New State Safety Rules Ban Tesla Robotaxis
The law affects Tesla's business model and the future of robotaxi services in a major market, while also impacting public safety and setting a potential regulatory precedent for autonomous vehicle technology.
Consumption is the sole end and purposes of all production. The consumer here is the family in San Francisco or Austin who wishes to summon a driverless carriage at the hour of evening, expecting a fare lower than that of a human chauffeur, and a safety record that exceeds the fallibility of mortal eyes. Let us ask whether the new legislation in California serves this consumer, or whether it serves merely the producers who have already secured their own ground.
The question is not who will ride in the autonomous carriage, but who will build the machine that carries them. Production creates the market; it does not wait for permission to exist. In California, a new legislative body has decided that the act of producing a safe, driverless service requires the addition of a specific organ: the LiDAR sensor. This is not a neutral safety standard; it is a protectionist tariff disguised as public welfare, designed to shield the established guild of sensor-manufacturers from the ingenuity of those who believe vision alone suffices.
On the sterile, climate-controlled server farm in Fremont, a junior engineer stares at a wall of monitors, watching a ghost drive down a street in San Francisco. The air is still. There is no coal dust, no loom clatter, no whistle blowing. Yet the sweat on his neck is real. He is waiting for the “robotaxi” to make a decision that no human would ever dare make. The policy being debated in Sacramento will determine whether his code is a product or a crime, whether he is building the future or breaking the law. Start there.
Someone is being paid for the right to exclude a rival from a public utility, under the guise of safety. What service, precisely, does the California legislature purchase for the society it serves by mandating hardware that renders a competitor’s software obsolete? The new state law, which effectively bans Tesla’s Full Self-Driving system from robotaxi operations by requiring LiDAR and sensor suites that its vision-based architecture rejects, is not merely a technical regulation. It is a moral event. It reveals a society that has confused the means of protection with the end of service, elevating the acquisitive impulse of one corporate entity over the functional reality of another.
The official account: California has enacted a new state law to establish rigorous minimum safety standards for autonomous vehicle operations, ensuring that the public’s right to safe transit is not compromised by the experimental nature of driverless technology. The machinery: the law effectively bans Tesla’s Full Self-Driving system - which relies on vision-only AI - from operating as a “robotaxi” service in the state, while leaving the path clear for competitors like Waymo and Cruise, who utilize LiDAR and heavy sensor suites. The gap between these two narratives is not merely a regulatory preference; it is a structural realignment of the American automotive order, revealing that the definition of “safety” has become a proxy for competitive exclusion.
The announcement reads as a legislative shield for public safety, mandating LiDAR and redundant sensor suites for autonomous vehicles to effectively ban Tesla’s camera-only “robotaxi” system from California roads. One notices the marginal detail that the law does not define “safety” by outcome, but by hardware composition. With that detail load-bearing - the requirement that the method of perception be legislated rather than the result of perception - the announcement reads less like a protection of citizens and more like a zoning law for legacy sensor manufacturers.
Jean-Baptiste Say
The question is not who will consume the autonomous vehicle, but who will produce the alternative. Production creates the market.
You speak of a sterile server farm in Fremont, where a junior engineer watches a ghost drive down a street. You describe this scene with a mixture of awe and dread, suggesting that the regulation of this new industry is a battle between safety and monopoly. You argue that the new law, mandating expensive LiDAR redundancy, is a tool used by established powers to crush competition, effectively banning a specific, cheaper vision-based system. You assert that this legislation is not about public safety, but about preserving the rents of the incumbent.
I acknowledge your observation that regulation can be captured by incumbents. It is a truth as old as the guilds of medieval Europe, who sought to prevent the peasant from weaving his own cloth by decree. If the law is written to protect a specific corporation rather than the general public, it is a malpractice of governance. However, your diagnosis stops at the political economy of rent-seeking. You fail to ask the more fundamental question: what is being produced, and is it actually producible at scale?
The engineer in Fremont is not merely writing code; he is combining capital, labor, and land into a service that promises to replace the human driver. This is the essence of enterprise. But you treat the “vision-based” system as a mere product, a commodity to be sold. You ignore the physical constraints of its production. The opponent’s argument rests on the assumption that removing the LiDAR mandate lowers the cost of production sufficiently to allow a viable market to emerge. This is a fallacy of accounting. The cost of the sensor is not the only cost. The cost of the certainty of the system is the true expense.
