Welcome to My Weird Prompts, Episode 201. So Daniel sent us this one — he wants to talk about the evolution of automotive safety standards, specifically anti-lock braking systems and emergency traction controls, and how safety is still improving today. And the thing that grabbed me right away is this: your car can now detect a patch of black ice before your foot even leaves the accelerator pedal. That's not a metaphor. That's shipping in production vehicles right now.
It's wild when you say it like that. The car is perceiving things you haven't perceived yet. We've gone from "try not to die in the crash" to "avoid the crash" to "predict the crash before the driver even knows there's a threat." That's the fundamental paradigm shift of the last fifty years.
Where do we even start with this?
I think you have to break it into three eras. The mechanical era — seatbelts, crumple zones, airbags. All designed around the assumption that the crash is happening, and your job is to survive it. Then the reactive electronic era — that's where ABS and stability control live. The car detects something going wrong and intervenes in milliseconds. And now we're in the predictive electronic era — sensor fusion, vehicle-to-everything communication, over-the-air updates that make the car safer while it sits in your driveway.
"Survive the crash" to "avoid the crash" to "predict the crash." That's the arc. And the bridge between era one and era two was a system that could pump brakes faster than any human ever could.
So let's start with that first leap: anti-lock braking systems. How did a 1978 Mercedes option become a mandatory safety feature on every car sold today?
Lay it out for me. What's the actual mechanical problem that ABS solves?
Okay, so think about what happens when you slam on the brakes in a panic stop. Without ABS, the wheels lock up. The tire stops rotating and just skids across the pavement. Two bad things happen immediately. One, you lose steering control — a sliding wheel can't generate lateral force, so turning the steering wheel does nothing. The car just goes straight, wherever its momentum is carrying it. Two, on most surfaces except maybe deep gravel, a locked wheel actually produces less braking force than a wheel that's right at the edge of locking up.
Why is that? Intuition says maximum friction equals maximum stopping power.
It's the difference between static friction and kinetic friction. A rolling tire maintains static friction with the road — the contact patch is momentarily stationary relative to the pavement. A sliding tire has kinetic friction, which is lower. The peak braking force happens at somewhere around ten to twenty percent slip — the wheel is turning slightly slower than the car's speed, but it's not locked. ABS tries to keep every wheel right in that sweet spot.
It's not just about steering. It's about actually stopping shorter.
On most surfaces, yes. But here's where the misconception comes in — and this is important. On loose gravel or deep snow, ABS can actually increase stopping distance by fifteen to twenty-five percent. Because a locked wheel on a loose surface builds up a wedge of material in front of it — gravel or snow piles up and adds resistance. That's the "plowing effect." ABS prevents that wedge from forming.
The tradeoff is: longer stopping distance on gravel, but you can steer around the thing you're trying not to hit.
And the data bears out that this is a good trade. A 2012 NHTSA study found ABS reduces fatal crash risk by six percent on dry roads but thirty-four percent on wet roads. On wet pavement, maintaining steering control is everything.
Okay, so walk me through the actual mechanism. How does a 1978 Bosch system pump brakes faster than a human?
The first production ABS was Bosch's system on the 1978 Mercedes-Benz S-Class, the W116. It was a two-channel system — it controlled the front wheels individually but the rear wheels together as a pair. Modern systems are four-channel, each wheel modulated independently. Here's the control loop. Each wheel has a speed sensor — usually a Hall effect or magnetoresistive sensor reading a toothed ring on the hub. That sensor feeds data to a microcontroller dozens of times per second. The controller is running what's called a slip-ratio algorithm — it's comparing each wheel's speed to the vehicle's reference speed and calculating the percentage of slip.
When it detects a wheel about to lock?
It enters a three-phase cycle: pressure hold, pressure release, pressure reapply. It does this at fifteen to thirty hertz — fifteen to thirty times per second. The fastest human driver might manage two or three pumps per second in a panic situation. And the human is pumping all four brakes together. ABS is doing per-wheel modulation. Your left front wheel might be on dry pavement while your right front is on ice. The human foot can't possibly handle that split.
