Baku Doesn't Trust Supercomputers: Antonelli, Street Circuits, and Unverifiable Predictions

On the streets of Baku, the ancient stone walls of the Icherisheher fortress sit so close to the racing line that the faintest touch is enough to end any driver's race. I remember the first time I stood at the Turn 1 braking zone, where F1 cars arrive at over 300 km/h and are forced into a nearly right-angled corner. No run-off. No escape. Only concrete, barriers, and the roar of engines echoing between the walls of a city thousands of years old. That is why I always stay behind after every race weekend in Baku. Not to wait for the result - the result is already on the timing screens. But to look at what remains: the scuff marks on the barriers, the scattered tyre marbles, and the eyes of the mechanics as they push a broken car back to the garage. A season opens with a heartbeat in the garage, before the lights even go out. The 2026 season is entering the Azerbaijan Grand Prix with a picture that is almost settled. Kimi Antonelli, Mercedes' Italian driver, leads the standings by 81 points over his teammate George Russell. That margin is worth three race wins - a gap that modern F1 history rarely sees closed in the closing stretch. Antonelli has taken eight wins this season. Russell, whom the media calls the 'closest contender,' has only two wins and seven podiums. Charles Leclerc of Ferrari sits fifth, 44 points behind Russell. Lewis Hamilton, who moved to Ferrari, ended his winless run this year but also retired in Madrid with a brake failure. Ahead of Baku, a prediction piece appeared across many platforms with a provocative headline: 'Supercomputer predicts the Azerbaijan Grand Prix result from P22 to P1.' The piece cites 'Grok' - a language model - and 'Supercomputer,' offering a detailed forecast for each driver: Antonelli wins, Verstappen third, Russell off the podium, Hamilton sixth. Baku is the 17th round of the 24-race 2026 calendar. The circuit is 6.003 km with 20 corners, including the longest straight of any street track - 2.2 km along the Caspian shore. Top speeds can exceed 350 km/h, yet a single mistake at Turn 15 - the narrowest corner - can undo an entire weekend's work. No telemetry data, no tyre analysis, no track-condition information appears in the prediction piece. They are simply numbers presented as if they had been verified. The problem lies in the presentation. A language model operates on word probabilities, not on telemetry, tyre data, or track conditions. It can produce a highly persuasive paragraph about the Baku race without knowing anything about Baku. Look at the numbers themselves. Baku has hosted eight Formula 1 races, and in those eight, six different drivers have won. It is one of the highest-variance circuits in modern F1 history. Yet the prediction presents an outcome that seems certain enough to leave almost no room for surprise. I have followed street circuits for more than three decades, from Monaco to Singapore, Long Beach to Baku. I keep the rhythm of every race not with a stopwatch, but by never missing a single session in thirty years. There is an unwritten rule: on street circuits, the performance gap between cars compresses, and the human factor - reflexes, split-second decisions, the ability to read the track - is pushed to the decisive level. On a circuit where the barrier sits one metre from the racing line, a better driver can compensate for a lesser car. Antonelli is having a remarkable season. But precisely because he leads by 81 points, predicting he will win in Baku carries no new information. It is like forecasting rain in Da Nang in October. It might be right, but nobody learns anything from it. The 81-point gap between Antonelli and Russell at Mercedes is one of the largest teammate gaps in Brackley's history. It is more than a statistic. It reflects a strategic truth: when a team has a leading driver with a safe margin, resources - from car upgrades to strategy calls - tend to tilt toward that driver. Russell is not only fighting Antonelli. He is fighting a team structure optimised for its leader. What stands out is where the prediction rates drivers lower than reality. Max Verstappen of Red Bull is placed third, despite five podiums in his last seven races, including third in Italy and second in Spain. If any driver can break Antonelli's dominance in Baku, it is Verstappen - a man who has proven he can extract the maximum from an imperfect car. But there is a variable the piece mentions without developing: Verstappen's car may not be strong enough. In his last seven races, his podiums have not come from a speed advantage but from situational handling. In Baku, with the 2.2 km straight and DRS, top speed is decisive. If Red Bull lacks straight-line speed, Verstappen may have to fight his own car before fighting his rivals. Another detail the prediction ignores: Lando Norris of McLaren, with two wins and five podiums this season, is not mentioned among the leading contenders. Neither is his teammate Oscar Piastri. Meanwhile, Pierre Gasly of Alpine and Liam Lawson of Racing Bulls are placed in the midfield - a classification reflecting the resource gap between teams. There is a trend in sports media I have observed over decades: turning predictions into weighty facts. When an outlet cites a 'supercomputer,' readers tend to believe it more than the opinion of an expert. But the nature of both is prediction. The only difference is that one wears the cloak of technology. The 'supercomputer prediction' format has become a genre in sports media. It appears before every major race, following the same formula: take recent form data, add a slight weighting for leading drivers, and present the result as if it were scientific analysis. What it actually produces is a false sense of certainty - a sedative for readers facing the unpredictable. The deeper issue is not whether the prediction is right or wrong. It is how prediction frames thought. When readers are told the result has been 'calculated' in advance, they are less inclined to notice what is actually happening on track. I was in Russia in 2026, following Vietnam's U23 team on their training camp. The Russian winter of 2026 was so cold that I remembered every breath of the stands for four years afterwards. One evening in Moscow, I sat in a press conference listening to experts analyse the team's chances. Numbers were produced. Predictions drawn. And then, in the actual match, everything unfolded in a way nobody could have scripted. Goalkeeper Bui Tien Dung, then 21, produced a performance no statistic could have predicted. That is why I never trust absolute predictions, whether they come from humans or machines. The track does not read predictions. It reflects only what happens in each moment. In Baku, there is another factor the prediction fails to mention: the likelihood of a safety car. With narrow street characteristics and heavy braking zones, the safety-car probability in Baku is higher than at most circuits on the calendar. A well-timed safety car can invert the entire picture - turning the leader into the chaser, or vice versa. It is the kind of variable no language model can handle, because it depends on human decisions made in split seconds. Isack Hadjar of Racing Bulls is another case worth noting. He returns from a wrist injury, and the prediction forecasts a dip. But in sport, a comeback from injury often generates an unpredictable psychological drive. I have seen drivers return from injury and race as if they had nothing to lose - and often they are the most dangerous. Liam Lawson is another example. He was promoted to Red Bull to cover Hadjar for three races, scored a sixth place, then returned to Racing Bulls. It was not a form-based move but a systematic arrangement between two teams within the same organisation. This operation - a driver being moved between the senior and junior teams - is a structural feature of the Red Bull system. It allows the team to assess drivers in real conditions before making long-term decisions. The Baku race takes place in a few days. Antonelli may win. Or may not. What I will be watching is not the result on the timing screen, but how the teams handle the unforeseeable: tyre strategy on a street circuit, the decision to call a safety car, and each driver's ability to adapt as the track changes lap by lap. What a supercomputer cannot calculate is the moment a driver realises his car is not good enough, and decides to push to the limit. Audiences remember the wins; I remember the eyes of the mechanics when the last car returns to the garage at night. That is the moment sport becomes interesting. And no algorithm can write it in advance.