I was at a red light in Bellevue, phone in hand, doing what every adult in America does at a red light. The car was stopped. The engine idling. I wasn't moving.

A police officer was in the lane beside me. The moment the light turned green and we both pulled away, his lights came on.

He was polite. I was cooperative. He told me he was ticketing me for using a phone while driving. I told him I wasn't driving—the car was stationary, at a red light, in a line of other stationary cars. He agreed that this was true. I asked him, genuinely curious, whether he thought it was actually dangerous. Whether a person sitting still at a light, looking at a phone, had ever caused an accident. He thought about it. He didn't have an answer. What he said instead was something to the effect of:

I don't make the laws. I just enforce them.

He handed me a $136 ticket and drove away. Of course, he was right. He didn't invest the law or set the consequences.

So I sat there fuming. Thinking: who is harmed here? Not in theory. I mean literally. Moreover, is there a single documented case of a stationary car at a red light, driver looking at a phone, causing an accident?

Which got me thinking (usually a dangerous thing). What if the 'who is harmed' were the actual basis for whether a law existed? Like if no-one could produce evidence of harm to anyone, then i shouldn't be fined for it, should I?

How would a data scientist approach this? Let's take a hypothetical. Today we have a driving speed of 60mph on the freeways. We might want to test whether 70mph would actually cause any more harm. My guess that we all have an intuition about this but no-one really knows for sure. As a good data scientist, I would look at the data. Running a controlled experiment on increased speed limits would be the obvious thing to do. Roll it out slowly, and measure carefully, test in lots of different situations, times of day, street types, neighborhoods, etc. Not let our intuitions get in the way of the results. If the data shows not meaningful increase in harm, then we know that 70mph is safe. When that changes, when we have new data, we can adjust the speed limit again. We'd make this measurements systematically to constantly adjust our baseline. Wouldn't that be a better way to set the speed limit?

This isn't an argument against distracted or speed laws. It's about making our laws more dynamic and data-driven.

Because here's what the officer's sentence actually describes: a system where nobody is responsible for whether the law works. Not him. Not the council member who voted for it and moved on years ago. Not the lobbyist who pushed it. The law simply exists, self-perpetuating, collecting fines, because laws almost never get evaluated against their original purpose and almost never go away.

We treat laws like antique furniture. They arrive, get arranged, and then sit there gathering dust to the point where they don't work anymore and no-one understands why we got them in the first place.

We don't ask of a law what we ask of a medication: does this work, at what dose, for whom, with what side effects? A drug entering the market has to prove efficacy. A law should be no different. It should be evaluated continually. Right now, it only has to survive until the next election, which mostly means avoiding anyone powerful.

And this would go both ways. Cannabis, which serious pharmacologists rank below alcohol, and far below the opioids sold at any pharmacy with a prescription—is a Schedule I felony, filed beside heroin since 1970 under a law Nixon's own advisors later admitted was aimed at Black people and anti-war activists, not at pharmacology. Meanwhile the federal tipped minimum wage is $2.13 an hour. It hasn't moved since 1991. Thirty-four years. The documented harm—poverty, wage theft, workers unable to report harassment without losing their income, is enormous. The restaurant lobby is more enormous. So the law stays.

Same failure, opposite directions. One law does too much, aimed at the wrong target, for reasons that were never about harm. The other does too little, calibrated to political tolerance rather than human outcomes. The mechanism is identical: set the rule once, never check, enforce or ignore based on who's funding the apparatus.

I found an old idea in economics called Pigouvian taxation, after the British economist Arthur Pigou He proposed in 1920 that we "price an activity at the exact cost of the harm it causes". Not a round number that sounds serious, the actual cost. A factory dumping into a river pays what the pollution costs the people downstream: their health, their lost income, their water.

Sweden built a version of this into road safety in 1997 and called it Vision Zero. Speed limits set by road type and measured fatality data, not intuition or negotiation. They cut road deaths by half over twenty years.

One of the important features is that we constantly consider the data. The limit isn't carved into stone. It moves when the data moves. A stretch of road starts producing fatalities, the limit drops. It proves safe, the limit rises. The law is a living thing, re-aligned to reality on a schedule.

Some obvious objections

Switching sides, for a moment.

Who owns the metric? Who ever controls the measurement controls the law. If this is not democratic and transparent, or worse, allowed to be influenced by the people who benefit from it, then it's not a law. It's a tax.

Measuring relative harm How do we measure the harm of a law? If we decrease the speed limit and start taxing everyone small amounts of money for every infractions, that is an annoyance to a large amount of people. How do you compare that to the death of a child? No-one will ever advocate for something that has the potential of killing children.

Measuring subjective experiences Some people will will feel more strongly about raising the speed limit to 80mph because it effects their daily commute. Some people will feel strongly that higher speeds increase their noise pollution. How do you measure that? Is that measurable in a way that is not subjective? Once it's subjective then it's open to manipulation and corruption.

Conclusion

So no, I don't have a clean answer to any of this. You can't fully separate the harm of a dead child from the harm of a million tiny inconveniences. You can't build a metric nobody will try to game. Judgment never disappears.

But notice that every one of those objections is already true. We already trade lives for convenience, every speed limit above a walking pace is a quiet decision that some number of deaths is an acceptable price for getting to work on time. We already let metrics get captured: the pharmaceutical industry's studies set opioid policy, the restaurant lobby sets the tipped wage. We already govern by subjective feeling dressed up as fact. The current system doesn't avoid a single one of these problems. It just commits them with the lights off, once, and then walks away for fifty years.

A dynamic law doesn't remove human judgment. It does something smaller and more honest: it forces the judgment back into the open, on a schedule, and makes someone answer for it again and again. The officer who pulled me over couldn't tell me who was harmed. Neither could the law. The least we can do is build a system where someone, somewhere, is still required to ask.