Here is an appropriate TED Talk for today. Enjoy!
Thursday, November 28, 2013
Monday, November 4, 2013
What Went Wrong at healthcare.gov
Until healthcare.gov is fixed, the Obama administration is being very careful what they say about why the website failed. There will be plenty of time to assign blame after the website is working, but that is not stopping critics from painting the worse picture possible. The critics focus on the testimony from the website contractors, CGI and QSSI, that the system was extremely complex and only two weeks were scheduled for end-to-end testing, but no one raised a red flag. So they conclude that everyone involved must have been incompetent.
Before rushing to judgment, we need to take a closer look. What we call the healthcare.gov website is actually three major systems, the user interface, the data hub and the marketplace. The user interface, what you see in your web browser, was working three months before the October launch. It was developed in just 90 days by a small company called Development Seed. They used the open source approach favored by today's most technically saavy startups. In early July, Obama demoed the user interface in this video. The video showed that the website provided a clean and simple user interface where people could get background information before they bought insurance in October. The user interface was considered a tremendous success and may have given Obama the impression that the rest of website rollout would be equally successful, and he could focus on other things, like the threat of a government shutdown and the debt default.
To actually buy insurance, healthcare.gov had to get data from hundreds of insurance providers and several federal agencies like the IRS and Social Security. Many of the existing databases were build on older technology and presented special integration problems. On top of all that, data access had to be in realtime while the user waited. The network which provided this access, called the Data Hub, was the most complex part of the system and probably received the most management attention because of all the players involved. The use of the Data Hub had to be well documented so the state health exchanges independent from healthcare.gov could access the same data. The complex job of developing the Data Hub was performed by QSSI and it worked when healthcare.gov launched. If it didn't work, none of the independent state exchanges could function.
What actually failed at healthcare.gov was the marketplace, the part which connected the user interface to the Data Hub. The two parts of the marketplace QSSI's EIDM and CGI's FFM were proven systems from other CMS programs. What broke these systems was the unexpected large number of users shopping for insurance. Users first had to register with the EIDM. so it broke first. QSSI was able to fix the EIDM in a few days, but then the overload problem passed to the FFM which has performed sporadically. More testing would not have revealed the problem if the tests were conducted at the anticipated maximums. They only discovered the true maximum number of users after they went live.
Before putting the blame on CGI, it should be noted that CGI also did the health exchanges for six of the states not on healthcare.gov, including the much praise exchanges of Kentucky and California. The FFM could have worked well if the number of users were in the expected range. Since CGI won the FFM contract, many more states decided to use healthcare.gov. Once the number of users passes a certain threshold, the FFM would have to be redesigned. CMS could have allowed for the possibility of much more users just to be safe, but that could have significantly increased the cost and deplete the reserves in their budget. They needed sufficient reserves to cover unexpected problems after the launch that could have happened anywhere besides the FFM. Given the likelihood of Congress allocating more money, CMS management had to take reasonable risks. Balancing risks is what all management does and sometimes things don't work out.
Before rushing to judgment, we need to take a closer look. What we call the healthcare.gov website is actually three major systems, the user interface, the data hub and the marketplace. The user interface, what you see in your web browser, was working three months before the October launch. It was developed in just 90 days by a small company called Development Seed. They used the open source approach favored by today's most technically saavy startups. In early July, Obama demoed the user interface in this video. The video showed that the website provided a clean and simple user interface where people could get background information before they bought insurance in October. The user interface was considered a tremendous success and may have given Obama the impression that the rest of website rollout would be equally successful, and he could focus on other things, like the threat of a government shutdown and the debt default.
To actually buy insurance, healthcare.gov had to get data from hundreds of insurance providers and several federal agencies like the IRS and Social Security. Many of the existing databases were build on older technology and presented special integration problems. On top of all that, data access had to be in realtime while the user waited. The network which provided this access, called the Data Hub, was the most complex part of the system and probably received the most management attention because of all the players involved. The use of the Data Hub had to be well documented so the state health exchanges independent from healthcare.gov could access the same data. The complex job of developing the Data Hub was performed by QSSI and it worked when healthcare.gov launched. If it didn't work, none of the independent state exchanges could function.
What actually failed at healthcare.gov was the marketplace, the part which connected the user interface to the Data Hub. The two parts of the marketplace QSSI's EIDM and CGI's FFM were proven systems from other CMS programs. What broke these systems was the unexpected large number of users shopping for insurance. Users first had to register with the EIDM. so it broke first. QSSI was able to fix the EIDM in a few days, but then the overload problem passed to the FFM which has performed sporadically. More testing would not have revealed the problem if the tests were conducted at the anticipated maximums. They only discovered the true maximum number of users after they went live.
Before putting the blame on CGI, it should be noted that CGI also did the health exchanges for six of the states not on healthcare.gov, including the much praise exchanges of Kentucky and California. The FFM could have worked well if the number of users were in the expected range. Since CGI won the FFM contract, many more states decided to use healthcare.gov. Once the number of users passes a certain threshold, the FFM would have to be redesigned. CMS could have allowed for the possibility of much more users just to be safe, but that could have significantly increased the cost and deplete the reserves in their budget. They needed sufficient reserves to cover unexpected problems after the launch that could have happened anywhere besides the FFM. Given the likelihood of Congress allocating more money, CMS management had to take reasonable risks. Balancing risks is what all management does and sometimes things don't work out.
Subscribe to:
Comments (Atom)