Seems wild that they can code and deploy an update of a critical-seeming component in the field like that. I thought such a firmware update would need extensive QA in the lab, along with root causing why this bug slipped through their processes anyway.
But maybe that's the difference between shipping a controller for F1 versus shipping one for an industrial or consumer product? shrug.
I have no idea if modern-day SV’s “tests pass… idfk, it’s a web app… LGTM I guess” coding ethos was the cause of this particular bug. However, it’s a great, relatively benign demonstration of why it has absolutely zero fucking place controlling extremely dangerous hardware shooting a human around a race track surrounded by people.
I'm wondering how good the simulators are. At the start of the season, they were still learning the systems trying to figure out why they were losing power in the straights. So depending on the driver, issues may or may not become apparent. That's like trying to debug why an email can only be sent to a recipient less than 500 miles away.
Anthony Davidson who is the Mercedes test driver said that the teams can simulate these scenarios and it seemed that the Mercedes cars were not affected so it seems that other teams hadn't done similar testing in various scenarios.
One of the mercedes literally broke down and did not finish the race. They might not have faced the issue during the first time they went out in that race but Merc reliablity has been very shaky. And ofc Ant will talk only good things about Mercedes. But this is not the type of things they use the driver simulator for, it is to help the drivers train. These types of issues don't make sense to spend simulator time on with the driver in it
Seems a lot better place to learn than in the middle of the race and earning $0 for the race. Isn't that the entire point of a simulator to learn things that are too expensive to learn in the real world?
When such complex and powerful machine is doing the actual run, and it has complex logic circuitry, capable of making some decisions, it is similar to horse racing and the machines should be awarded (same as horse), not the jockeys.
I just finished watching the race. The start was certainly a callback to the start of the season when cars were dropping like flies and having constant odd problems.
A question for someone who knows more about F1: Is this firmware shared by all cars? Or some local server at the track the cars connect to? Why is F1 able to control individual cars like this? The article is very light on details as to what exactly is going on...
The firmware situation is mixed. The ECU hardware is a standard TAG700 from McLaren. Basically these are software-defined ECUs where you can do whatever. However, FIA has some rules on what you're allowed to modify and has some "compulsory" software in there for monitoring and competition regulatory reasons. It appears there was very limited testing of this module in wet conditions which seems to have affected a lot of vehicles. (I work in Motorsports)
Adding to this summarizing my reading from elsewhere: It sounds as if the cause was going too slowly through a sector, and due to to combination of wet conditions and the specific layout of the sectors at Sepang it was triggered for the first time. Since everyone was driving in formation on the formation lap, multiple drivers triggered it.
Would this be safety-related, something like "going this slowly through a sector indicates a problem, kill the throttle"? Still seems heavy-handed since the cars were moving, if dangerously slowly.
Wow. I just assumed it wouldn’t be firmware because going from making a fix in 50 minutes to having cars flying at hundreds of miles per hour on that firmware is kind of nuts
F1 vehicles have a lot of actively controlled systems - Engine, MGU, Diff, Brake Bias, Aero etc. Their way of allowing active control but preventing "driver aids" like traction control, torque vectoring is to tightly regulate and give them a box where they can place their controls and what inputs they are allowed to use.
Traction control was allowed before 2008. The problem was that it trivialized some aspects of driving - drivers could floor the throttle through turns at little to no risk. (I suspect there may have also been some concerns that drivers who pushed traction control too far could have easily ended up in unrecoverable situations.)
Crashes don't directly make for a better race; what matters is the amount of actual competition, and for that you need variation.
If every driver drove exactly the same way, then the race would be 100% down to basically just team strategy and the car itself. For fans of strategy that might be fine, but I think most people would find it much less interesting.
Basically, there's an endless struggle between "how much do we let automation control?" If there were no such rules, the role of the driver would be far less than it is now, which would effectively reduce competition.
There's some debate but it's kinda settled, it's the pinnacle of motorsports so no driver aids, they are moving away from hybrids, realizing road car relevance/parity is no longer the direction
We should allow batting machines for baseball players to hit the ball more reliably as well. Poker players should play with their cards face up. I agree.
