Four decades ago, in 1977, a conspiracy began bubbling up from the basements of the vast network of hospitals belonging to the Veterans Administration. Across the country, software geeks and doctors were puzzling out how they could make medical care better with these new devices called personal computers. Working sometimes at night or in their spare time, they started to cobble together a system that helped doctors organize their prescriptions, their CAT scans and patient notes, and to share their experiences electronically to help improve care for veterans.

Within a few years, this band of altruistic docs and nerds—they called themselves "The Hardhats," and sometimes "the conspiracy"—had built something totally new, a system that would transform medicine. Today, the medical-data revolution is taken for granted, and electronic health records are a multibillion-dollar industry. Back then, the whole idea was a novelty, even a threat. The VA pioneers were years ahead of their time. Their project was innovative, entrepreneurial and public-spirited—all those things the government wasn't supposed to be.

Of course, the government tried to kill it.

Though the system has survived for decades, even topping the lists of the most effective and popular medical records systems, it's now on the verge of being eliminated: The secretary of what is now the Department of Veterans Affairs has already said he wants the agency to switch over to a commercial system. An official decision is scheduled for July 1. Throwing it out and starting over will cost $16 billion, according to one estimate.

What happened? The story of the VA's unique computer system—how the government actually managed to build a pioneering and effective medical data network, and then managed to neglect it to the point of irreparability—is emblematic of how politics can lead to the bungling of a vital, complex technology. As recently as last August, a Medscape survey of 15,000 physicians found that the VA system, called VistA, ranked as the most usable and useful medical records system, above hundreds of other commercial versions marketed by hotshot tech companies with powerful Washington lobbyists. Back in 2009, some of the architects of the Affordable Care Act saw VistA as a model for the transformation of American medical records and even floated giving it away to every doctor in America.

Today, VistA is a whipping boy for Congress; the VA's senior IT leadership and its overseers in the House and Senate are all sharpening their knives for the system, which they caricature as a scruffy old nag that fails the veterans riding on it. Big commercial companies are circling, each one putting forward its own proprietary technology as the answer to the VA's woes. The VA leadership seems to agree with them. "We need to move towards commercially tested products," VA Secretary David Shulkin told a congressional committee on March 7. "If somebody could explain to me why veterans benefit from VA being a good software developer, then maybe I'd change my mind."

You'd have to be a very brave VA administrator, and perhaps a foolhardy one, to keep VistA in 2017: The system's homegrown structure creates security and maintenance challenges; a huge amount of talent has fled the agency, and many Congress members are leery of it. Because it serves nearly 9 million veterans at 167 hospitals and 1,700 sites of care, however, the wrangling over VistA concerns much more than just another computer software system. The men and women who created and shaped VistA over the decades were pathfinders in efforts to use data to reshape the multi-trillion-dollar U.S. health care system. Much of what they've done continues to serve veterans well; it's an open question whether the Beltway solution to replacing VistA, and the billions that will be spent pursuing it, will result in a system that serves the VA—and the nation—as well in the long run.

What's clear, though, is that the whole story of how VistA was born, grew and slid into disrepair illustrates just how difficult it can be for the government to handle innovation in its midst.

YOU COULD SAY that VistA—which stands for the Veterans Information Systems and Technology Architecture—began as a giant hack.

Its birth occurred in 1977, far back in the era of paper medical records, with a pair of computer nerds from the National Bureau of Standards. Ted O'Neill and Marty Johnson had helped standardize a computer language, originally developed at Massachusetts General Hospital, called MUMPS, and the two men were hired by the VA to see whether MUMPS could be the basis of a new computer system connecting the VA's hospitals. Computerizing the one-on-one art of medical care seemed like a sacrilege at the time, but the VA, struggling with casualties of the Vietnam War, was underfunded, disorganized and needed all the help it could get.

O'Neill and Johnson began recruiting other techies to the effort, some of whom were already working in VA hospitals in places such as St. Petersburg, Florida; Lexington, Kentucky; and San Francisco. Though they were on an official mission, their approach—highly decentralized, with different teams trying things in various hospitals—ran against the grain of a big bureaucracy and aroused the suspicions of the central office. The project soon had the feeling of a conspiracy, something that nonconformists did in secret. They gave themselves an internal nickname—the Hardhats. People who followed the project recall being struck by just how idealistic it was. "This will sound a bit hokey, but they saw a way to improve health care at less cost than was being proposed in the central office," says Nancy Tomich, a writer who was covering VA health care at the time. As bureaucratic battles mounted, she says, "I remember how impressed I was by these dedicated people who put their personal welfare on the line."

