The Risk You Delegate Is the Risk You Create

NHTSA attributes roughly 94% of crashes to human error—distraction, impairment, fatigue, misjudgment. This number barely fluctuates. It is the baseline cost of humans operating vehicles, and it has been stable for decades. No institution is accountable for this. No regulator faces scrutiny when the number holds steady.

The Delegation Move

Rather than each individual choosing their own exposure to the new system—trying it, assessing it, adjusting—the risk gets delegated. "Someone should make sure this is safe." That someone becomes an institution: a regulatory body, a standards agency, an approval framework. The individual steps back. The institution steps in.

This feels like responsibility. It is its opposite.

The moment risk delegates to an institution, it transforms in kind. The regulator now holds concentrated accountability for any failure of the new system. If they approve FSD and a crash makes headlines, the approval is scrutinized. Careers end. Investigations launch. The cost of a wrong "yes" is visible, attributable, and catastrophic for the institution.

The cost of a wrong "no"—the 110 daily deaths that continue under the old baseline—is invisible. No one is accountable for the status quo. No regulator is investigated for the crashes that would not have occurred had the better system been deployed a year earlier. That cost is borne by strangers, distributed, unattributable, and structurally invisible to the institution making the decision.

So the institution optimizes rationally: minimize attributable novel failure. This directly maximizes unattributable ongoing harm. Not through malice, not through incompetence—through the geometry of how accountability concentrates when responsibility delegates.

The Loop

The institution now exists to manage the risk. Its funding, authority, and relevance depend on the risk continuing to require management. The more complex the approval process, the more necessary the institution becomes. The longer the delay, the more resources flow toward oversight rather than deployment. The institution doesn't conspire to perpetuate risk—it doesn't need to. Its survival structure does the work automatically.

Meanwhile, the individual who delegated the responsibility loses capacity. "Is this system safe?" becomes "has the regulator approved it?"—a different question entirely, one that substitutes institutional judgment for direct assessment. The individual's ability to evaluate risk atrophies because the delegation made it unnecessary. This increases dependence on the institution, which increases the institution's necessity, which increases its conservatism, which increases the delay.

And the individual's own driving—the actual source of 94% of crashes—never enters this loop. That risk was never delegated because no one thinks of their own driving as a risk to be managed by someone else. It sits outside the accountability structure entirely, invisible precisely because no institution holds it.

The loop closes: the act of delegating risk generates the conditions that sustain it. The institution requires the risk to justify its existence. The individual's abdication removes the direct engagement that would diminish the risk. The delegation, undertaken to reduce harm, becomes the mechanism that perpetuates it.

The Evidence

This structure is visible everywhere in the autonomous driving landscape.

Europe is years behind North America on deployment—not because the technology is less capable, but because fragmented regulatory frameworks multiply the delegation layers. Each jurisdiction creates its own institutional gate, each gate optimizes for its own accountability, and the aggregate delay compounds across borders. The body count of that delay is real, distributed, and attributed to no one.

Every FSD-involved crash triggers institutional response: investigations, probes, potential restrictions. The institution must respond—its legitimacy depends on visible action when attributable failure occurs. No equivalent response exists for the thousands of human-error crashes that occur in the same period. The asymmetry isn't bias. It's structural. The institution can only see what its accountability structure makes legible.

Tesla's data, for all its limitations, represents something the institutional framework cannot easily process: a direct, fleet-scale measurement of comparative risk. The institution's tools are incident review, standards compliance, and approval gates—none of which are designed to weigh aggregate displacement of baseline harm. The data speaks to a question the institution isn't structured to ask.

What This Reveals

The conventional framing positions institutions as the safeguard between dangerous technology and a vulnerable public. The structure operating beneath that framing is different: the institution's survival requires ongoing risk, the individual's delegation removes direct risk-reduction capacity, and the resulting delay produces the very harm both parties sought to avoid.

This isn't an argument against regulation as such. It's a structural observation: delegating risk to concentrated accountability structures converts a distributed, diminishing problem into a concentrated, self-sustaining one. The distributed version—millions of individuals choosing exposure to an improving system—shrinks as the system improves with each mile driven. The concentrated version—institutional gates that optimize for zero attributable failure—has no mechanism to shrink, because the cost of the status quo is never attributed. Restriction is a selective tax is the same densification under capacity-gap costume: the hold taxes those who accept the seal, and mostly those who enforce it, while the frontier continues at the pace of those who do not treat the seal as ground.

The risk you delegate is the risk you create. Not as paradox. As mechanism. The 40,000 annual deaths persist not because we lack a better system, but because the structure through which we collectively manage risk is itself the primary obstacle to reducing it. The climate problem registers only as perception is the same apparatus under planetary costume: solution layers whose funding and necessity require residual on a closed suite that an open multi-factor field will not finish.

This is how the mechanism works. What you do with it is yours.