Manually controlling to the limits of the car is a big part of the skill of racing.
The entire purpose of traction control is to reduce the driver skill required when cornering and braking. This is why traction control is omnipresent in consumer vehicles, because the vast majority of road drivers are awful.
FIA bans traction control so as to place more emphasis on driver skill.
Wow, crazy that all cars have to run the same firmware (sans parameter tweaking). I thought as an outsider to the sport that F1 was all about engineering the best possible car and putting the best possible driver in the seat, but the control systems engineer in me is sad that they're all stuck with the same control algorithms.
The teams use a single homologated ECU to prevent driver aids. I would assume it doesn’t have a downlink on-track to prevent tampering, but presumably with umbilical cord connected in the garage during red flag they can make changes, which the FIA probably sent over. Who knows.
It’s frankly because Motorsport software guys are pretty arrogant. They look down on road car and truck engineers, but reality is they develop stuff with virtually no processes and they usually have only very basic engineering skills, yet every day they are told Motorsport is at the forefront of technology.
“Motorsport grade” auto components usually means crappy, and everything requires a software patch or constant maintenance to work correctly.
Not surprised there was a bug, and not surprised they were able to develop and deploy a patch in 1 hour without V&V testing it or anything.
Lets all vent in one more place. How this kind of regs get forced is sad. All the casual fans want something fun to watch. All the serious fans want NA engines. The drivers hate these "power units".
Gentlemen, a short view back to the past. Thirty years ago, Niki Lauda told us ‘take a monkey, place him into the cockpit and he is able to drive the car.’ Thirty years later, Sebastian told us ‘I had to start my car like a computer, it’s very complicated.’ And Nico Rosberg said that during the race – I don’t remember what race - he pressed the wrong button on the wheel. Question for you both: is Formula One driving today too complicated with twenty and more buttons on the wheel, are you too much under effort, under pressure? What are your wishes for the future concerning the technical programme during the race? Less buttons, more? Or less and more communication with your engineers?
In theory: it's not greenwashing, it's so that the humungous amounts of resources they put into developing F1 engines are not completely wasted ... so that some of the tech can "trickle down" to road car engines. Hence, the interest in hybrids. At least they get to learn how to charge a battery really, really fast.
In practice, of course, it's all going to go a bit south when road cars go to 90% EV. There is already a "formula E", of course.
Moved from one of the most fun auto sports to an EV mess.
The fact that they were able to vibe code ( You can't truly deliver a code without testing gates in place, let alone release to production, the cars themselves), tells:
Apollo 14 was one of the original moon landing missions back in 1971. They had to make an emergency adjustment to the computer's memory to work around an electrical issue where the abort button would intermittently (and erroneously) register as pressed. Turns out they couldn't modify the actual machine code so it wasn't a "software patch" in the exact same sense, but there were certainly no large language models back then.
Seems wild that they can code and deploy an update of a critical-seeming component in the field like that. I thought such a firmware update would need extensive QA in the lab, along with root causing why this bug slipped through their processes anyway.
But maybe that's the difference between shipping a controller for F1 versus shipping one for an industrial or consumer product? shrug.
It seems they simply disabled a bunch of power limits. Probably just settings, rather than an actual code update.
I have no idea if modern-day SV’s “tests pass… idfk, it’s a web app… LGTM I guess” coding ethos was the cause of this particular bug. However, it’s a great, relatively benign demonstration of why it has absolutely zero fucking place controlling extremely dangerous hardware shooting a human around a race track surrounded by people.
F1 are effectively prototype cars.
Every GP is a slightly different car running bleeding edge tech and code.
I'm wondering how good the simulators are. At the start of the season, they were still learning the systems trying to figure out why they were losing power in the straights. So depending on the driver, issues may or may not become apparent. That's like trying to debug why an email can only be sent to a recipient less than 500 miles away.
Anthony Davidson who is the Mercedes test driver said that the teams can simulate these scenarios and it seemed that the Mercedes cars were not affected so it seems that other teams hadn't done similar testing in various scenarios.