In 1978, with personal computers just starting to appear in the homes of nerdy hobbyists, the Hardhats bought thousands of personal data processors and distributed them throughout the VA. Software geeks and physicians were soon exploring how patient care could be improved with these new devices. A scheduling system was built in Oklahoma City, while technicians in Columbia, Missouri, built a radiology program, and the Washington, D.C., VA's Hardhats worked on a cardiology program. In Silicon Valley, Steve Wozniak was building a computer in his garage that would overturn an industry; at the VA, these unsung rebels were doing something that was equally disruptive in its own way—and threatening to the VA's central computer office, which had a staff and budget hundreds of times greater and planned to service the data-processing needs of the VA hospitals and clinics by means of leased lines to regional mainframe centers. While the bureaucrats in the central office had their own empire, Tomich recalled, the Hardhats—some of them straight-looking guys with burr haircuts and pocket pen protectors, some scruffy, bearded dudes in T-shirts—were "in the field planting seeds, raising crops and things were blossoming,'' she says.

The Hardhats' key insight—and the reason VistA still has such dedicated fans today—was that the system would work well only if they brought doctors into the loop as they built their new tools. In fact, it would be best if doctors actually helped build them. Pre-specified computer design might work for an airplane or a ship, but a hospital had hundreds of thousands of variable processes. You needed a "co-evolutionary loop between those using the system and the system you provide them," says one of the early converts, mathematician Tom Munnecke, a polymathic entrepreneur and philanthropist who joined the VA hospital in Loma Linda, California, in 1978.

So rather than sitting in an office writing code and having the bureaucracy implement it, the computer scientists fanned out to doctors' offices to figure out what they needed. Doctors with a feel for technology jumped into the fray. "I got involved because it solved my problems," says Ross Fletcher, a cardiologist at the Washington, D.C., VA—where he is now chief of staff—since 1972. Working in close consultation with their clinical partners, sometimes coding at home at night or in their spare time, the computer experts built software that enabled doctors to legibly organize their prescriptions, CAT scans and patient notes, and to share their experiences electronically. Fletcher, who had studied a little computer science in college, worked with a software developer to help create an electronic EKG record. "The technical staff was embedded with clinical staff. I had lunch with the doctors, and in the parking lot in the morning we'd report what we'd done the night before," says Munnecke.

Munnecke, a leading Hardhat, remembers it as an exhilarating time. He used a PDP11/34 computer with 32 kilobytes of memory, and stored his programs, development work and his hospital's database on a 5-megabyte dish the size of a personal pizza. One day, Munnecke and a colleague, George Timson, sat in a restaurant and sketched out a circular diagram on a paper place mat, a design for what initially would be called the Decentralized Hospital Computer Program, and later VistA. MUMPs computer language was at the center of the diagram, surrounded by a kernel of programs used by everyone at the VA, with applications floating around the fringes like electrons in an atom. MUMPS was a ludicrously simple coding language that could run with limited memory and great speed on a low-powered computer. The architecture of VistA was open, modular and decentralized. All around the edges, the apps flourished through the cooperation of computer scientists and doctors.

"We didn't call it 'agile development,' but it was agile," says Howard Hayes, another VA IT veteran who served as CIO for the Indian Health Service, which adopted VistA. "Tight relationships between user and programmer, and sometimes they were one and the same." Instead of top-down goals and project sign-offs, teams of techies and doctors kept working to improve the system. "The developer did something, the user tried it, called him up or walked down the hall and says 'It really needs to do this.' The next day they had another build," says Hayes.

The VA's centralized computer department, which relied on contractors, was not amused. Its leadership wanted control, and they believed, with a position remarkably similar to current-day criticisms of the VA's IT work, that it made more sense to let the outside experts move the ball than have "garages" full of unconventional nerds and upstart doctors. The Hardhats were sharing records among doctors and hospitals. They were digitizing X-ray images. They were doing everything much less expensively and more successfully than the central office. They had to be stopped. In 1979, Ted O'Neill was fired (he drove a cab for a while, and later became a real estate agent). The main Hardhats office was shut down, and "pretty much everybody in the Washington part of the organization headed for the hills," says Munnecke.

