of collective behavior especially. The group’s emergent strategies mutate to accommodate the machine.

Now apply that to the workplace. An AI tracks exactly how many minutes you spend away from your keyboard, enforcing company policy to the literal second, blind to the fact that you were pacing the hallway working through the best idea you’ve had all month. When algorithms enforce society’s rules without understanding the intent behind those rules, human behavior contorts itself in bizarre ways just to stay on the right side of the line.

Reshaping, not replacement

This is also why I’m not a pessimist about our technological future. Even with the tracking cameras and the microchipped ball and the mutating tactics, what’s on the pitch is still human behavior.

Nobody is going to pay $5,000 for a stadium ticket, paint their face, and cry tears of joy or agony watching robots play each other. While today’s robots are clumsy, tomorrow Unitree or Boston Dynamics or some other tech giant will build a football superintelligence: humanoid machines that execute a perfect bicycle kick every single time, at a level of physical performance no human can approach.

Maybe a niche market of tech-fans will want machine-on-machine matches, but I don’t see it replacing the World Cup. We watch sports to see humans overcome their own humanity: the physical limits, the fatigue, the emotional volatility, the flashes of unpredictable genius.

That, I think, is the lesson for the broader world. This technological shift points toward human activity that will be irreversibly shaped by AI without being replaced by it. If we want to understand what AI will do to society, the science-fiction doomsday scenarios (terminators, all human labor rendered obsolete) are there in the background, and it’s good some people are worrying about those possibilities, but what is certain are shifts in human incentives and the concomitant behavior change.

We’re social, empathetic creatures. We value things because other humans are involved in them.

The likelier outcome of the AI revolution is what we’re watching on the pitch right now: an uncomfortable reshaping of our incentives, rules, and everyday behavior as we learn to live alongside this technology. Nobody is coming to take the pitch away from us, but we’ll be playing a far more scrutinized version of the beautiful game.