Interesting that both Mercedes also had great starts relative to the Red Bull that was affected, wonder if that’s somehow related
One of the mercedes literally broke down and did not finish the race. They might not have faced the issue during the first time they went out in that race but Merc reliablity has been very shaky. And ofc Ant will talk only good things about Mercedes. But this is not the type of things they use the driver simulator for, it is to help the drivers train. These types of issues don't make sense to spend simulator time on with the driver in it
Seems a lot better place to learn than in the middle of the race and earning $0 for the race. Isn't that the entire point of a simulator to learn things that are too expensive to learn in the real world?
It was probably just "config" changes or changing some hard-coded constants. So relatively straightforward.
I think those engineers had some memories of their FSAE / FSUK days during that patch.
When such complex and powerful machine is doing the actual run, and it has complex logic circuitry, capable of making some decisions, it is similar to horse racing and the machines should be awarded (same as horse), not the jockeys.
I just finished watching the race. The start was certainly a callback to the start of the season when cars were dropping like flies and having constant odd problems.
Yes. And wasn’t this the first wet race of the season as well?
A question for someone who knows more about F1: Is this firmware shared by all cars? Or some local server at the track the cars connect to? Why is F1 able to control individual cars like this? The article is very light on details as to what exactly is going on...
The firmware situation is mixed. The ECU hardware is a standard TAG700 from McLaren. Basically these are software-defined ECUs where you can do whatever. However, FIA has some rules on what you're allowed to modify and has some "compulsory" software in there for monitoring and competition regulatory reasons. It appears there was very limited testing of this module in wet conditions which seems to have affected a lot of vehicles. (I work in Motorsports)
McLaren sold McLaren Applied, the makers of the TAG700, in 2021 and it’s now known as Motion Applied.
Adding to this summarizing my reading from elsewhere: It sounds as if the cause was going too slowly through a sector, and due to to combination of wet conditions and the specific layout of the sectors at Sepang it was triggered for the first time. Since everyone was driving in formation on the formation lap, multiple drivers triggered it.
Would this be safety-related, something like "going this slowly through a sector indicates a problem, kill the throttle"? Still seems heavy-handed since the cars were moving, if dangerously slowly.
Wow. I just assumed it wouldn’t be firmware because going from making a fix in 50 minutes to having cars flying at hundreds of miles per hour on that firmware is kind of nuts
So basically like the anticheat rootkits used in online games to detect ilicit software controls on the player pc's?
F1 vehicles have a lot of actively controlled systems - Engine, MGU, Diff, Brake Bias, Aero etc. Their way of allowing active control but preventing "driver aids" like traction control, torque vectoring is to tightly regulate and give them a box where they can place their controls and what inputs they are allowed to use.
The FIA has this software to stop teams from creating a form of traction control (which has been banned for a good while now)
Why? To increase chances of crashes that produces more views?
Traction control was allowed before 2008. The problem was that it trivialized some aspects of driving - drivers could floor the throttle through turns at little to no risk. (I suspect there may have also been some concerns that drivers who pushed traction control too far could have easily ended up in unrecoverable situations.)
There's some interesting discussion of the history and technical aspects at: https://www.f1-forecast.com/pdf/F1-Files/Honda/F1-SP2_25e.pd...
Crashes don't directly make for a better race; what matters is the amount of actual competition, and for that you need variation.
If every driver drove exactly the same way, then the race would be 100% down to basically just team strategy and the car itself. For fans of strategy that might be fine, but I think most people would find it much less interesting.
Basically, there's an endless struggle between "how much do we let automation control?" If there were no such rules, the role of the driver would be far less than it is now, which would effectively reduce competition.
There's some debate but it's kinda settled, it's the pinnacle of motorsports so no driver aids, they are moving away from hybrids, realizing road car relevance/parity is no longer the direction
I am not big in F1, but my understanding is that the primary goal is to have the driver be more relevant over the car/firmware producers.
Then why not have standardised vehicles and have the driver be the only variable?
Presumably because they also see at least some value in technical innovation.
It's possible for driver skill to be important without being the only thing that matters.
