In Friday’s Formula 1 news digest after the Singapore Grand Prix, there was one line I got stuck on: the FIA admitted that sprint qualifying had been ruined by a pop-up window on a race control staffer’s computer. They did not manage to deal with it in time, and the operation that had to be performed afterward was left undone — under time pressure, it was simply forgotten.
Sprint qualifying is a short standalone session that decides the grid positions for Saturday’s sprint, a mini-race over a limited distance. This season, DRS, the rear wing that opened on the straights, was removed from the championship — and replaced by Overtake Mode: when a car is less than a second behind, it receives nearly half a megajoule of extra energy per lap and can deploy it at a higher speed, up to 355 km/h rather than 345. Race control activates the mode manually, and that manual step once became the most expensive action of the weekend.
This is what caught me: in 2026, a modal window — a window that hijacks your attention, demands an answer, and hangs over everything else — can still stop a system where every second is measured in tens of thousands of dollars.
It all happened in the first part of sprint qualifying. Cadillac’s Sergio Pérez crashed, a red flag came out, the cars returned to racing — and Overtake Mode did not come on. Messages reading “overtake enabled” and “overtake disabled” flashed across the timing screens, but after the restart the mode remained off. The timing system also malfunctioned, delaying the start of the second segment, though the FIA stressed that the two failures were unrelated.
The official statement gave the cause as “human error in race control.” Nikolas Tombazis, FIA Director of Single-Seater Racing, gave Sky Germany the details. “Usually, re-enabling is a manual process, because we want to make sure it is safe to activate the mode. A pop-up window appeared on somebody’s computer, and they didn’t manage to respond to it in time. Then, under time pressure and other pressure, it was simply forgotten.”
When the problem was noticed, the mode was switched on for 27 seconds, then switched off again, on the grounds that this would be fairer. It did the opposite: that burst of energy went only to the cars that happened to be at the right point on track at the time. Half the field gained half a second per lap, half did not; Carlos Sainz put it more precisely: “Half the cars did a lap with an engine that gave them half a second per lap more than the others.”
There was nothing to fix it with: Williams asked the stewards to put all 22 cars through to the second segment, giving everyone a clean slate — refused, they had no regulatory authority to do so; restarting the first segment was impossible anyway, there would not have been enough sets of tires. Aston Martin kept a protest option open, but the result stood. Tombazis called what happened almost the worst-case scenario for the FIA outright: the governing body’s top priorities are safety and fairness, and here the result had been compromised.
The context makes it worse. A week earlier, at the Malaysian Grand Prix, a software failure shut down the hybrid engines on the formation lap, and the restart was delayed by 48 minutes — after a software update had been uploaded to every car at once. F1 has spent several weeks living in a world where software glitches and forgotten manual procedures mandated by the rules are standing in the same queue of risks.
Race control is a sealed, windowless room staffed by more than thirty people: three rows of desks, with local specialists from the FIA’s US division at the front. Blue tables crammed with communications equipment travel to every race alongside the scales and the safety car, so that the systems are identical everywhere — Miami, Singapore, Las Vegas.
Inside, routine is assembled from sharply defined roles. The video analyst finds replays and onboard footage at the race director’s request. The messaging-system operator runs the third page of the timing screens — the one where teams receive official stewards’ messages. One person keeps the ERS status of every car on screen: if a hybrid system dies, the light on the car may go out, and until then someone has to know its last known state.
When there is a serious impact, the system automatically throws up an alert with the force of the hit and the location on track. All around it — no fewer than twenty-five surveillance cameras on a shared display.
This room rests on the Racewatch system, which the FIA has been building for around fifteen years with Australia’s Catapult. Some sixty feeds from track cameras, plus onboard cameras, become a file for each incident: video, audio, and contextual data are assembled automatically. Incidents can be initiated from race control, by teams, from the remote operations center, or by the stewards.
The system has a documented pedigree of mistakes. The FIA once communicated with teams by email, until then-Ferrari boss Stefano Domenicali received an email about wet-weather tire types before the start of a wet race — since then, the governing body has had its own messaging system. After Abu Dhabi 2021, where lapped cars behind the safety car were counted incorrectly, a safety-car queue widget appeared: laps, gaps, cars peeling off and cars in the pits are gathered automatically. After the start sensor at Saudi Arabia 2024 failed to see Lando Norris start, leaving no basis on which to penalize him, additional cameras were installed on the grid, and a folder with ready-made clips began assembling itself within 40 seconds of the lights going out — and the regulations were amended immediately.
Chris Bentley, who is responsible for information systems in FIA single-seater racing, describes the philosophy plainly: “If there’s a problem, we learn from it to make sure it doesn’t happen again.” His colleague on the same team draws a parallel with aviation: the system must give the pilot exactly as much information as they can absorb, not a millimeter more. The manual step after a red flag is a deliberate choice: before activating the mode, someone must make sure it is safe. But the system has no way of knowing whether that step was carried out.