But, remarkably, the project didn't die. There were still Hardhats dispersed throughout the VA system who had been hired locally and couldn't be sacked, and they carried on. A regular Monday morning telephone tree, which Munnecke created by patching together six people at a time from different parts of the country, each of whom would patch in others, kept them moving together.

The project took on the feeling of an insurgency, and the establishment began to retaliate. In Columbia, Missouri, Hardhat Bob Wickizer got back to his computer room from lunch one day to discover that his new desktop had been unplugged and put in a crate. In Washington, D.C., paper medical records were used to start a fire in the computer room that housed the code developed by Hardhats. "There were fires in the computer rooms. Sand in gas tanks. Not a pleasant fight," says Fletcher. "Fascinatingly enough, it occurred within a government institution."

Munnecke and his colleagues fretted about being fired for developing something that was far superior to anything that existed at the time. The participants called themselves "the conspiracy," and they developed kernels of essential software, which they shared via 300-baud modems (which transmit 300 words a minute) at 3 a.m. on Sunday mornings, or via disk packs the size of laundry baskets. "They went through a period of persecution," says Phillip Longman, who wrote about it in his book on the VA health system, The Best Care Anywhere. "And that was good, because it created bonding experiences."

Then one day in 1981, VA Chief Medical Director Donald Custis visited the Washington VA medical center and found administrators and practitioners using the unauthorized software. Astonished, he blurted out, "It looks like we have an underground railroad here," according to Munnecke, who was so tickled by the phrase that he had membership cards printed up with a microchip glued onto a train engine, and a title, "VA MUMPS Underground Railroad." Munnecke handed the cards to friends, colleagues and higher-ups he was trying to woo. And eventually, the rebels won over some leaders, including VA Administrator Robert Nimmo, who gave them a budget. The money allowed the applications developed piecemeal at hospitals across the country to be shared, recalls George Timson. By 1985, most hospitals were being updated regularly through a database management system Timson developed called FileMan. They all used the same structure of application building, which meant they were integrated; the mainframers had advocated a more traditional approach of buying commercial modules and lashing them together through interfaces.

"The combination of cheap computers and an efficient language allowed the VA to leapfrog over the existing technology," says Stephan Fihn, the VA's current director of analytics and business intelligence. "We competed to make it better," recalls Fletcher. "They'd holler at us because we were late on some project. … Late for what? There were no other options." And after beating back the mainframe people, Munnecke could crow with his foot planted on the chest of the enemy. "Every one of their systems is totally dependent on a specific vendor, incompatible with every other system they have developed," he said in a 1982 speech that sounds strangely familiar to students of current, commercial EHR systems, notorious for being "walled gardens" that have trouble sharing data with one another. "Every one of our systems is vendor-independent and compatible with every other of our systems."

New features could be added as quickly as they were developed in far-flung basements. Responding to doctors' requests, developers created tools that were good at organizing care for the millions of veterans with chronic illnesses. Doctors could find a patient's records easily and compare outcomes in patients who'd undergone different treatments, finding the best treatments in different settings and leading to improved guidelines for populationwide care. As it grew, VistA genuinely changed medicine: The Hardhats created databases that allowed researchers to aggregate clinical cases, which led to discoveries linking blood pressure to stroke and the arthritis medicine Vioxx to heart attacks, among others. They helped make VA clinics the best in the world to avoid amputation if you were a diabetic. They also gave the world the patient wristband, whose bar codes assured that the right patients got the right dose at the right time. VA patients in New Orleans were the only ones in the city who emerged from Hurricane Katrina in 2006 with their records—including medication lists—intact.

All that customization was a great joy to the doctors who helped create it, and who used it. But it also held the seeds of a monstrous problem—all those thousands of pieces of code, like all code, would need to be updated and integrated with new computer technology. Sometimes the original coder had moved on, and trying to update his or her work was like editing a book written decades before, in another language, by a dead author.