Traction control removes of the most fundamental skills of a driver. It’s like allowing ABS.
We should allow batting machines for baseball players to hit the ball more reliably as well. Poker players should play with their cards face up. I agree.
Manually controlling to the limits of the car is a big part of the skill of racing.
The entire purpose of traction control is to reduce the driver skill required when cornering and braking. This is why traction control is omnipresent in consumer vehicles, because the vast majority of road drivers are awful.
FIA bans traction control so as to place more emphasis on driver skill.
They all run the same ECU and the same firmware on it, yeah. Teams can control some parameters though. It's to stop any kind of traction control.
Wow, crazy that all cars have to run the same firmware (sans parameter tweaking). I thought as an outsider to the sport that F1 was all about engineering the best possible car and putting the best possible driver in the seat, but the control systems engineer in me is sad that they're all stuck with the same control algorithms.
The teams use a single homologated ECU to prevent driver aids. I would assume it doesn’t have a downlink on-track to prevent tampering, but presumably with umbilical cord connected in the garage during red flag they can make changes, which the FIA probably sent over. Who knows.
No this all happened during the formation lap. So teams could clearly push settings changes live.
I can't not read it as "They vibe coded a patch in 50 minutes". Had they used Sol 6.1 they would still be waiting
Still waiting for 6.1 sol to patch a UI bug as am reading this
Is 6.1 Sol that bad? Or just slow?
Auto software engineer here.
It’s frankly because Motorsport software guys are pretty arrogant. They look down on road car and truck engineers, but reality is they develop stuff with virtually no processes and they usually have only very basic engineering skills, yet every day they are told Motorsport is at the forefront of technology.
“Motorsport grade” auto components usually means crappy, and everything requires a software patch or constant maintenance to work correctly.
Not surprised there was a bug, and not surprised they were able to develop and deploy a patch in 1 hour without V&V testing it or anything.
"Meet the F1 developer who fix a critical bug in 50 minutes with AI agent"
2000 PRs per day btw.
For anyone not willing to click through, by "Bahrain" they actually mean Malaysia.
Time to go back to NA engines, the hybrids don't give the rock concert wall of sound
Lets all vent in one more place. How this kind of regs get forced is sad. All the casual fans want something fun to watch. All the serious fans want NA engines. The drivers hate these "power units".
Can you repeat the question?
I don't follow motorsports very closely, sorry, but racing hybrids sounds extremely queer.
They might as well not do it at all.
You cannot greenwash auto racing as "sustainable" when by its very nature it need not exist. It's done for sport.
It's like people who hunt exotic animals agreeing to put suppressors on their rifles as if that somehow makes it more acceptable.
In theory: it's not greenwashing, it's so that the humungous amounts of resources they put into developing F1 engines are not completely wasted ... so that some of the tech can "trickle down" to road car engines. Hence, the interest in hybrids. At least they get to learn how to charge a battery really, really fast.
In practice, of course, it's all going to go a bit south when road cars go to 90% EV. There is already a "formula E", of course.
Make race cars analog again
Claude, make my car fast, make no mistakes
Honestly at this point, bring back the V10's.
The Enshittification era keeps on giving.
Moved from one of the most fun auto sports to an EV mess. The fact that they were able to vibe code ( You can't truly deliver a code without testing gates in place, let alone release to production, the cars themselves), tells:
1: The teams no longer own the car
2: F1 has become Formula E 2.0
F1 teams have never fully owned the car, you’ll notice they all look pretty similar. The FIA has a lot of rules around the construction of the cars
FIA, the FIFA of motorsport.
I’ve really enjoyed this year.
Gotta love how people assume software patches have never been prepared in an emergency before LLMs. Ever heard of Apollo 14?
Apollo 14? Is that a new model?
Apollo 14 was one of the original moon landing missions back in 1971. They had to make an emergency adjustment to the computer's memory to work around an electrical issue where the abort button would intermittently (and erroneously) register as pressed. Turns out they couldn't modify the actual machine code so it wasn't a "software patch" in the exact same sense, but there were certainly no large language models back then.