Last week I was following someone home from the hardware store when their brake lights started strobing like a nightclub. Except nobody else seemed to notice.

Turns out about 1 in a thousand people can see light flickering faster than normal humans are supposed to. I'm one of them. And some car lamps? They're not actually "on"—they're flickering hundreds or thousands of times per second.

For most people, this is fine. Their brain smooths it out. For me, it's like driving behind a mobile disco. All great but honestly annoying and distracting.

But here's where it gets weird: It's causing some genuinely bizarre problems that reveal a spectacular failure of coordination in the automotive industry.

The Pokémon Problem

Remember that Pokémon episode in 1997 that sent like 600 kids to hospitals with seizures? That wasn't a one-off freak accident. Flashing lights can trigger photosensitive epilepsy in about 1 in 4,000 people.

After that incident, the broadcasting industry implemented rules: no rapid flashing above 3 Hz, especially with red. Video games followed suit. Even TikTok videos get flagged for strobing now.

Car lamps interior and exterior? Somehow got a pass.

So there are roughly 1,325 people in New Zealand at seizure risk from flickering automotive lights. Most car manufacturers stay above the 'danger' zone (3-60 Hz), but cheap aftermarket LED clusters? They're all over the map.

The Temporal Acuity Spectrum

Roughly 90% of gamers will complain about input lag and framerate stutter on a 60Hz display after using 120Hz. Ask any serious gamer about their monitor refresh rate and prepare for a lecture.

Of those, a smaller subset can perceive differences at 144Hz and above. Competitive gamers, FPV pilots, and people with naturally high temporal acuity live in a different perceptual world. I'm sure they are equally annoyed.

A brake light at 1% duty cycle (99% off-time) flickering at 45Hz is genuinely terrible—sharp, stabbing pulses. But 50% duty cycle at 20Hz? Annoying but less throbbing. Depends on a few variables.

For me, driving at night in more recent years is neurologically exhausting.

The Robot Vision Problem

This is where the story gets both hilarious and expensive.

Autonomous vehicles use cameras to see brake lights. But cameras don't see light the same way humans do—they sample discrete points in time rather than integrating a continuous image. Rolling shutter sensors scan line by line, global shutter sensors capture the entire frame at once, but both are taking snapshots rather than seeing a smoothed-out average.

When an LED is flickering on and off quickly, the camera might catch it "on" at the top of the frame and "off" at the bottom, or entirely "off" if the snapshot hits during a dark period. The result? The computer literally cannot tell if the brake light is on or off.

This is not a theoretical problem. This is an "engineers pulling their hair out" problem.

So camera manufacturers have spent the last decade developing expensive specialized sensors to work around flickering lights:

• Extended exposure times that capture multiple PWM cycles to average out the flicker
• Split-pixel architectures (dual photodiodes per pixel) that sample simultaneously—one for short exposure, one for long
• LOFIC (Lateral Overflow Integration Capacitors) that collect signal electrons in large storage capacitors
• Extreme dynamic range sensors (minimum 12 f/stops beyond standard automotive cameras) to capture both dim road backgrounds and blinding LED pulses without clipping
• HDR and LFM Engine (HALE) combination algorithms that simultaneously provide high dynamic range and LED flicker mitigation
• On-sensor processing that combines four independent exposure channels into a single high-fidelity image

Multiple companies—OmniVision, Sony, ON Semiconductor—have entire product lines dedicated to this problem. Split-pixel LFM is now a standard feature requirement for automotive imaging sensors, with some sensors operating up to 125°C specifically to handle this issue.

The Beautiful Part

The same companies make both parts!

Bosch, Continental, Valeo—these mega-suppliers sell:

• Lighting systems with PWM-flickering LEDs
• Camera systems with specialized flicker-compensation sensors

Different divisions. Different budgets. Nobody asking: "Wait, are we spending millions solving a problem we created?"

It's like if the same company sold you shoes that hurt your feet and also sold expensive orthotic inserts to make the shoes comfortable. Except in this case, they don't even realize they're doing it. I don't think.

Why Do They Flicker Anyway?

Two reasons: simplicity and cost savings.

To make an LED dimmer without changing the current, you pulse it: on for 10% of the time, off for 90%, and most eyeballs average it out to "less brightness."

This technique is called Pulse Width Modulation (PWM). It's simple, it's cheap, and it's been the go-to solution since LEDs entered mainstream lighting.

The problem tho, "most eyeballs" is not "all eyeballs."

The Fix That's Been Sitting There
Since before LED's

There are three ways to make LEDs not flicker I can think of at the top of my head. All proven. All cheap. Mostly ignored.

Option 0: Add a capacitor ($0.20 per light)

A capacitor stores electrical charge and releases it smoothly. Add one to the LED circuit and it fills in the "off" periods of the PWM cycle. The LED current stays steady. Even Dr. Emmett Brown knew this!

Option 1: Use a constant-current driver ($0.40-1.80 per light)

Instead of pulsing the LED on/off, use a circuit that regulates the current, thus the light output. Simple. Elegant. No flicker.

Option 2: Analog/linear dimming ($0.30-0.60 per light)

Use a variable resistor or transistor in linear mode to reduce current. Less efficient than Option 1 (wastes energy as heat), but dirt simple and still cheaper than developing $640+ camera sensors.

Total cost per vehicle (20-30 lights): $6-54

Total cost for specialized automotive camera sensors with LFM: $640-2,400+ per vehicle

Total human cost: Not on the balance sheet.

So Why Hasn't This Been Fixed?

Because nobody owns the whole problem. Probably.

The lighting division optimizes for: "Make auto lamps cheaper."

The camera division optimizes for: "Make cameras that work with auto lamps."

The safety regulators say: "Well, the lamps meet brightness standards, so..."

The automotive industry spends considerable resources developing cameras with LFM capabilities, HDR sensors, and sophisticated algorithms—all to work around flickering that costs $6-54 per vehicle to eliminate at the source.

It's not malicious. It's not a conspiracy. It's just... nobody's looking at the whole system. Probably.

The Broader Picture

Medical impact: About 1,325 people in New Zealand have photosensitive epilepsy. Flickering lights in the trigger frequency range (3-60 Hz) exist in the wild, mostly in cheap aftermarket lamps. Most OEM lamps stay above this range or don't flicker, but there's no regulation requiring it.

High temporal acuity: Between 50,000-250,000 people in New Zealand can perceive flicker above 200 Hz. The experience varies wildly based on duty cycle and frequency—1% pulses at 45Hz are far worse than 50% duty cycle at 20Hz. It's not just "can you see it," it's "how unpleasant is it."

Autonomous vehicles: Every major manufacturer developing self-driving tech has implemented specialized camera sensors with LED Flicker Mitigation capabilities. Sony, OmniVision, ON Semiconductor, and others have made this a standard feature in automotive imaging products.

Economic absurdity: The automotive industry has invested heavily in developing sensors with 140 dB HDR and LFM technology to work around a problem that costs $6-54 per vehicle to eliminate at the source.

And every major supplier? They're selling both sides of this equation.

What Happens Now?

Probably nothing.

The lights work "fine" for regulatory purposes. The cameras compensate. Most people don't notice. The few who do get written off as edge cases.

Meanwhile, Bosch's lighting division and camera division will continue their excellent work, completely independent of each other, solving and creating the same problem in parallel.

It's perfect. In a spectacularly stupid way.