But such problems would take years to become acute. In the meantime, the system's popularity began to spread beyond the VA. In the 1980s, the Finnish health system and some German hospitals adopted VistA, and, later, hospitals in Jordan, India, Australia and Japan. In the U.S., the Indian Health Service essentially cloned and then adapted the system. Over the years there would be many suggestions that the Pentagon also use VistA—the DOD runs its own massive health care system, unconnected to the VA, for active-duty service members—but it always resisted. Munnecke says he tried to wire the Pentagon with VistA at least three times. "Each time, I was technically correct and politically incorrect," he says. The problem, in his view, was that the Pentagon didn't want to run on the same system as the VA. If it did, Congress would start asking why it had to pay for redundancies. "Why do you need two hospitals? Someone's going to get riffed. The person who cooperates most gets the least turf. It's all turf. If I sound bitter, it's because I've been beating my head against the wall so long." The conflict between VA and DOD, it turned out, would end up becoming VistA's fatal flaw.

WHEN KEN KIZER, who had run California's Medicaid program, took over the VA health care system in 1994, it was under widespread attack for poor access and quality of care, to the extent that some GOP leaders were calling for its privatization. The IT system also had problems—most of the programming had been focused on developing applications but not modernizing the core, Munnecke would say later. Kizer, who knew nothing about electronic health records, funded a feasibility study to install a commercial system but discovered that the VA's system was clearly superior to anything available. So he went in a different direction: He hired one of the original Hardhats, the brilliant Rob Kolodner, as his chief health informatics officer, and, armed with a $400 million annual budget, Kolodner oversaw the implementation of a bright new user interface and record-tracking software. Kizer also implemented quality performance measures and started coordinated care projects well ahead of the rest of health care, and his staff built a web service, which for the first time allowed providers to see a veteran's electronic records from anywhere in the country. By 1999, Kizer testified in Congress that the new system he'd built—it was rechristened VistA from Decentralized Hospital Computer Program—wasn't just working out for the VA, it could be the basis of a national commercial health IT system.

But the 2001 administration change was not kind to VistA. Kizer's creation cost a lot to support, and several members of Congress, concerned by IT spending at the VA, sought to enforce a law requiring that the IT system in each federal agency be run by a single CIO. In other words, they wanted VistA's work to be handled from a central office, a recurring theme in VistA's development. VistA's innovative approach, indeed its value to doctors, resulted from being brewed in small, decentralized units throughout the sprawling VA system. But whenever projects missed their deadlines, critics of the system would say it lacked accountability. "In the federal government, and in VA in particular, there are these cycles of decentralization and centralization, and you have to figure out where you are dropped into the cycle," says Gary Christopherson, who held various IT positions at the Veterans Health Administration, including CIO from 2000 to 2002. "It's not because there is necessarily a rationale for one or the other. You'd hear, 'They're out of control, we need to centralize.' Or, 'Centralization isn't working—we need to put power out to the field.' When I got there, we were in decentralization phase. There are always people in both neighborhoods."

Then too, there were few people in positions of power, particularly in Congress, who could really grasp the complex issues at play. "You have a subject that's vital to the fate of health care, the fate of the nation, but few people have the skill set to navigate it intelligently," says Longman. In 2000, Christopherson convened a White House discussion bringing together federal IT officials, doctors groups and private industry to standardize data so that different health systems would be able to communicate with one another. Everyone agreed it was a good idea. It was scrapped when the new administration took office and, in the way of new administrations, was determined to leave its own mark. "The discussion you hear today about standards, interoperability, information exchange, all those things, was agreed to back in 2000-2002 but still hasn't occurred," says Christopherson, who is now a sculptor in Wisconsin. "It's very sad, very tragic, very stupid."

In 2004, the Pentagon hired outside contractors to revise its medical-records system, implementing a clunky version of VistA known as AHLTA. The Pentagon's approach was strictly a top-down IT job, unlike the collaborative approach used with such success in building VistA, and it lacked finesse. Many military doctors considered AHLTA a disaster—unreliable, with frequent crashes and inaccessible data. Its initials, some said, stood for, "Aw hell, let's try again." The system ranked at the bottom of physician preference lists. At the top of the list? VistA. (The Pentagon system was like "running on sand," residents told Ross Fletcher, while VistA was "running on asphalt.")

Meanwhile, the enemies of VistA in the bureaucracy and industry were moving in. A 2005 Gartner study said the agency could save $345 million a year by consolidating the agency's health IT efforts within the central office. The agency was reluctant to fully centralize, but when a burglar stole a computer containing 26.5 million patient records from the home of a VistA engineer, the knives came out at a series of House Veterans Affairs Committee hearings. The security of veterans' records became an issue. Steve Buyer, the Indiana Republican who chaired the committee, was determined to recentralize. At one hearing, he called decentralization proponents "the gargoyles that defend bureaucracy and the old way of doing business."