That split-mu surface — one wheel on ice, one on dry — that's where things get interesting, right?
This is the hidden complexity most people never think about. If you brake hard with one wheel on ice and one on dry pavement, the braking forces are wildly asymmetric. The car wants to spin toward the side with more grip. Early ABS systems couldn't handle this well. Modern systems integrate yaw rate data — the car senses it's starting to rotate and adjusts brake pressure side-to-side to counteract the spin. It's not just preventing lockup anymore. It's actively managing vehicle dynamics.
ABS bleeds into stability control. The lines start blurring.
And the bridge between them is traction control. Let me explain the relationship, because people constantly confuse these systems. Traction control is essentially ABS in reverse. Instead of releasing brake pressure on a wheel that's locking up under braking, it applies brake pressure — or cuts engine torque — to a wheel that's spinning under acceleration.
If I'm on ice and I floor it, one wheel spins uselessly while the other sits there?
And here's the physics problem: an open differential sends torque to the wheel with the least resistance. That's the spinning wheel. The wheel with grip gets nothing. Traction control detects the speed difference between the driven wheels, applies the brake to the spinning wheel, and that braking torque creates resistance that the differential can work against — it forces torque to transfer to the wheel with grip. It's an electronic limited-slip differential, essentially.
That's clever. Using the brake as a torque-routing device.
It's fast. The first car in the US with standard ABS and traction control was the 1989 Ford Thunderbird Super Coupe. But the real synthesis came with electronic stability control. ESC adds two critical sensors that ABS and traction control don't have: a steering wheel angle sensor and a yaw rate sensor, plus a lateral accelerometer.
Now the car knows where you're pointing the wheel and where the car is actually going.
Those are two different things when you're losing control. The yaw rate sensor is typically a tiny vibrating structure gyroscope — it measures the car's rotation around its vertical axis. The steering angle sensor tells the computer where the driver intends to go. If the car is rotating faster than the steering angle predicts, you're oversteering — the back end is coming around. If it's rotating slower, you're understeering — the front end is plowing wide. ESC applies individual brakes to counter the rotation. Left-front brake to catch oversteer, right-rear to catch understeer — the specifics depend on the situation, but the principle is: use asymmetric braking to create a correcting torque.
This is the system that became mandatory.
The 1995 Mercedes-Benz S-Class, the W140, was the first production car with full electronic stability control — they called it ESP. The NHTSA started pushing for a mandate in 2004, and by 2012, ESC was required on all new passenger vehicles in the US. A 2006 NHTSA study found ESC reduces single-vehicle crash risk by forty-nine percent and rollover risk by eighty percent. Those are enormous numbers. ESC is arguably the most important safety innovation since the seatbelt.
Forty-nine percent reduction in single-vehicle crashes. That's not a marginal improvement. That's a different category of safety.
It works because it addresses the root cause. Single-vehicle crashes are usually loss-of-control events — the driver enters a corner too fast, or swerves to avoid something, and the car exceeds its handling limits. ESC catches it before the driver even realizes they've made a mistake. It intervenes in the first hundred milliseconds of instability, when the correction needed is small. A human driver doesn't notice the slide until it's already significant, and by then the correction needs to be much larger — and often it's too late.
That hundred-millisecond window keeps coming up. The car's reaction time versus the human's. It's the thread that connects all of these systems.
It's why the next generation is so significant. ABS and ESC gave the car the ability to react. But the newest systems don't just react — they predict. And that requires a completely different sensor architecture.
Let's talk about what's in a 2026 production car. What's the sensor stack actually look like?
Take something like the Mercedes EQS or the Volvo EX90. You're looking at four to six radar units, eight to twelve ultrasonic sensors, two to three lidar units, and eight or more cameras. All of this feeds into what's called a safety domain controller — a dedicated computer running sensor fusion algorithms that build a three-hundred-sixty-degree model of the world around the car, updated dozens of times per second.
