Japanese municipal websites go offline

Several Japanese municipal websites went dark on Wednesday, leaving residents in places from Ebetsu in Hokkaido to Dazaifu in Fukuoka staring at error messages instead of public notices. According to The Japan Times, the Internal Affairs Ministry asked prefectures to report on the disruption, while Chief Cabinet Secretary Minoru Kihara said the central government was collecting information to determine the cause. A supplier based in Gifu Prefecture that provides the underlying system is investigating, but officials said it was still unclear when service would be fully restored.

The incident is a familiar failure mode for public-sector IT: many separate organisations appear independent, but in practice sit on the same vendor stack, the same hosting arrangements and often the same procurement logic. When a shared platform fails—whether from a configuration error, a hosting outage, a DNS problem or a denial-of-service attack—the outage spreads horizontally across jurisdictions that have no operational connection except a contract. The Japan Times report notes that some local governments switched to posting updates on X and other social platforms, a workaround that is quick but also shifts official communication into privately run channels with their own moderation rules, outages and account-takeover risks.

The deeper vulnerability is not that a website can go down, but that the incentives in public procurement reward standardisation and low visible cost while underpricing resilience. A municipality buying a “proven” packaged system can treat uptime as the vendor’s problem—right up to the moment residents cannot access evacuation maps, tax forms or service announcements. In the private sector, a retailer that loses sales during an outage can quantify the loss immediately and fund redundancy accordingly. For a city hall, the cost is dispersed: citizens pay in time and uncertainty, while the budget line that would have paid for duplicated infrastructure is easy to cut.

Kihara said some municipalities were already relying on social media to communicate with residents. The websites were believed to be running on systems provided by the same Gifu-based company.

Tokyo Skytree reopened to visitors on Thursday after engineers traced Monday’s six-hour elevator stoppage to a damaged cable that supplied power and transmitted signals to the car. The tower’s operator and the elevator manufacturer said the cable was caught in a roller in a device meant to limit shaking, and that a fuse in the control panel melted and broke, according to The Japan Times. A cover has now been installed over the roller as a preventive measure.

The episode illustrates how “smart” systems typically fail: not through exotic software bugs, but through ordinary mechanical interactions that cascade into a full stop. Skytree’s elevator had the sensors and control logic to detect abnormal conditions, yet that intelligence does not move passengers when the physical system loses power or signal integrity. Once the car is immobilised, the most important design question is no longer optimisation but evacuation: what procedures exist, what equipment is available, and how much of the plan depends on other components functioning at the same time.

The operator’s explanation also shows how safety features can become new failure points. The cable was caught in a mechanism intended to stabilise the car during shaking—an accessory whose job is to reduce risk under stress conditions, but which introduced a new pinch point for a critical line. Engineers cited the cable’s twist direction and strong winds on the day as possible contributors, a reminder that edge cases are often the ones that matter in public infrastructure: rare combinations of weather, wear and motion that do not show up in routine operation.

Responsibility, meanwhile, is divided by design. The operator runs the site, the manufacturer supplies the elevator, and the maintenance regime sits somewhere in between, often subcontracted and governed by service-level terms rather than clear personal accountability. In this case, the fix was a cover over a roller.

The cable was damaged after being caught in the roller of a shaking-control device, and the tower was closed for several days before reopening.