In 2006, two important things happened to VistA: It won the Harvard Business School's coveted Innovations in American Government Award—given each year to promote creativity in the public sector—and its budget disappeared. Congress ordered that all IT work at VA report to the agency's chief information officer, creating new bureaucratic barriers to getting things done and putting power in the hands of an office whose priorities were different from those of IT workers in the field. "We became the only health care system in the world where the health care CIO didn't report to the CEO of health care," recalled a senior official. As funding for health IT development dried up, hundreds of VistA experts left to join the private sector, taking their coding memory, their Fingerspitzengefuhl, with them. The collaborative relationships between doctors and developers were over. Several blue-ribbon panels since have said that the reorganization, aimed at efficiency, smothered VistA innovation. "Modern management techniques killed it," says the former official. "We always wondered whether it was a plot to help the private vendors. But whether it was or not, it had that effect."

When Buyer left Congress in 2011, he took an assignment for the giant contractor McKesson—lobbying Congress for the company, a major producer of commercial EHR systems—on health IT and Veterans Affairs issues.

VISTA CONTINUED TO serve the VA, and well enough to continue to receive high user ratings in surveys. Of course, the world around it had changed a lot by 2009. Health IT had blossomed into an industry—big companies like Epic, Cerner, GE and Siemens were selling to big hospital systems, which appreciated their strengths at handling billing—an area where VistA lacked good applications, since the VA was both the provider of care and the agency that paid for it. But VistA was still well ahead of the industry standard for clinical care. It had things like computerized physician order entry, in which the physician electronically requests things like drug prescriptions and radiological scans. Its population health programs had produced a wealth of recommendations for the best treatment of chronic conditions. (MUMPS, VistA's language, was old but is also still the basis for major commercial EHRs, such as those produced by Epic, Allscripts and McKesson.) VistA was so far ahead of the pack, in fact, that in the year President Barack Obama was elected, Rep. Pete Stark (D-Calif.), the chairman of the Ways and Means Committee, had introduced a bill that would provide VistA for free to every hospital and doctor in the United States. Geoff Gerhardt, a member of Stark's staff, felt the idea needed tweaking and got the congressman to support an alternative that would give the Centers for Medicare and Medicaid Services money to incentivize doctors who "meaningfully used" EHRs—thereby giving birth to the term that defined Obama's EHR subsidy program, the HITECH Act.

The Obama administration included the HITECH subsidies in its $831 billion stimulus program. But Stark was the only big proponent of open-source software at the table, and his proposal to give away VistA to doctors en masse ran into trouble. Republicans and Democrats like Anna Eshoo, who represents Silicon Valley, were not particular fans. "There weren't a lot of members who felt strongly [in favor]," Gerhardt recalls, "and when you have Microsoft opposing it …" Instead of directing the administration to distribute VistA to doctors, the final language included a sort of "public option," whereby HHS would make available a version of the open-source software to doctors. It never did. Senior administration officials felt the government lacked the expertise to provide software directly.

And so, when $35 billion was teed up for doctors, VistA was not available to them. A few startups, like MedSphere, had adapted the software and used it to wire some hospitals and other medical practices, at a price that was pennies on the dollar compared with the bills for the big EHR systems, which had started out as administrative billing software. So nearly all of that money was funneled into purchases of commercial EHRs.

BACK AT THE VA, VistA was running into different kinds of problems. Roger Baker, who became the VA's CIO in 2009, was an Obama transition team member with a background in corporate IT. To deal with the perception of disorganization and cost overruns at VistA, Baker set up the Project Management Accountability System, which required each IT project, such as modifications of VistA's lab and prescription software, to be up and running within six months and capped each project at two years. The federal government loved the accountability system, declaring it a model government program. It used lots of record-keeping, monitoring and strictly enforced business rules to keep projects from spinning out of control and burning up truckloads of money. And it succeeded at that, but at the same time became so burdensome that it strangled initiative. "It's like having a building, refusing to properly maintain it, and then when it inevitably starts to fall apart you say, 'See, there is no reason to maintain this building. It's falling apart,'" says Fred Trotter, CEO of DocGraph and one of the country's leading health IT hackers.