Sensor fusion being the same Kalman filter techniques that keep your phone's GPS accurate, just applied to a much harder problem.
Same mathematical foundation, vastly more complex implementation. Each sensor type has different strengths. Cameras are great at classification — is that a pedestrian or a mailbox? — but terrible at ranging in bad weather. Radar is excellent at measuring distance and velocity and works in fog and rain, but it's low-resolution. Lidar gives you precise three-dimensional point clouds but it's expensive and can struggle with reflective surfaces. The fusion algorithm weights each sensor's confidence based on conditions and cross-validates. If the camera and radar both say there's an object at forty meters, confidence is high. If they disagree, the system has to decide which to trust.
This is where predictive safety comes in. The car seeing things you literally cannot see.
The Mobileye EyeQ6 system, which started shipping in 2025, has a feature they call "visibility of the invisible." It uses radar multipath processing. Here's how it works: when a radar pulse hits an object, some of the signal bounces directly back to the sensor. But some bounces off the road, off nearby cars, off buildings — and those multipath reflections carry information about objects that are occluded, hidden behind other vehicles. The EyeQ6 can detect a pedestrian behind a parked van up to a hundred meters away by analyzing these secondary reflections. The driver can't see the pedestrian. The camera can't see the pedestrian. But the radar, combined with the processing algorithm, can.
The car is braking for a child you haven't seen yet. That's not a hypothetical.
It's in production. And that's just onboard sensing. Add vehicle-to-everything communication and it gets even more interesting. The FCC allocated spectrum for C-V2X in the 5.9 gigahertz band — thirty megahertz of dedicated bandwidth for safety communications. An intersection equipped with an infrastructure unit can broadcast a "red light violator approaching" alert to every vehicle within three hundred meters. Your car knows someone is about to run a red light before you can see the light, before you can see the other car.
The safety envelope extends beyond the car's own sensors. It's a network effect.
The network gets more valuable with every vehicle and every intersection that joins it. A car approaching a blind curve can receive a warning from an oncoming vehicle it can't see. An ambulance can broadcast its approach path and every car within range can clear the lane before the siren is audible. This is infrastructure that's being deployed right now in pilot programs across the US and Europe.
Then there's the software side. Over-the-air updates changing how safe a car is after you've bought it.
This is one of the biggest shifts, and I think it's underappreciated. In the old model, a car's safety features were frozen at manufacture. Whatever software was flashed onto the ABS controller or the airbag module on the assembly line — that was it for the life of the vehicle. Maybe a dealer could apply a recall update if you brought it in. Now, Tesla's 2024.44 over-the-air update improved automatic emergency braking pedestrian detection by twenty-two percent in controlled IIHS testing. A software update. No hardware changed. The neural network weights were retrained on more data, validated, and pushed to the fleet.
Twenty-two percent is a meaningful number. That's lives.
It raises an interesting question about vehicle ownership. A car that gets safer over time — that's a fundamentally different product than what we grew up with. My first car was as safe on the day I sold it as the day I bought it. Actually less safe, because components wear. Now you can buy a car and six months later it's better at avoiding accidents than when you drove it off the lot.
Which also means the regulatory framework has to adapt. You can't crash-test a car once and certify it for life if the software changes every quarter.
The regulators are scrambling to catch up. The NHTSA just finalized the rule mandating automatic emergency braking on all new passenger vehicles by September 2029, including pedestrian detection at speeds up to forty-five miles per hour. They've also proposed a rule for driver monitoring systems that detect impairment — not just camera-based eye tracking, but analysis of steering wheel torque patterns. A drowsy driver makes characteristic micro-corrections that are different from an alert driver's inputs.
The steering wheel as a biometric sensor. That's clever because it works even if you're wearing sunglasses or a face mask.
