Daniel sent us this one — he's been wanting to take a defensive driving course, maybe an advanced one, and he'd always pictured the curriculum as skid-pad training, sliding cars around, catching oversteer. Then he watched an actual course video and realized it's almost entirely about observation, not vehicle dynamics. He's asking four things: what these courses actually teach, what separates a trained driver from a minimally licensed one, whether there's measurable accident reduction, and — the one that really hits — whether they prepare you for the dangerous unpredictability of other drivers, which he says is especially relevant where we are in Israel.
That last question is the one most people don't think to ask until they've been driving somewhere like Jerusalem for about a week.
Where a turn signal is less a request and more of a surprise announcement.
The thing that jumps out at me is how right his observation is about the skid-pad assumption. Almost everyone pictures defensive driving as a playground of threshold braking and J-turns and wet-track slides. But police pursuit instructors — and this is documented — spend roughly seventy percent of their training time on scanning, spatial mapping, and decision-making under uncertainty. Not on the Hollywood stuff.
Seventy percent of the time looking and thinking, not drifting.
And the average driver has no idea that's the split. They think the course is going to make them a stunt driver and instead it rewires how they see the road.
Which is actually more useful, because most of us are not being pursued.
But here's why this matters right now — traffic fatalities have been climbing in a lot of countries, insurance premiums are spiking, and the gap between "I passed a driving test once" and "I've actually been trained" has never been more expensive. The average following distance in the US is one point five seconds. The recommended safe distance is three to four seconds. That gap — that literal gap between bumpers — is a pretty good metaphor for the whole problem.
Most drivers are tailgating the entire concept of safety.
They're tailgating their own survival. And Daniel's prompt gets at something deeper too, which is that most of us were trained to operate a vehicle — steering, braking, parallel parking — and then licensed as if that's the whole job. But driving in actual traffic is almost entirely about predicting what other humans are about to do wrong.
Which is a skill nobody tests for. The driving test is "can you stay in your lane" not "can you spot the guy who's about to merge into you without looking three seconds before he does it.
That's precisely what these courses target. The National Safety Council's Defensive Driving Course has been running since nineteen sixty-four — over sixty years of curriculum refinement — and it's built around the Smith System, five rules developed by Harold Smith in nineteen fifty-two. Still used by about eighty percent of Fortune five hundred fleet training programs.
Nineteen fifty-two. So we've known the core principles for over seventy years and most drivers still haven't heard of them.
And the five rules are all cognitive — aim high in steering, get the big picture, keep your eyes moving, leave yourself an out, make sure they see you. Not a single one of those is about what your hands are doing.
"Make sure they see you" is practically a philosophy of urban survival.
In Israel traffic it's practically a prayer.
Daniel's mental model was skid-pad heroics, and the reality is a course that teaches you to narrate hazards out loud and count following distance in seconds. Which, honestly, sounds less fun but significantly more alive.
That narration technique — commentary driving — is the same method used in UK police pursuit training. You literally speak aloud everything you observe, every potential hazard, every developing situation. It forces active scanning and exposes the gaps in your awareness immediately because if you're not saying anything, you're not seeing anything.
The gap between what you think you're noticing and what you're actually noticing, made audible.
Which is humbling. The UK introduced a hazard perception test in two thousand two, and the data shows trained drivers identify developing hazards two to three seconds earlier than untrained drivers. Two to three seconds is an eternity in a potential collision.
That's the difference between "oh" and "oh no.
It's not just reaction time — it's that trained drivers have automated the basic control of the vehicle. A novice uses about ninety percent of their mental bandwidth just keeping the car between the lines. A trained driver has freed up that bandwidth for prediction, for reading the road ahead, for spotting the pedestrian who's about to step off the curb before they've even turned their head.
The skill isn't faster hands. It's a quieter brain.
That's beautifully put. The brain that isn't busy managing the steering wheel can see the intersection three cars ahead.
Which brings us to the structure of what we want to dig into here. Daniel gave us four clear questions: the actual curriculum, the skill gap between licensed and trained, the accident data, and the other-drivers problem. We'll walk through each of those, and I want to spend real time on that last one because it's the part that keeps people up at night — the driver who does something so incomprehensible that no amount of your own skill feels like enough.