Baker, to his credit, was committed to making VistA flourish again. "When people are passionate about a product, they want to keep working it," he told me. "They do outrageous amounts of work to make the product as good as possible. In the 1980s and 1990s, that happened with VistA. By the time I got there, the passion wasn't at the VA anymore." To reignite the original spirit of the VistA team, he created a sort of "X Prize," an initiative for a relatively inexpensive project that would bring in open-source hackers, including some former VistA programmers, to wrap VistA in more contemporary code. The Veterans Health Administration put out bids for that work in 2011.

For the VistA lovers, the idealists who still thought open-source technology was the logical system for medical records, this was perhaps the last best chance to restore a treasured common property. Munnecke was one of them. So were Longman and a techie colleague at the New America Foundation, Sascha Meinrath. They created a consortium and enrolled Red Hat, a federal contractor and leading open source developer, as well as several former Hardhats to help them shape the bid. They were sure they would get the job, although midway through the process, the entire Wisconsin congressional delegation—Epic, the big EHR maker, was based outside Madison—wrote Baker, urging him not to focus exclusively on open source developers. In the end, the bid went to another consortium led by well-known contractors. It created a nonprofit organization that serves as a forum for open-source EHR developers, but, while full of idealists, the organization is underfunded and relatively toothless.

The loss of that contract stings Meinrath, now a technology professor at Penn State University, because he's convinced that VistA's open-source technology could have been used across the health care system. "If we'd been able to do this, [the] [disaster] wouldn't have happened," he maintains. "We could have built a backbone record-keeping system that was standardized and extensible to all the systems that all the different states were using."

Baker tried to strengthen VistA by attempting, once again, to convince the Pentagon to incorporate the VA's software in a shared health-records project. Even entering the project, many VA officials were convinced it would fail because they sensed the DOD was not really interested. The climax, after $564 million in spending, came at a November 2012 meeting at which the VA and Pentagon leaders of the project presented slides to VA Secretary Eric Shinseki and Defense Secretary Leon Panetta. "If you've ever been on the receiving side of a full fusillade from Leon Panetta. … It was not a lot of fun," says a former VA official who attended the meeting. "If we could have gotten DOD in, we could have gotten the critical mass we needed to rebuild VistA." With the failure of the shared records project, VistA's future also seemed to slip away.

THE VA STILL runs on VistA, and IT teams are still working to improve the interface for clinicians and to improve connections between the Veterans Health Administration, the Pentagon and other VA programs, such as the one that enrolls vets in benefits. But there's a bull's-eye on the program's back. Since the failure of the integration effort, and the subsequent 2014 scheduling fiasco, in which VA officials in Phoenix falsified records to make the agency look better at timely treatment of veterans, congressional committees have repeatedly called VA and Defense officials to testify. The solons blast the bureaucrats over the scheduling system and shake their fists about the supposed lack of interoperability between Defense and the VA—though in fact, experts say, there is far better transfer of information between the VA and DOD than between any two other U.S. health care entities. And minds are made up about VistA. "It's like an old Buick that gets you from Point A to Point B. But wouldn't we rather be riding inside an air-conditioned new Cadillac?" the new Veterans Affairs Committee chairman, physician Phil Roe of Tennessee, said at a hearing last year.

An updated user interface for VistA, called the Enterprise Health Management Program, seems to be progressing, although when it was implemented at the VA in Hampton Roads, Virginia, in a March 2015 pilot, it caused a system crash that affected clinical services at the hospital for a month. Not enough good computer expertise was left at VistA to manage the task, according to one former senior official. With the dramatic growth of commercial EHRs, the skills and training and innovation had moved out of VistA and out of open source EHRs, says David Waltman, chief information strategy officer at the Veterans Health Administration. A blue-ribbon panel on the VA's future urged the agency to dump VistA. The membership of the panel included conservative-funded veterans groups and was led by leaders of hospital systems that have spent hundreds of millions to install Epic software. Not a single VistA expert was on the panel or consulted by it.

At a summer meeting of the open source EHR forum, the VA's then-chief information officer, Laverne Council, indicated it was time to move on.