The Volvo EX90, launching as a 2026 model, has what they call the Driver Understanding System. Two interior cameras tracking eye gaze and head position, combined with capacitive sensors in the steering wheel — it knows if your hands are on the wheel and where you're looking. If it detects microsleep — those two-to-three-second lapses where your eyes close and your head dips — it escalates through warnings, then brings the car to a controlled stop if you don't respond.
The EU has been ahead on some of this. They passed the General Safety Regulation in 2024, right?
Yes, mandating intelligent speed assistance, driver drowsiness detection, and advanced emergency braking for all new vehicles. The speed assistance piece is interesting — it uses GPS and camera-based sign recognition to know the speed limit, and it can either warn the driver or actively limit speed. You can override it by pressing harder on the accelerator, but the default is compliance.
That tension between safety and driver autonomy is the uncomfortable part of all this. At what point does the car refuse to let you make a mistake?
It's already happening in subtle ways. ESC doesn't ask permission — it intervenes. AEB doesn't ask permission — it brakes. The Mercedes Drive Pilot Level 3 system, which is approved for use in Germany and parts of the US, has redundant braking with two independent brake actuators, each capable of bringing the car to a full stop, plus a third backup on the steering rack. When that system is engaged, Mercedes — not the driver — is legally responsible for any accident.
That's a big deal. The manufacturer accepting liability.
It changes the entire incentive structure. Once the manufacturer is on the hook, they have every reason to make the system as conservative and safe as possible. Which might mean the car refuses to do things a human driver would attempt.
With all this technology in the car, what should a person actually do as a driver? Let's get practical.
Three things I'd tell anyone listening. First, when you're buying a used car, prioritize models with ESC and AEB. ESC was mandated after 2012 in the US, so anything from model year 2013 or later should have it. AEB became increasingly common after 2020, but it's not universal yet — check the specific vehicle. These two features are the highest-impact safety additions after seatbelts. The ESC numbers alone — forty-nine percent reduction in single-vehicle crashes — that's worth a slight premium on the purchase price.
Understand the limitations. ABS increases stopping distance on loose gravel and deep snow. Traction control cannot overcome physics on black ice — if there's zero friction, there's zero friction. The car is a safety multiplier, not a safety guarantee. I've seen drivers get overconfident in winter conditions because their SUV has all-wheel drive and traction control and stability control. None of that helps you stop on ice. None of it.
The all-wheel-drive overconfidence problem. People think it helps them stop.
All-wheel drive helps you go. It does nothing for stopping. That's entirely about tires and friction. Which brings me to the third thing: keep your car's software updated. Over-the-air safety updates are now as critical as oil changes. Check your manufacturer's recall and update portal quarterly. If your car doesn't support OTA updates, check for technical service bulletins related to safety systems when you get it serviced. A lot of improvements are being made to existing vehicles through dealer-applied updates, and most owners never know about them.
The software-defined car means maintenance isn't just mechanical anymore.
The average driver has no framework for thinking about that. We're trained to check tire pressure and change the oil. Nobody trained us to check if our automatic emergency braking algorithm is current.
Let's pull back for a minute. All these systems — ABS, traction control, ESC, AEB, V2X, driver monitoring — they're all converging toward something. What's the destination?
I think we're approaching what you could call a safety singularity — the point where the car's perception and reaction time exceeds human capability in essentially all driving conditions. We're not there yet. A human is still better at interpreting ambiguous situations — is that plastic bag blowing across the road something I need to avoid, or can I drive through it? But for the core safety functions — detecting imminent collisions, maintaining control in emergency maneuvers, monitoring for driver impairment — the machine is already better.
The next frontier isn't better sensors. It's better decision-making.
The ethical dilemma that engineers are grappling with right now: how does the car choose between hitting a pedestrian and swerving into oncoming traffic? That's not a sensor problem. The sensors can see the pedestrian and the oncoming truck perfectly. It's a decision problem. What's the right choice when all options cause harm?
The trolley problem, but at sixty miles an hour with a hundred milliseconds to decide.