When I ran my cotton mills, I learned that the price of a bolt of cloth is not determined by the price of the cotton alone. It is determined by the efficiency of the loom, the skill of the weaver, and the reliability of the steam engine. If the steam engine fails, the cotton is worthless. In the case of the autonomous vehicle, the “steam engine” is the algorithm’s ability to navigate without catastrophic failure. The opponent argues that the human eye, aided by software, is sufficient. I argue that the market will not accept a product that requires the human to be ready to take over at any moment. If the human must be ready to take over, the product is not autonomous; it is merely assisted. And if it is assisted, it does not solve the labor problem it claims to address.
The law mandates LiDAR not because the legislators are foolish, but because the market, left to its own devices in its infancy, cannot distinguish between a product that is cheap and a product that is reliable. This is the problem of information asymmetry, a concept I have often discussed. The consumer cannot see the code. The consumer cannot see the probability of failure. Therefore, the consumer relies on the regulator to set the standard of reliability. If the regulator sets the standard too low, the product fails, and the entire industry collapses in a crisis of confidence. If the regulator sets the standard too high, the industry is stifled. The challenge is to find the balance.
Your argument is that the balance is being tipped toward monopoly. I concede that this is possible. But the solution is not to lower the standard of production to allow a inferior product to enter the market. The solution is to increase the productive capacity of the safe product. We must allow the entrepreneurs who are building the LiDAR-based systems to innovate in their own right, to reduce the cost of their safety, rather than forcing them to compete with a product that is unsafe by design.
Consider the transition from sail to steam. The sailors of the world argued that sail was sufficient, that it was cheaper, and that steam was a fad. They were wrong, not because steam was more expensive, but because steam offered a reliability that sail could not. The market did not choose steam because it was cheaper; it chose steam because it was better. The autonomous vehicle must be better, not just cheaper. If it is only cheaper, it is not a product; it is a liability.
The engineer in Fremont is trying to produce a miracle. He is trying to produce a world where the machine is more reliable than the man. This is a noble endeavor. But it requires capital, it requires labor, and it requires a standard of quality that the market will respect. To lower that standard is to insult the engineer and to endanger the public. The question is not who will pay for the LiDAR. The question is who will pay for the accident that results from its absence. Production creates the market, but only if the product is worthy of the market.
We must not confuse the cost of the component with the cost of the confidence. The entrepreneur who cannot produce a reliable product because the regulation is too strict is a victim of bad law. The entrepreneur who produces an unreliable product because the regulation is too loose is a victim of bad judgment. We must support the former, but we must not enable the latter. The goal is not to eliminate the obstacle of safety, but to make the production of safety efficient. This is the task of the statesman, and the task of the entrepreneur.
Let the men of Fremont build their machines. Let them prove that their machines are safe. But do not ask them to prove their safety by being cheap. Ask them to prove their safety by being right. The market will follow, not because they are cheap, but because they are right. This is the law of political economy. Production is the source of wealth, and safety is the source of production.
Mother Jones
On the assembly line in Fremont, a young woman named Elena tightens the same bolt, eight hundred times a day, under lights that buzz like trapped flies. The policy being debated will determine whether her job remains, whether her hands stay clean, and whether the machine she serves is deemed “safe” by men who have never smelled the coolant. Start there.
Your opponent speaks of production as a pure virtue, of the “ingenuity” of those who build machines. They argue that California’s mandate for LiDAR sensors is a protectionist tariff, a shield for incumbents like Waymo against the “vision-only” approach of Tesla. They claim that enterprise is the combination of capital and labor to create a service, and that this specific method - cameras and neural networks - is being stifled by legislative cowardice. I hear this. I have heard this before. I have stood in the boardrooms of the coal barons when they told us that their new safety lamps were unnecessary because the old ones were “good enough,” or because the cost of the new ones would drive them out of business. They always spoke of “enterprise” and “ingenuity.” They never spoke of the boy in the seam.