"Technologists accept change and hate old stuff," she said. VistA had become a victim of the factors that led to its strength. Its distributed development resulted in a thousand flowers blooming, each one a different species, and in the foggy ruins of time no one could identify the people who wrote the code or how it could be amended. Managing all of those applications across a 167-hospital national system, some said, was a nightmare—more than half of the VA's $4 billion annual IT budget goes to maintaining existing software. Under its current tentative plans, the VA will continue to update VistA's interface through 2018. After that, the VA is likely to swap out VistA for a commercial EHR. The speculation is that bid would go to Leidos and Cerner—if their current partnership with the DOD continues to move along on schedule. Some believe the fix is in already, because a number of Leidos IT specialists already work inside the VA. At least two other big EHR vendors, Epic and Allscripts, say they'll also bid to take over the VA's EHR duties.

This is bitter fruit for many VistA fans. Some still say the system could be fixed for $200 million a year—the cost of a medium-sized hospital system's EHR installation. "I don't know if there even is an EHR out there with data comparable to the longitudinal data that VistA has about veterans, and we certainly do not want to throw that data out if a new EHR were to be used," says Nancy Anthracite, a Hardhat and an infectious-disease physician.

Replacing VistA could be a colossal task—the military is spending $4.3 billion to switch to a Cerner EHR by the end of 2022, and Baker has estimated it could cost the VA $16 billion to rip and replace VistA. To maintain the functions that doctors like about it would require intricately lacing together new and existing software. Pitching all the good old stuff because it's too old and complex to integrate with a new EHR would be tremendously wasteful and frustrating to doctors. A decision on whether to switch to a commercial system has been set for July 1. Some, including former CIO Baker, think the VA should wait until at least September, at which point the IT world will have a clearer picture of how the military's implementation of a Cerner EHR is progressing.

"If they change the system and it becomes user unfriendly, that would be a major disaster," says Ross Fletcher. "I would hope that the creativity and leadership are not supplanted. Creativity is important."

OUTSIDE THE VA, the story of digitization of American medicine isn't any smoother than the story of VistA. It's true that an industry that used to run on paper records and folders is now mostly digital—a transformation accelerated by the $35 billion in subsidies offered as part of the stimulus package. But the promises of electronic records—improving care, improving health and making the industry more efficient—have only been partly borne out. Today the verdict is that electronic health records may have made medicine safer, but they've actually reduced, not improved, its efficiency.

Doctors now spend an average of $32,000 a year each on health IT installation and maintenance, and roughly 40 percent of their time working on the computer. Ask doctors what really bugs them and it doesn't take long to get to the software: clunky, vulnerable to hackers, and built by competing players without agreed-upon standards, so patient information often ends up locked in the records of competing software companies and hospital systems. The highly educated doctor delivering you cutting-edge medicine is stuck on a computer system you'd have been annoyed to find on your desktop back in 2005.

It's reasonable to see this, of course, as the growing pains of a new industry, much like computers themselves, which were originally a mishmash of competing systems until they standardized around just a couple. The alternative seems unrealistic: a centrally designed system that has gone through years of testing and improvement. But in one of Washington's strange ironies, the government really did develop such a system. And the government is about to spend billions of dollars to scrap and replace it.

The four-decade struggle over the VA's health IT system reveal some philosophical and logistical quandaries that arise in big health-care management systems, in government IT—and in the economy in general. Should the evolving pieces of a complex technological program run independently, or as an integrated, centralized whole? Should they be customized to user needs, or standardized for simplicity's sake? Does the answer lie in open-source or proprietary technology? Does passion or accountability promise the most success? Problem solving or planning? Creativity or control? And who should be trusted to come up with the correct decisions—experienced careerists or outside analysts? Or politicians? And finally, who best serves the veteran: the private sector or the government?

There are still VistA die-hards who think it could be revived were it not for politics and money. "It's hard to argue that VistA is an old racehorse when it still comes in first in the races," says DocGraph's Trotter, himself a Hardhat. For the most part, however, even the men and women who built and patched and rejuvenated VistA over the years have given up on it. It has been kicked around and neglected too long, like a former Kentucky Derby candidate chewing up pasture while awaiting a trip to the glue factory. Even so, a few dream that in a parallel universe, a place with less grandstanding about brave veterans and incompetent bureaucrats, a place with fewer lobbying dollars and more humility, the racehorse could keep on running.

"Perhaps this was a golden era of the kind of, 'Let's all make the world a better place by working together' attitude," Munnecke says. "It seems terribly naive today, but it was a driving force back then."