We'll look at what the evidence actually says about whether you can train for that. Because the intuition is "you can't prepare for random stupidity," but the curriculum disagrees.
It turns out random stupidity has patterns.
It absolutely does. And we'll get into the Cochrane Review data, the Pass Plus numbers from the UK, and some techniques that are genuinely useful the moment you pull out of your driveway. But first I want to sit with this observation-versus-handling split, because it's the foundation everything else rests on.
Let's nail down the distinction, because Daniel mentioned he wasn't sure if defensive and advanced driving are the same thing. They're not, but they overlap in ways that confuse people.
One sounds like you're protecting yourself, the other sounds like you're learning to chase someone.
Defensive driving is typically classroom-based, focused on hazard recognition, attitude, and the cognitive habits that prevent collisions before they start. The National Safety Council's Defensive Driving Course — the DDC — has been running since nineteen sixty-four. It's the most widely recognized curriculum in the US, and it's built almost entirely around the Smith System we mentioned.
Defensive driving is the theory layer. The "think before you're in trouble" course.
Advanced driving courses are different — they usually involve on-road or track-based coaching, they include vehicle dynamics like threshold braking and weight transfer, but even those courses still prioritize observation. The UK has something called the IAM RoadSmart advanced test, and a core component is commentary driving — literally narrating hazards aloud the entire time you're behind the wheel.
Which sounds awkward until you realize it's the same technique police pursuit instructors use.
It's the backbone of their training. And that's the thing Daniel picked up on from those YouTube videos — the police instructors look like they're teaching high-speed maneuvering, but the actual curriculum is seventy percent scanning, spatial mapping, and decision-making under uncertainty. The J-turns and the speed are the garnish, not the meal.
The skid-pad fantasy is basically the movie trailer for a film that's actually about eye movement.
That's a perfect way to put it. And it's worth underlining why: if you can't read the road three hundred meters ahead, no amount of car control saves you from the hazard you never saw coming. The observation comes first because it has to.
Which sets up what we're actually going to walk through. Daniel asked four things — what's taught, what separates a trained driver from a licensed one, does it actually reduce crashes, and can you train for other people's chaos. We'll take each of those in order, and I want to land hard on the data piece because people love to assume these courses work without actually checking.
The Cochrane Review numbers are interesting — they show a clear benefit but also a shelf life, which nobody talks about.
A shelf life on safety is a terrifying concept.
It really is. But first let's get inside the curriculum itself, because once you see what's actually being taught, the rest of it starts making sense.
Let's pull apart the defensive driving curriculum, because once you see the structure, Daniel's skid-pad assumption almost feels like someone designed the course specifically to correct it. The National Safety Council's DDC is built on three pillars — collision prevention theory, the Smith System, and something called the SEE strategy. Search, Evaluate, Execute.
Search, evaluate, execute. That sounds like a robot's to-do list.
It's deliberately mechanical. The idea is you don't trust your instincts — you run a conscious cycle every time you approach a potential hazard. Search the scene, evaluate what could go wrong, execute your response. And the Smith System is the operating system underneath that cycle. Five rules Harold Smith laid out in nineteen fifty-two.
We're still using them because nothing better has come along, or because they're just that solid?
The rules are: aim high in steering — meaning look further down the road than feels natural, not at the bumper in front of you. Get the big picture — don't fixate on one car, scan the whole scene. Keep your eyes moving — a static gaze is a blind gaze. Leave yourself an out — always have an escape path. And make sure they see you — eye contact, horn taps, headlight flashes, positioned so you're visible. Eighty percent of Fortune five hundred fleet training still runs on this. Seventy-plus years and nobody's found a better foundation.
What strikes me is that every single one of those rules is about where your attention is, not what your feet are doing.
That's the whole point. The curriculum treats the car as something you already know how to operate. What you don't know is how to see. And the advanced courses double down on this — the IAM RoadSmart test in the UK makes you do commentary driving, which is exactly what it sounds like. You narrate every hazard you spot, out loud, the entire drive. "Pedestrian approaching crossing from left, car in side street edging forward, cyclist in wing mirror preparing to overtake.
You're basically a sports commentator for your own potential demise.
