The saga of has been a symphony of government inefficiency. The effort, directly overseen by the IT department of the Centers for Medicare and Medicaid Services, involved no fewer than 55 contractors. The process was thick with lawyers and political interference. In violation of current best practices in the software world, the code was kept almost entirely secret; other engineers weren’t able to point out its flaws, and it wasn’t tested rigorously enough. The Obama administration has been assailed for not calling in Silicon Valley’s top minds to collaborate, but that misses the fundamental problem: The best coders in the Valley would’ve never agreed to work under such deadening, unpleasant conditions.

The flawed product poses real dangers to the law’s success. The front-end shopping experience—which is what people mean by “the website”—is busted, but there are also persistent errors in the systems that authenticate users, determine eligibility for subsidies or for Medicaid, and hand off purchase information to insurers. Obamacare’s coverage expansion sits atop a complex digital infrastructure, and that infrastructure is failing.

For all its deficiencies, isn’t the worst disaster a government has experienced on a major IT project. That distinction belongs to the U.K.’s endeavor to create an electronic medical records system for its National Health Service. The effort, which began in 2002, tore through about $10 billion before the government admitted it simply couldn’t be salvaged. In an editorial at the time, the liberal Guardian newspaper declared, “The government is an inept purchaser of private services: indecisive, ponderous, overambitious, and wasteful. Mass centralisation does not reduce costs, but it kills flexibility.”

“NHS just got tremendously burned by the rollout of electronic health record systems,” says Fred Trotter, the founder of health-care software developer Not Only Development and the co-author of Hacking Healthcare. “It’s probably the biggest catastrophe before They also had a big-bang, all-at-once, one-big-vendor strategy. That did not work.”