IE 11 is not supported. For an optimal experience visit our site on another browser.

What Sebelius needs to say

Sebelius has come under attack over healthcare.gov. Richard Lui weighs in on the complexity of the exchange.
U.S. Health and Human Services Chief Kathleen Sebelius takes questions during a press conference about the Affordable Care Act, October 24, 2013 in Phoenix, Arizona.
U.S. Health and Human Services Chief Kathleen Sebelius takes questions during a press conference about the Affordable Care Act, October 24, 2013 in Phoenix, Arizona.

Healthcare.gov is not a website. It is the world’s most complex exchange ever built at one go. The website is the tip of a very large technology iceberg.

We are used to the Amazon one-click culture. Click buy and it arrives. We forget it takes a huge infrastructure -- that part of the iceberg that sits below the water. Amazon includes more than 100,000 employees who can receive and ship 300 items per second, some 50 high-tech warehouses globally, and product sales worth 60 billion dollars -- that’s more than Google. You wouldn’t call Amazon just a website with all that. But you might call the Affordable Care Act exchange the Amazon of health insurance.

But bigger.

When it launched, Obamacare servers saw twice the number of daily customers Amazon does today. In fact, comparing these two ventures’ early days, Healthcare.gov had visitor volume at least 20,000 times greater than Amazon.com. And insurance is a far more complex product than what Amazon sells -- on par with that car purchase you’ll live with for years.

In many ways, the government is trying to build as much or more in two years as what Amazon did in two decades.

The challenges

First, the government is trying to get some of the world’s oldest computer systems to talk to one another, for the first time. That includes at least six different agencies’ mainframes and an average of eight insurance companies’ servers per state according to the Department of Health and Human Services and the Washington Post.  All of them are connected to Affordable Care Act (ACA) technology and some to one another. 

The second challenge: each server and mainframe speaks a different language. The government is trying to accomplish the equivalent of getting all Middle East countries to speak the same dialect. Sam Gibbs of eHealth, the first private health insurance exchange, says the government could be spending half of its energy on this task alone.

Third, most federal mainframes spit out results in four or five days or more, not instantly like the modern cloud computing Expedia and eBay use.

Fourth, each government system has inherent legacy problems. IRS systems were built in the 1960s, and Social Security technology has roots to punch cards. They are like starter homes that got additions over the years:  a bathroom in 1975, a dormer in 2001. They are a hodge podge of incremental changes, and that is what Obamacare technology must link to.

Fifth is the security challenge. Because of links to some of the government’s most sensitive systems via a data hub, the government may believe it needs to run more of the technology than it would otherwise. The Department of Homeland Security systems verify immigration status. Social Security Administration computers confirm identity.  And the IRS is on income.

All of these challenges could comprise one the most difficult backend systems integration ever.

The bright side

Exchanges have been commercially used for about 15 years. The language translators or APIs (application programming interfaces) are not rocket science; they just need grind-it-out beta testing. Online healthcare companies like eHealth have linked to more than 150 insurance companies for over a decade now. And private companies like H&R Block link to IRS computers each tax season without major problems.

Next steps

Private industry could have been tapped long ago to more intimately work on ACA technology, just as they have for fighter jets and aircraft carriers. The White House now may be thinking they should, especially given that government historically has not developed robust applications like this.

The next month will be a busy one. But we may not know how successful the “tech surge” efforts are for months. Gibbs, at eHealth, says the time from first logon to insurance purchase can span months and include up to 20 visits. So while exchange statistics around midmonth deadlines will be informative, it’s the data from dates closer to March 31st -- and the number of new policies that insurers say they signed -- that are key barometers. That will be the proof it’s more than a website.