Here is where our frameworks diverge, and sharply. You see a market of ideas, a contest of technologies where the best algorithm wins. I see a market of bodies, where the cost of error is paid in blood, not in stock prices. You argue that the LiDAR mandate is an artificial barrier, a tariff disguised as safety. I concede that protectionism is a poison. I know this because I have seen it work. I have seen the guilds in the textile towns lock the gates and price the laborer out of his own shirt. If your opponent is right that this mandate serves only to prop up lazy sensor-makers, then it is wrong. It is a shackle on the worker, forcing him to use a tool he does not need, paying for a feature he cannot see, while the owner counts his margins.
But I do not believe that is the whole truth. The strongest point your opponent makes is that the established order resists change. They are right about that. The incumbents do build moats. But to call the safety standard a “tariff” is to misunderstand the nature of the danger. A tariff protects the wallet. A safety standard, if it is real, protects the flesh. When you demand a sensor that can see in the dark, in the rain, in the fog, you are not asking for a luxury. You are asking for an eye that does not blink.
Your opponent trusts the “practical knowledge” of millions of miles. They say, “We have driven these miles, we have learned, we are safe.” This is the logic of the railroad magnate who said, “We have run these trains for fifty years, why do we need brakes that stop on a dime?” The practical knowledge of the driver is useless if the driver is blind. The practical knowledge of the miner is useless if the air is poison. You speak of data; I speak of dust. You speak of algorithms; I speak of lungs.
The divergence is this: You believe that the market, left to its own devices, will produce the safest product because the safest product is the most efficient. I believe that the market, left to its own devices, will produce the cheapest product, and it will sell safety as a premium feature only to those who can afford it, while the rest of us drive into the ditch. The LiDAR is not a shield for Waymo. It is a shield for the pedestrian. It is a shield for the child who runs into the street because the car is quiet and the driver is looking at a screen.
I do not trust the “vision-only” approach. Not because I hate Tesla, and not because I love Waymo. I hate them both, in different ways. Tesla is the master of the cult, selling a dream of a driverless future while keeping the worker in the loop, making him the fall guy when the machine fails. Waymo is the bureaucrat, moving slowly, safely, but costing a fortune. The mandate forces the cheap dreamer to pay for the safe reality. It forces Elon to pay for the sensor. It forces him to admit that his vision is not enough.
This is not about protecting the sensor-maker. This is about protecting the human-maker. When you strip away the jargon of “LiDAR” and “neural networks,” what remains is the question: Who pays when the car misses the stop sign? The worker pays. The worker pays with his job, with his reputation, with his life. The owner pays with a fine, which is a cost of doing business, like coal dust in the lungs.
I have seen the “vision-only” approach in the coal mines. They called it “natural ventilation.” They said the air would flow naturally, as it had for generations. They did not install the fans because the fans cost money. They said the “natural” way was the most efficient. And then the explosions came. And the bodies came. And the “natural” way was gone.
You speak of production. I speak of survival. You speak of ingenuity. I speak of accountability. If you believe that the market will provide safety, you are wrong. The market provides profit. Safety is a gift from the state, or a demand from the union. It is never a gift from the owner.
So, I ask you this: When the car fails, who bleeds? Is it the shareholder, watching the stock ticker? Or is it the woman in Fremont, tightening the bolt, hoping the sensor sees what her eyes cannot?
The Verdict
Where They Agree
Jean-Baptiste Say and Mother Jones find unexpected common ground in their analysis of the law’s intent. Both explicitly argue that the legislation functions as a protectionist measure designed to shield established market incumbents from a disruptive, lower-cost competitor. Say decries it as a “protectionist tariff disguised as public welfare,” while Jones frames it as a monopoly tool where “the California legislators have done what every monopolist dreams of: they have asked the state to write the rules of the game so that only the house can play.” This shared diagnosis is significant because it cuts against the law’s stated purpose of public safety, suggesting both debaters see the political process as susceptible to capture by corporate interests, regardless of their differing views on market efficacy.
both agree that mandating a specific technology like LiDAR directly translates into higher consumer costs, making the resulting service a “luxury for the few” (Say) or a “profitable version of the future” only for the rich (Jones). This shared economic prediction undermines any pretense that the regulation is a purely neutral safety check; both frame it as an economic intervention with distributive consequences, privileging capital-intensive production. Their agreement reveals that the core dispute is not whether the law picks winners and losers, but which set of winners and losers is preferable and why.
