t was an ambitious federal technology project, tying together a string of government databases in one Web portal. But the launch was a giant mess: contract missteps, broken deadlines, cost overruns. Users complained the site was balky and unworkable.

That was the rocky rollout last year of SAM.gov, a General Services Administration project intended to combine nine different contracting databases. The site was delayed two months and, even then, performance was so poor that it had to be taken down for repairs. In March, the GSA apologized after finding a flaw that could have exposed users' personal information.

Another federal website, the spectacularly flawed HealthCare.gov, has created a political crisis for the Obama administration. But as SAM.gov and other projects demonstrate, technology failures have become the rule in the federal government, not the exception.

Websites crash, attempts to modernize systems founder and military systems costing hundreds of millions are abandoned before ever being used.

The Obama administration has tried to confront the problem, appointing top technology officers who scrapped and consolidated some flagging projects and pushed for more agile procedures. But the reforms have been modest.

Last month, Obama acknowledged that the administration had not been able to fix a procurement system that he said was "just generally not very efficient."

"In fact, there's probably no bigger gap between the private sector and the public sector than IT," he said.

Experts point out that troubled projects aren't unusual in either. But the government's problems, involving taxpayer money, are pervasive and add up to billions in waste. Washington will spend more than $76 billion this year on information technology. A federal report in January found that 700 projects, accounting for $12.5 billion, were in trouble.

"Clearly the federal IT system is broken," said Michael Krigsman, an industry analyst who is often hired to analyze project failures. The problems seen with HealthCare.gov are not unique, he said.

"There are whole strings of failed IT projects where everybody points fingers at everybody else, and there's no one who ultimately takes responsibility."

The problems start with a byzantine contracting system that matches up poorly with the fast-evolving information world.

In the time it takes federal agencies to figure out what they need and get Congress to appropriate money, the technology can become obsolete or the costs increase dramatically.

The appropriations process, said one tech industry lobbyist, is stuck in a "horse and buggy" era.

"It's almost impossible for the government to find what it wants and get Congress and the (Office of Management and Budget) to buy that thing in a timely fashion," added Trey Hodgkins, senior at the Information Technology Industry Council, a trade group.

Another problem: Years of budget cuts and outsourcing have left many federal agencies short on technology talent. They are often slow in defining what they need at the start of a project, then unable to manage trouble when it happens.

A year ago, the Air Force scrapped a system that was supposed to improve supply management. The project had run for seven years at a cost of more than $1 billion. Afterward, a review team found it had suffered from mismanagement, a churn in directors and "simply an underestimation of the sophistication needed in tackling the enormous and complex effort."

In the case of SAM.gov, or System for Award Management, the GSA in 2008 began to consolidate databases into one big Web portal for use by private companies looking for government business as well as the government and civilian employees who hand out those contracts. The agency chose IBM Federal as the main contractor, giving the company $74.4 million over eight years.

But the GSA seriously underestimated costs. The agency thought IBM was providing the computers and software to host the website, but later realized officials had left that out of the contract, according to a report last year by the Government Accountability Office. They had to pay IBM another $38 million.

Meanwhile, Congress was slow to come up with extra money, the GAO found, leading to further delays and still-higher expenses to keep the old systems running.

Just as with HealthCare.gov, the SAM system proved to be buggy in late testing; the GSA delayed the launch, but that wasn't long enough to fix the problems.

In a statement, the GSA said the system was much improved, with fewer errors and shorter wait times. IBM said, "The system has been up and running successfully for the past year and IBM continues to support and innovate as part of the ongoing contract."

People from the technology world who have worked in the administration say the contracting process is inevitably tilted toward a few giant companies that can assemble experts in writing proposals - and who often file protests when things don't go their way.

"It's people with the best lawyers, not people with the best programmers," said Clay Johnson, chief executive of the nonprofit Department of Better Technology. "When your legal team becomes part of your sales force, you know you've got a problem."

Inside the agency building HealthCare.gov, the problems soon became obvious. The design of the website was supposed to be locked down by early 2012, one of the milestones that "must be completed to ensure the (exchange) will be ready," according to a contract document. But requirements kept shifting.

A consultant report from this spring said the design was "still presumed to be open" with six months to go, leading to "materially higher risk of system instability."

"It's like trying to build an airplane while you're flying it," said Dan Schuyler, former director of technology for the Utah Health Insurance Exchange and now a director with Leavitt Partners, a healthcare consulting firm. The Medicare and Medicaid agency that oversaw HealthCare.gov didn't have the capacity to run the project and tried to "build it from the ground up" after the law was passed, he said.

Daniel Goure, vice president of the Lexington Institute, a nonprofit policy research group based in Arlington, Va., said the health care site was "one you could have predicted would fail."

"Too many requirements, too many players, government trying to be in charge, starting too late - they were guaranteed not to make it on time," he said. "How come nobody told the government, you are riding for an incredible fall on this one?"