15.8 C
New York
Sunday, March 17, 2024

Why Joe Biden's Free Covid Test Website Wasn't a Dumpster Fire

Another Nor’easter. How many of these snow cyclones can you take before you move west’er? It’s a song of ice or fire.

The Plain View

On December 21, President Joe Biden pulled an Oprah—antigen-style. He was touting the January launch of a website where every family in the country would be able to apply to receive four free rapid at-home Covid tests. While the announcement was an embarrassing turnaround from the day his press secretary Jen Psaki mocked the very idea of gifting the American people with those vital tests, the move was quickly lauded; millions of Americans were frustrated at their inability to score these at their local drugstores. They were also miffed at the cost, which sometimes bordered on price gouging.

But the big question was whether the feds could pull off a website to handle the inevitable scramble for a half a billion tests. After all, eight years earlier a different dot-gov site triggered one of the biggest tech debacles in history, when the nonfunctioning healthcare.gov almost sank Obamacare—and Obama.

But these days, a relatively new government unit, the United States Digital Service, is in place. Its new administrator, Mina Hsiang, actually served on the small team that rescued healthcare.gov for Obama. That site's meltdown—it typically crashed before a user could even do anything—exposed a long history of government haplessness. Locked into antediluvian IT protocols and a contracting system void of accountability, government tech’s default was usually failure. The USDS was formed in August 2014 to address that overall problem and apply the modern principles used to fix healthcare.gov. During the pandemic, the USDS became even more important, as citizens accessed more services online. Although Hsiang had left for private industry in 2018, “Coming back was a no-brainer,” she says via email.

Covidtests.gov would be a high-profile test of how effectively the federal technologists implemented the lessons from past mistakes.

The first lesson was applied even before the program was officially underway. At one time, a presidential announcement like that would have caused a mad scramble in the agencies involved. But hard and bloody experience has changed the way the executive branch works. This time, even before Biden made his public promise, the people charged with actually building the site had, as Hsiang says, “a seat at the table” and were able to shape expectations from the beginning. “We did a bunch of work to make sure that it was technically feasible before we decided how we were going to implement it,” says Natalie Kates, who is the Covid lead for USDS.

They decided that the project should be sited and built at the United States Postal Service, which not only had the national database of valid addresses, but would ultimately deliver the packages. When the Postal Service’s CIO, Pritha Mehra, learned about the project in December, she was given estimates that demand might peak at a million users an hour. Mehra, a 31-year veteran of the service, concluded that was a lowball prediction and multiplied the number by 20, striving for a fail-safe capability. “Think about it—free Covid tests,” she says. “Look at the numbers of people that are trying to buy them. And so we read 20 times the demand that had been projected, and I told my team that’s what we’re going to build to.” She had no problem recruiting that team. “This is a technologist’s dream, to be able to do this,” she says.

Mehra knew it would be a challenge to the service’s architecture, which involved a combination of its own data centers and outside cloud providers. Her team set up a system with triple redundancy, beefing up the architecture, separating the customer experience process from the order fulfillment, and caching data multiple times in the process. And doing endless load testing. “Believe me, there was a lot of work behind what seemed like a very simple site,” she says.

The whole project, from conception to a soft rollout on January 18 (a day earlier than announced), was completed in three weeks. The website itself was built by a relatively tiny team: three from USDS and around 15 in the Postal Service, a dramatic contrast to the human waves the government used to marshal on such projects. And it worked—a conclusion verified by the lack of outrage at its performance. Outside analytics indicate that more than 68 million people visited the site during its first week. Even more striking, at one point on the 18th, the site was handling 700,000 visitors at the same instant. By a large margin, this exceeded all traffic on other government sites combined. Best of all: The government now says that approximately 60 million people ordered tests.

I was one of those visitors, and I marveled at how easily the postal service cleared the admittedly low bar for a successful exchange on a government site. I typed in my name and address, and that was it. Within seconds, I got an email confirming my request.

Was the rollout perfect? Of course not. Some addresses didn’t register because they were listed as commercial buildings. The postal service tells me that it’s on the case and is handling those issues. Also, there have been issues with the response time at the phone number provided for those who don’t have web access. But don’t blame the geeks for that!

Of course, the registration is only step one in actually getting the tests delivered; Will Postmaster General Louis DeJoy’s much criticized operation get those tests to our homes? We’ll have to wait and see, but some people are reporting the arrival of their packages even before the promised window of seven to 12 days.

Success! If you wanted to be a Grinch, you’d note that four lousy tests won’t make a dent in what we really need to be safe during a pandemic that doesn’t seem to end. But at least this time around, we can’t blame the website.

Time Travel

In June 2014, I wrote about the team that rescued the Obamacare website, and their efforts to revamp the process after the first open enrollment period. The new version was dubbed Marketplace Light, and the principles the team followed are the same ones that made sure covidtests.gov would not be the disaster that healthcare.gov was.

The Marketplace Light crew got some of their work into HealthCare.gov during the open enrollment period. This included improvements to the registration screens that applicants confronted when they visited the site. These screens were crucial—if they were confusing or presented frustrating obstacles, some applicants would simply exit. On the early iterations of the site, this happened a lot, in many cases because of poor design. “For example,” one engineer says, “you had to choose a user name that had a special character such as an underscore or a dollar sign, which normally is a requirement applied to passwords, not to user names. So a lot of people were confused by that.” Worse, if someone chose a user name that already existed, the applicant would not find out until the end of the third page—at which point the system would make the applicant start over.

The Marketplace Light team fixed this by using something apparently in short supply during the original design of HealthCare.gov: common sense. “We just did the most basic simple thing that anybody would do,” a team member says. “Make one page, and use the email as the user name.”

Ask Me One Thing

Maury asks, “What would you have covered if it were not for the early venture into the hacker scene?”

Thanks for your interest in my career, Maury. I addressed this somewhat in a previous question about my early choices, sharing that at one point I had to choose between writing a book about hackers or one about cheesy nightclub singers. But I didn’t go down the sliding-doors path to alternate outcomes, so let me do that now. Undoubtedly, in some version of the multiverse I’m busking on a street corner, hoping to collect enough quarters for a burrito. And in another multiverse, my failure to cover hackers would have led to my writing about sports, politics, or some other subject. But upon some reflection, I think that even if I hadn’t been clued in to the rich journalistic opportunities that hackers provided, I probably would have wound up writing about tech anyway. It’s the major story of our time, and even someone dumb enough to write a book about crooners would have eventually figured that out. I might not have met Steve Jobs so early, but I bet you still would have seen my byline in the early days of WIRED.

You can submit questions to mail@wired.com. Write ASK LEVY in the subject line.

End Times Chronicle

Climate change is making the Russian permafrost less … perma. Hello to vast underground puddles, collapsing buildings, and the release of toxins previously buried in frozen tundra. How about massing some troops to stop that, Comrade Putin?

Last but Not Least

Is meat consumption overrated in its role in human development? Something to chew on.

Gilad Edelman interrogates Philip Rosedale, founder of Second Life, on what the metaverse will really be like.

You can’t get more old-school geeky than a discussion about what happens if a space elevator breaks.

Here are 14 alternatives to playing Wordle. I have a 15th alternative—read a book.

Don't miss future subscriber-only editions of this column. Subscribe to WIRED (50% off for Plaintext readers) today.

Related Articles

Latest Articles