It's humbling, because the moment you stop talking, the instructor knows you've stopped looking. It exposes your awareness gaps in real time. This is the same technique police pursuit instructors use — they're not evaluating how fast you can flip the car around, they're evaluating whether you saw the delivery truck pulling out three intersections ahead.
Which brings us to the cognitive load piece, because you mentioned earlier that a novice driver uses about ninety percent of their mental bandwidth just keeping the car straight.
Right — and that's not a metaphor. When you're new, basic vehicle control consumes almost all your working memory. Steering angle, pedal pressure, mirror checks, gear shifts — each one is a conscious decision. A trained driver has automated all of that. The control layer drops into the background, and suddenly you've freed up bandwidth for prediction. You're not reacting to the brake lights in front of you, you're anticipating that the car three ahead is drifting toward the lane line.
The training isn't making you a faster driver, it's making you a less busy driver.
Which is exactly what the UK hazard perception test quantifies. Introduced in two thousand two — drivers watch video clips from the perspective of being behind the wheel, and they click the moment they spot a developing hazard. Trained drivers score twenty to thirty percent higher than untrained drivers. And the time gap is the really striking number: trained drivers identify hazards two to three seconds earlier.
Two to three seconds at highway speed is what, the length of a football field?
At sixty miles an hour, two seconds is about a hundred and seventy-six feet. That's more than half a football field. It's the difference between controlled braking and a panic swerve. And the test doesn't measure hand speed — it measures how early your brain registers the threat.
Daniel's YouTube video — the police instructor weaving through traffic — what looks like incredible reflexes is actually just someone who saw the gap forming five seconds before the camera did.
That's it exactly. The physical vehicle handling — threshold braking, weight transfer, skid recovery — that stuff is taught in advanced courses, and it matters. But it's secondary. The primary skill is reading the road. Anticipating where a hazard will emerge, not reacting once it's already in your lane. The Smith System rules are all cognitive because the mechanical stuff only helps you if you've already spotted the problem.
Which makes me wonder about the tradeoff between classroom and on-road training. If the skill is seeing, can you really learn that in a room with a PowerPoint?
That's the tension. Classroom courses like the DDC are great for installing the mental framework — the rules, the cycle, the theory. But the evidence suggests the habits don't stick nearly as well without coached on-road practice. The commentary drive, for example, you can explain in a classroom in ten minutes. Actually doing it for an hour while navigating real traffic is a completely different experience.
Because the classroom doesn't have pedestrians who might step off the curb.
Or the emotional load of real consequences. When you're watching a video clip, you know it's a test. Your adrenaline isn't up, you're not managing speed and steering and a conversation with a passenger. On-road coaching puts the cognitive techniques under realistic load, and that's where the automation actually gets built. The classroom plants the seed, but the road waters it.
The ideal is probably both — the theory first, then the coached practice. But most people just do the classroom bit and call it done.
Which connects directly to what we're about to look at with the accident data — because the classroom-only approach shows a benefit, but it decays. And that decay is the part nobody talks about when they hand you the completion certificate.
We've seen the curriculum and the cognitive model behind it — but the question Daniel's really asking is what actually changes in a driver who's been through this. What's the gap between someone who passed the basic test and someone who's done the course?
Because on paper they both have licenses. They both passed. So what's different?
The difference is that a minimally licensed driver has demonstrated basic control and rule knowledge — they can parallel park, they know the speed limit, they can merge without hitting anything. A course graduate has practiced what the instructors call defensive driving habits. It's the difference between knowing the speed limit and automatically adjusting your speed based on sight lines, weather, traffic density, and the behavior of the cars three positions ahead.
One driver knows the rule, the other has internalized the conditions under which the rule is insufficient.
And there's a fallacy people fall into — the ten thousand hour rule applied to driving. They think more hours behind the wheel automatically makes you safer. But it's not total hours, it's deliberate practice of specific scenarios. Emergency braking on a wet road, scanning an intersection for red-light runners, positioning yourself to be visible to a truck's mirrors. You can drive for twenty years and never practice those things consciously.
You can drive for twenty years and just reinforce bad habits for twenty years.