In September 1997, the Smart Ship missile cruiser USS Yorktown was left dead in the water off the coast of Virginia for roughly two and a half hours. A sailor calibrating a fuel valve entered zero into the database of the ship’s SMCS monitoring system — and the zero crashed the remote terminals, taking the engineering local-area network down with them. “For two and a half hours, the ship was what we call ‘dead in the water,’” Commander John Siley recalled. The repair took a couple of hours and came down to a single sentence from the crew: “We just had to change that zero to any other number — and everything would come back up.”
First came the search for someone to blame. WIRED’s headline read like a verdict: “Sunk by Windows NT.” Several months later, the Navy shifted the blame from the operating system to the human being. Both were right, and both missed the point: the system calmly accepted an impossible value and had no way back except through a sailor who knew the trick.
A modal window works by the same logic, only instead of a zero there is an innocent notification. It captures input, hangs over the workspace, and demands a decision: answer it or close it. Everything beneath it keeps running, but the operator’s task has changed: they are now occupied with the window. If a crucial system operation is manual, any distraction means a missed step, and a missed step in qualifying means half a second per lap.
I was ready to find an explicit ban in the regulations: surely pop-ups had been prohibited during critical phases for ages. There is no ban — and that turned out to be the most substantial finding of all.
NASA-STD-3001, a body of interface requirements for crewed systems, puts it differently. “Pop-up menus shall not obscure critical information or information required to perform a task.” Pop-up messages must be movable or dismissible by the crew — or dismissed automatically. Every command must receive on-screen confirmation that the system accepted it.
Irreversible actions require a confirmation dialog, while activating critical controls requires at least two independent actions. Alerts are arranged by priority, with a persistent queue of unacknowledged events, and flashing is reserved only for what requires immediate attention.
The FAA’s advisory circular on flight-deck interfaces is even more candid: pop-up windows are explicitly listed there as a legitimate control element, with one requirement — any feedback from a graphical interaction must be “obvious and unambiguous.” The document does not contain the word “modal,” and the word “dialog” does not appear at all.
So industry does not ban pop-ups — it bans them from obscuring what is needed and requires every input to receive an explicit response. That is stricter than an outright ban, because it removes responsibility from human memory. The truth is that in Singapore, the pop-up window apparently did not obscure anything at all: its crime was simply appearing. The window remains unnamed — neither the FIA statement nor Tombazis’s comments say whether it was an operating-system notification, a corporate software window, or something else entirely.
The cost of interruption is not measured in lost time alone. In a CHI-2008 study, people performed interrupted work faster and no less well — but paid for it with greater stress, frustration, a sense of time pressure, and effort. “Interrupted work may be performed faster, but at a cost,” the authors concluded.
That cost showed itself most clearly not in offices, but in intensive care units. Over 31 days of observing 461 intensive-care patients, monitors produced 2,558,760 alarms: 187 audible alerts per bed per day, while 88.8% of classified arrhythmia alarms were false, and 93% of true ventricular tachycardia events did not last long enough to require intervention. After a high-profile death in 2010, nurses explained that they simply did not remember the bradycardia alarms; the regulator called it alarm fatigue, and ever since, the issue has remained near the top of medical-technology hazards.
The window in race control belongs to the same class: the more often notifications are dismissed reflexively, the more likely the manual step that follows will drop out of memory. An operator who sees hundreds of service windows over a shift develops one workable reflex — close it. A modal window teaches its own insignificance, and any critical step manually placed after it inherits that insignificance in full.
The FIA already has a working recipe, and applies it to itself: turn every mistake into a widget. Incorrectly counting lapped cars behind the safety car — a queue widget. A start missed by the start sensor — cameras and an automatic folder, followed by a regulatory amendment. The same logic applies to the Singapore window: the state of Overtake Mode after a red flag should be hanging in plain sight as a status line: something that, under the rules, ought to be on is currently off. Two independent confirmations and a clear system response to an accepted command — exactly the requirements NASA wrote into its standard decades ago.
Tombazis himself promised a review “to make sure it doesn’t happen again,” and in F1 that is not an empty phrase — after every one of the incidents listed above, the system really did change.
The answer to the question raised by the hook is dry and short: a pop-up window can still stop a critical system as long as it is followed by a person who has not been distracted. Removing a single window from the interfaces is pointless — what matters is the step that is nowhere reflected in the system’s state. The messages “overtake enabled” and “overtake disabled” flashed on the timing screens during that very session, but as ordinary text in the general feed; they were not a task. The window is not guilty of anything: it simply appeared. The guilty party is the chain in which, after it, nobody looked at the status line.