The apparent death of California high-speed rail has led to a spate of stories asking, “What went wrong?” The answer to that question is a lot simpler than most people think: what went wrong is that the state even considered building the project in the first place.

Back in 1997, a graduate student at the University of California, Berkeley named David Levinson — who has since gone on to do a lot of incredible work on transportation access — asked whether high-speed rail would cost more or less than flying or driving. The state had not yet made cost estimates, so Levinson estimated that the line from San Francisco to Los Angeles would cost about $10 billion. Based on that, he found that high-speed rail would be “significantly more costly than expanding existing air service” and also more expensive than driving.

Just three years later, California’s first official estimate was that the SF-LA line would cost $20 billion. The state should have stopped right there. Yet, having made that estimate, the momentum was in place to continue even as estimates rose to $33 billion, then $68 billion, then $77 billion and more.

High-speed rail is presented as some kind of new technology race that America is in danger of losing if it doesn’t start building right away. In fact, American railroads began experimenting with high-speed trains back in the 1930s. Japan’s bullet trains date to 1964, 55 years ago. By that time, we had already surpassed them with jet airliners.

Today, air travel is far less expensive than train travel, with airfares averaging under 14 cents a passenger mile, barely more than a third of Amtrak fares even though Amtrak receives much bigger subsidies, per passenger mile, than the airlines. Racing to build a faster train would be like subsidizing the manufacture of new IBM Selectric typewriters because China developed a faster electric typewriter — no matter how fast, typewriters have been rendered obsolete by word processors.

The reason high-speed rail is expensive is that it requires a lot of precision infrastructure that is costly to build and costly to maintain. By comparison, airlines require very little infrastructure and while highway infrastructure is relatively inexpensive: for the cost of one mile of high-speed rail, California could build eight to ten miles of four-lane freeway.

Moreover, the growth in costs was totally predictable. Bent Flyvbjerg has shown that megaprojects like this just about always go way over the original projected cost because planners are guilty of optimism bias — they always make optimistic assumptions about each part of the project. Flyvbjerg went even further, saying that often planners weren’t just optimistic but engaged in “strategic misrepresentation” in order to get the projects approved.

California rail planners, for example, justified the high costs with unrealistically high ridership projections. Although the California corridor has a lower population that is less optimally arranged than the Boston-to-Washington corridor, California was predicting that its high-speed rail line would attract 55 million riders a year, fifteen times the number carried by Amtrak’s high-speed Acela. Based on this, the planners predicted the trains would collect so much revenue above their operating costs that private parties would gladly invest $7.5 billion in the project. Such private money never materialized.

Whether light rail, heavy rail, commuter rail, or high-speed rail, it turns out to almost always be true that construction and operating costs are higher, ridership and revenues are lower, and projects take longer to complete than the original projections — a combination of problems known as the planning fallacy. Moreover, maintenance costs are almost never considered since rail infrastructure wears out after about 25 to 30 years, by which time the original planners are retired and won’t be blamed when the systems start falling apart. This is why rail transit today is suffering far more from crumbling infrastructure than the nation’s highways.

Though these problems are repeated over and over, many cities and states today are planning more such urban and intercity rail projects. The state of Washington is thinking about starting a state high-speed rail authority. Cities like Austin and Durham want to build light rail even though transit ridership in those regions (and just about everywhere else) is declining.

The reason these projects go forward is the city or state hired some engineering firm to do a “feasibility study” — and then that firm spent part of its income from the study to lobby for the project, knowing that if it goes forward it is likely to get some of the future contracts. A plan to extend Portland’s light rail into Vancover, Washington died after it was revealed that the main lobbyist for the project was paid by the contractor who was hired to write the environmental impact statement.

Such revelations rarely come in time: Honolulu is building a rail line whose costs have tripled partly because the transit agency turned most project administration over to consultants — paying them $505,000 a year per person — who made more than 270 contract changes that added a half a billion dollars to the total costs.

The bottom line is that states and cities should not even ask whether urban or intercity passenger rail projects are feasible. I can tell you at the start that they are not: unless you are in Tokyo or Hong Kong, buses, cars, and planes are always superior to passenger rail. Once you start spending money on it, it is hard to stop. The only thing that stopped California is that it literally didn’t have any more money to spend.