Most people do. The average US following distance is one point five seconds — that's not a beginner number, that's an experienced driver number. Experience without training just makes you more confident in your bad habits.
Which brings us to the data question. Daniel asked whether any of this actually reduces crashes, and I know you dug into the Cochrane Review on this.
Ker and colleagues published a Cochrane Review in two thousand five analyzing twenty-one separate studies on defensive driving courses. They found a ten to twenty percent reduction in crash rates among drivers who took the courses. That's real, that's measurable.
Ten to twenty percent is meaningful but not revolutionary. It's not "take this course and never crash.
And here's the part nobody talks about — the effect decays. After six to twelve months without refresher training, the benefit starts fading. The habits don't self-maintain. You go back to your old following distance, your old scanning patterns.
The safety has a shelf life, like I said. You have to renew it.
That's where advanced courses with on-road coaching show something more durable. The UK's Pass Plus scheme — an advanced driving program for newly licensed drivers — showed a twenty-five to thirty-five percent reduction in crash rates among participants. And the effect held longer because the training happened in real traffic with an instructor in the passenger seat.
The classroom plants the seed but the on-road coaching makes it stick. That tracks with what we said earlier.
The Cochrane Review itself noted some methodological weaknesses in the studies — not all of them were randomized controlled trials, some relied on self-reported data. But the pattern across twenty-one studies is consistent enough that the conclusion holds: these courses reduce crashes, the effect is real, and it fades without maintenance.
Which raises the question of whether this should be mandatory for license renewal. But we'll get to that. Daniel's fourth question is the one I want to spend time on — because even if you're perfectly trained, you're still sharing the road with people who are not.
He specifically mentioned Israel, which is — let's be diplomatic — a high-variance driving environment.
High-variance is the most generous description of Jerusalem traffic I've ever heard.
Defensive driving courses address this directly through something called risk compensation. The idea is you don't assume other drivers will follow the rules — you assume they'll make the worst possible move, and you position yourself accordingly. Not out of paranoia, but as a predictive model.
Expecting the worst isn't pessimism, it's a forecasting tool.
And the techniques are specific. Scanning intersections for red-light runners before you enter — not after the light turns green, but as you're approaching. Leaving a safety gap of three to four seconds in front, not the one point five most people use. The two-second rule in dry conditions, four seconds in rain, six in snow or heavy traffic.
The advanced courses add another layer — mirror checks every five to eight seconds, and positioning for visibility. I read that one technique is stopping behind the car in front's rear tires so you can see their wheels turning.
That's a classic. If you can see their wheels, you know immediately if they're about to move. It also gives you room to pull around them if they stall or if someone approaches from behind too fast. It's a small adjustment that creates options.
Options being the whole point of "leave yourself an out.
Then there's the psychological layer. These courses explicitly address road rage and cognitive biases — the illusion of control, optimism bias, the belief that you're a better driver than you are. Trained drivers are taught to recognize when their emotions are escalating and to consciously de-escalate. You're not winning anything by retaliating against someone who cut you off.
The Smith System rule five — "make sure they see you" — that's proactive, not reactive. You're using eye contact, horn taps, headlight flashes before the situation becomes dangerous.
That's the shift. An untrained driver honks after someone has already done something stupid. A trained driver taps the horn to prevent the stupid thing from happening in the first place. It's the difference between a reaction and a communication.
You're not just driving your car defensively — you're managing the attention of everyone around you.
Which sounds exhausting when you describe it that way, but the whole point of automating these habits is that it becomes background processing. You're not consciously thinking "mirror check, mirror check" — your eyes just go there every few seconds because the pattern is burned in.
That's the answer to Daniel's concern about other drivers. You can't control what they do, but you can train yourself to see it coming earlier and to have already positioned yourself with an escape route. The randomness doesn't go away — your margin for handling it just gets wider.
We've covered the theory, the data, and the techniques — but I want to turn this into something you can actually use on your drive home tonight. The single most effective technique from any of these curricula is the commentary drive.
Which sounds like a podcast you do alone in your car.
It kind of is. Narrate your observations out loud for ten minutes every time you drive. "Car approaching from side street, pedestrian near the crosswalk, brake lights three cars ahead." It forces active scanning and exposes the gaps immediately — if you stop talking, you've stopped looking.
