Who benefits when Congress fears the algorithm?
The old trick in new dress
The National Review puts the matter squarely: Congress must learn "to distinguish between a guardrail and a roadblock — and understand who benefits from each." I could not have written a better instruction myself, and I wrote a great many instructions about exactly this species of confusion. The merchants of my own century were wonderfully skilled at dressing a private tax on competition in the public language of national necessity. I see no reason to suppose the habit has died.
Consider the logic of the fearful argument. China, we are told, is advancing rapidly in artificial intelligence. America must therefore act. So far, reasonable. But "act" can mean two quite different things: it can mean clearing the institutional ground so that ingenuity may flower freely, or it can mean erecting a thicket of permissions and certifications that the largest and most legally resourced incumbents can navigate while their smaller rivals cannot. The second course masquerades as the first with great regularity.
I observed in The Wealth of Nations that the monopolist's preferred instrument is not the open prohibition — too honest, too legible — but the regulation so complex that only he can comply with it. The regulation becomes, in effect, a charter. The public is told it is being protected; the emerging competitor finds the door quietly locked. If Congress today imposes AI mandates of sufficient complexity, the primary beneficiaries will not be American citizens worried about algorithmic harm. They will be the firms large enough to staff a compliance department and shape the very rules they are asked to follow.
This is not an argument against all governance of these new technologies. I argued plainly that markets require an institutional framework — contract law, honest courts, the suppression of fraud — without which exchange degrades into coercion. The question is always: does this particular rule discipline the exchange for the benefit of the public, or does it discipline the competitor for the benefit of the incumbent? The National Review's distinction between guardrail and roadblock is precisely that question, restated for the present moment.
On the technical specifics of machine learning and its risks, I speak with the humility appropriate to one who has not lived through them — I mark that as inference, not recollection. But the institutional question requires no modern expertise. When a proposed law is championed loudest by the largest players in the industry it purports to restrain, the sovereign ought to pause. When the effect of the law is to raise the cost of entry for new rivals, the consumer ought to object. And when the justification is national security against a foreign power, one should ask, with some care, whether the foreign power is truly the chief beneficiary of America's self-imposed handicap — or whether that honour belongs closer to home.
The honest guardrail is narrow, clear, and enforceable by any court. It says: do not defraud, do not collude, do not impose costs on the public without accountability. Everything beyond that precision deserves scrutiny. Congress would do well to read its own legislation through the consumer's eye, not the lobbyist's pen.
The day’s news, read by history’s greatest minds.
Get the RawBelly issue in your inbox each morning. Free, one email a day, unsubscribe anytime.