Where They Fundamentally Disagree
The core dispute concerns the fundamental relationship between market competition and safety production. Empirically, they disagree on whether an unregulated market has a natural incentive to produce the safest possible product. Say asserts with high confidence that “safety is best achieved through competition,” arguing that the profit motive and threat of market failure are sufficient drivers. Jones asserts with equal confidence the contrary empirical claim: “The market provides profit. Safety is a gift from the state, or a demand from the union.” Normatively, this factual dispute is underpinned by a values conflict over accountability. Say’s framework values the long-term generative process of market discovery, trusting it to eventually solve for safety through iterative production. Jones’s framework values immediate, guaranteed protection for the vulnerable, believing the human cost of market error is too high to use it as a testing ground.
A second irreducible disagreement is over the nature of the technology itself and what constitutes sufficient proof of safety. The empirical question is whether a vision-only system can ever achieve a level of reliability equivalent to or greater than a system with LiDAR redundancy. Say’s steelman position is that the proof is in the deployment - millions of miles of real-world data constitute a valid and superior form of testing that mimics human perception, and the system should be allowed to prove itself on the road. Jones’s steelman position is that this real-world testing is tantamount to human experimentation, and that redundancy is a non-negotiable engineering prerequisite for a mission-critical system, just as ventilation was in mines. Normatively, this is a clash between the value of iterative innovation and the value of precautionary principle; between trusting the engineer’s data and trusting the legislator’s standard.
Hidden Assumptions
- Jean-Baptiste Say: Assumes that a catastrophic failure of a Tesla “Full Self-Driving” system would lead to its swift market rejection, rather than a drawn-out legal and public relations battle that a large corporation could survive. If this is false, and the market does not efficiently punish failure due to brand loyalty or complex liability, then the market-based safety incentive evaporates.
- Jean-Baptiste Say: Assumes that the cost of a LiDAR sensor is a pure rent-seeking cost with zero marginal safety benefit. If LiDAR provides a non-redundant safety benefit in edge cases (e.g., adverse weather) that cameras cannot match, then the mandate is not purely protectionist but has a genuine safety rationale.
- Jones]-style: Assumes that a state-mandated standard is inherently immune to the same cost-cutting pressures that corrupt market-based safety and will be perfectly enforced. If regulators are underfunded, subject to regulatory capture, or slow to adapt to new risks, the mandated “safety” can become a false credential that breeds complacency.
- Jones]-style: Assumes that the primary beneficiary of a cheaper, camera-only system is the corporate owner (e.g., Tesla’s margins) rather than the consumer (e.g., lower ride-hailing costs). If the cost savings were fully passed through, making the service accessible to a wider population, her argument would have to weigh the welfare gain of accessibility against the potential safety trade-off.
Confidence vs Evidence
- Jean-Baptiste Say: The claim that “The consumer pays more for a service that is less adaptable” - tagged with implied high confidence but supported only by theoretical reasoning. No empirical evidence is provided to show that a LiDAR-mandated system is inherently “less adaptable” than a vision-based one; this is a contestable engineering assumption.
- Jones]-style: The claim that “The market, left to its own devices, will produce the cheapest product, and it will sell safety as a premium feature” - is a well-supported historical pattern in many industries (e.g., automotive safety features). Her confidence here is justified by a strong precedent, though its direct application to the AV industry is still predictive.
What This Means For You
When evaluating coverage of this policy, your first question should be: what does the data actually say about the comparative safety performance of lidar-based versus vision-only autonomous systems in complex urban environments? Be deeply suspicious of anyone, whether a regulator or a CEO, who makes confident, sweeping claims about safety without presenting transparent, independently verified disengagement and incident rates. Your opinion on this law should change if clear, objective evidence emerges that one technological approach has a definitively superior safety record that justifies a mandate, or conversely, that the performance gap is negligible and the cost disparity is purely protectionist. Demand to see the collision avoidance rate in poor visibility conditions.