You don't need a course to start doing that tomorrow.
That's the beauty of it. The second one is the two-second rule. Pick a fixed object — a sign, a shadow on the road — and when the car ahead passes it, count "one thousand one, one thousand two." If you reach the object before you finish, you're too close.
Most people are going to find they're at about one-point-five seconds without realizing it.
Which is exactly the US average. In rain, extend to four seconds. Snow or heavy traffic, six. That alone probably does more for collision prevention than any vehicle-handling drill.
Commentary driving and counting following distance — two things that cost nothing and require no instructor.
If you want the full benefit, find a local defensive driving course. Many are online now and take four to six hours. Even a half-day session can start rewiring those habits. For Daniel specifically — in Israel, look for courses certified by the Ministry of Transport or the Israel Police driving school. They offer civilian versions of the same observation-heavy training we've been describing.
The Israel Police driving school teaching civilians feels like the ultimate "if you can't beat them, teach them.
Honestly, given what we said about the skill decay from the Cochrane Review — six to twelve months and the benefit starts fading — the real takeaway is that this isn't a one-and-done thing. You take the course, you build the habits, and then you have to maintain them deliberately. The commentary drive isn't just a training exercise, it's a maintenance tool you can use forever.
The course gives you the framework, but the ten-minute narration keeps it sharp.
That's the part most people miss. They treat the certificate like a vaccine when it's actually more like exercise.
Here's the question I keep coming back to — if the skills decay after six to twelve months, and the average driver never takes a refresher, should defensive driving be mandatory for license renewal? Every five to ten years, you have to prove you can still see the road, not just that you could see it when you were seventeen.
Some countries already do this. Japan requires periodic training for drivers over seventy, and Germany has mandatory refresher courses for certain license categories. The Cochrane Review data makes a pretty strong case that one-and-done training isn't enough — the decay is real, and it's consistent across studies.
The pushback would be enormous. "I've been driving forty years, I don't need a course.
Which is exactly the optimism bias these courses are designed to address. The forty-year driver is often the one following at one point five seconds with their gaze fixed on the bumper ahead.
The bigger shift, the thing I think actually changes the whole conversation, is what happens as driver-assistance systems take over more of the task. Lane keeping, adaptive cruise, emergency braking — the car is doing observation now.
That flips the training model on its head. The skill stops being "how do I avoid the crash" and becomes "how do I supervise the system that's supposed to avoid the crash for me." It's a new kind of defensive driving where you're not scanning other drivers — you're scanning your own car's behavior for moments it's about to get it wrong.
Which is almost more demanding, because the system lulls you into trusting it, and then it hands control back to you in the worst possible moment.
There's already research on this — the automation disengagement problem. The driver who hasn't touched the wheel in forty minutes suddenly has to make a split-second decision. The observation skills we've been talking about atrophy even faster when the car is doing the watching for you.
The defensive driving course of twenty thirty-six might be less about the Smith System and more about staying cognitively engaged while a computer does the boring part.
That might actually be harder to teach, because the enemy isn't other drivers' unpredictability — it's your own complacency.
Which brings us back to Daniel's original instinct. He wanted the course because he recognized there's a gap between being licensed and being prepared. The curriculum surprised him by being about observation rather than heroics, but the instinct was right. The gap is real, the training works, and the shelf life is shorter than anyone wants to admit.
Now: Hilbert's daily fun fact.
Hilbert: In the early Renaissance, explorers off the coast of Cape Verde reported auroras so vivid they appeared green tinged with crimson — a color signature we now know comes from oxygen atoms at different altitudes: green from oxygen around one hundred to two hundred kilometers up, and deep red from oxygen above two hundred kilometers, where the atmosphere is thin enough to let the longer wavelength emerge.
Oxygen is two different colors depending on how high it is. That feels like it should matter for something.
Oxygen has range. I respect that.
This has been My Weird Prompts. Thanks to our producer Hilbert Flumingtop. If you enjoyed this, the best thing you can do is leave a review wherever you listen — it helps more people find the show.
If you want to send us a question like Daniel did, email the show at show at my weird prompts dot com.
We'll be back next week. Drive like you're narrating it.