The uncomfortable truth is that there's no consensus on the right answer. Different cultures have different preferences. A famous study — the Moral Machine experiment from MIT — found that people in different countries have systematically different preferences about who to protect in unavoidable crash scenarios. Should the car prioritize its occupants? Prioritize the young over the old? There's no universal ethics of crash optimization.
Which means the engineers are making ethical choices and encoding them in software, whether they want to acknowledge that or not.
Every line of code in the decision module is an ethical choice. "Brake force allocation in multi-target scenarios" — that's a dry engineering term for "who gets hit and how hard." The industry is uncomfortable talking about it, but it's inescapable.
There's a deeper question here — one that I think gets less attention. As cars become safer, do drivers become more reckless?
It's a well-documented phenomenon. When people feel safer, they take more risks. Drivers with ABS tend to drive slightly faster and follow slightly closer. Drivers with ESC tend to corner a little harder. The net safety improvement is still positive — you're better off with the systems than without — but it's not as large as the engineering numbers alone would predict.
There's a safety ceiling we can't cross without full autonomy. Because as long as a human is in the loop, the human will adapt to the safety margin and consume some of it.
That's the theory, and there's evidence for it. The question is where the equilibrium point lies. If ESC reduces crash risk by forty-nine percent in isolation, but drivers adapt and give back ten percent of that through riskier behavior, you're still at a thirty-nine percent net reduction. That's still enormous. The question is whether there are diminishing returns — does each additional safety system produce a smaller net benefit because drivers adapt more aggressively?
It also depends on the type of safety system. Some interventions are invisible to the driver — you might not notice the car applying individual brakes to correct a minor slide. Others are very noticeable — AEB slamming the brakes on is hard to miss. The invisible interventions probably don't trigger as much risk compensation.
The less the driver perceives the intervention, the less they adjust their behavior. Which argues for making safety systems as seamless and unobtrusive as possible.
What's coming in the next model year or two that pushes this further?
The 2027 to 2028 model year will likely see the first production cars with what the industry is calling predictive crash avoidance. These are systems that use V2X and cloud data to reroute the car away from high-risk intersections before the driver even approaches them. Imagine your navigation system knows that a particular intersection has a thirty percent higher crash rate during evening rush hour in wet conditions, and it's currently raining and it's five-thirty PM, so it routes you around it. You never know there was a risk. You just take a slightly different route home.
The safest crash is the one that never happens. That's almost tautological, but it's the direction everything is heading.
It's weird, right? The ideal safety system is one you never experience. You just arrive at your destination, unaware that three different interventions happened along the way. The car braked slightly for a pedestrian you never saw, tightened the seatbelt for an impending collision that a V2X alert prevented, and routed you around an intersection where statistically you would have been in danger. You get out and think "nice drive.
" Three near-misses you never knew about. That's where we're heading — and it's weird, wonderful, and a little unsettling.
The unsettling part is the loss of agency. We're not used to machines making life-and-death decisions on our behalf without our awareness or consent. But we're also not very good at making those decisions ourselves. The data is clear on that.
Now: Hilbert's daily fun fact.
Hilbert: In the 1930s, British geographer Frederick Bailey described the tidal bore on Nepal's Kosi River as a "mascaret" — a term borrowed from French that originally referred to the tidal surge on the Seine, derived from an Old Occitan word meaning "stained" or "soiled," because the advancing wave churns up mud and turns the water brown.
" That's actually a pretty good name for a tidal bore.
It really is.
The safest car is the one that never gets into a situation where safety systems are needed. We're building cars that can see around corners, talk to traffic lights, and make ethical choices in milliseconds. But the unanswered question is whether we're building safer drivers — or just more complacent ones. The 2027 models will route you around danger you never see. At some point, the car isn't just protecting you from the road. It's protecting you from yourself. And that's a different relationship entirely.
Thanks to our producer Hilbert Flumingtop. This has been My Weird Prompts, Episode 201. Find us at myweirdprompts.
If you found this useful, leave us a review wherever you listen. It helps more than you'd think.
See you next time.
