On May 15, 2026, Japan's Digital Agency published the latest follow-up to its roadmap for reviewing "analog regulations" — the roughly 10,000 statutory provisions that quietly require a human being to physically show up, look at something with their own eyes, or shuffle paper. The committee that ordered the review, the Digital Extraordinary Administrative Research Committee, spent 2021–2023 cataloguing rules that mandate visual inspection, on-site attendance, periodic in-person checks, full-time staffing, written postings, on-site document reading, and obsolete recording media. The goal is not to delete safety — it is to stop the law from prescribing how a check must be performed and let it specify what must be achieved.
This is unglamorous administrative housekeeping. It is also one of the most quietly important deregulatory programs in the democratic world, and it deserves more attention than it gets.
The strongest case for keeping a human in the loop
Start with the regulators' best argument, because it is a real one. Many of these rules exist for sound reasons. A statutory requirement that a qualified engineer personally inspect a bridge, a tunnel, or a hydroelectric dam encodes accountability: a named person is legally on the hook, and a trained eye can notice the anomaly no checklist anticipated. Paper postings and in-person filings create tamper-resistant, auditable records. When the failure mode is a collapsed overpass or a gas explosion, "we let a drone handle it" is not a reassuring sentence. Mandating the method is a crude but effective way to guarantee a floor of diligence, and stripping it out carelessly could open exactly the gaps critics fear.
Japan's program takes that concern seriously rather than waving it away — which is precisely why it works.
Regulating the goal, not the gadget
The flaw in method-mandates is not that humans are bad at inspection. It is that freezing one technology into statute guarantees the law will lag reality. The Digital Agency's own framework for visual-inspection rules is explicitly phased: Phase 1 is human visual inspection; Phase 2 permits remote inspection by cameras, sensors, and drones; Phase 3 layers AI image analysis on top for automated, more precise diagnosis. Crucially, the agency does not simply assert that drones are good enough. It runs technical-validation projects — drone bridge inspections, 3D point-cloud surveys of aging infrastructure, sensor monitoring of hydroelectric plants — to confirm safety and effectiveness before the underlying rule is rewritten.
That is what proportionate, evidence-based deregulation actually looks like. The accountability the steelman cares about is preserved; what changes is that a licensed inspector can now cover ten bridges a day from a control room instead of one by physically climbing it, and the marginal crack gets caught by an AI that never blinks. The safety outcome is held constant or improved; only the mandated means is liberalized.
The floppy-disk lesson
Nothing illustrates the cost of method-mandates better than Japan's war on the floppy disk. When the Digital Agency went looking, it found roughly 1,900 government procedures that still required submissions on floppy disks, CDs, or mini-discs — in fields as varied as utilities, mining, and aircraft manufacturing. By June 28, 2024, Digital Minister Taro Kono declared victory: the agency had scrapped 1,034 regulations governing physical recording media, leaving a single holdout tied to vehicle recycling, as The Register reported.
No one drafted those rules to be absurd. Each was a sensible specification of "submit data on durable media" — in 1995. The lesson is structural: the moment a regulation names a specific technology, it starts decaying, and it takes a dedicated cross-government campaign years later to dig it back out. A rule that had instead said "submit data in a machine-readable format" would never have needed rescuing.
A template other regulators should copy
The payoff is not merely tidier statute books. The committee's economic analysis projects meaningful efficiency gains — an August 2023 interim estimate put the savings from introducing drones, cameras, and sensors into inspection regimes at around ¥40 billion across public and private sectors, and the agency's dashboard attaches still larger cost-effectiveness figures to the visual-inspection category alone. Every hour an engineer does not spend driving to a remote substation is an hour redirected to higher-value work, and every digitized filing is a small subtraction from Japan's notorious administrative drag.
The deeper value is as a method. Regulators worldwide — including those now drafting AI, content-moderation, and platform rules — are under pressure to do something, and the easiest something is to mandate a specific tool: this filter, that audit format, this human-review step. Japan's program is a standing rebuttal. Specify the outcome you want — safe bridges, verifiable filings, accountable decisions — and let the regulated party reach it with whatever technology demonstrably works, subject to validation. That keeps the law durable as tools change, avoids quietly banning the better mousetrap, and spares some future minister another multi-year war against yesterday's mandate.
The critics are right that deregulation done carelessly creates risk. Japan's answer is not to deregulate carelessly. It is to validate first, phase the transition, and write rules around goals rather than gadgets. That is the proportionate path — and at 10,000 provisions and counting, it is also one of the largest live experiments in getting regulation out of technology's way without getting out of safety's.