Did Open Source Software Bring Down Healthcare.gov?
No, almost certainly not.
A lot of news articles about Healthcare.gov have mentioned Jekyll, an open source tool for generating database-free websites.
I’ve asked people who know Jekyll well and the consensus is that the problems lie elsewhere.
Jekyll simply generates HTML files. Jekyll doesn’t create slow database connections, put excessive load on the server or add too many external files. Jekyll was just a tool for the designers to use.
In other words, Healthcare.gov may have struggled for a variety of reasons from the database, the servers, bad management, bad workmanship, or excessive traffic, but none of these were the fault of Jekyll.
This Slate article seems to pinpoint connections to the Oracle database as a primary source of errors. This article produces a screenshot of problems related to Oracle Access manager.
There’s something mysterious about the whole Marketplace area of the site that interacts with the Oracle database(s).
The source code for Healthcare.gov is hosted on Github at https://github.com/CMSgov/healthcare.gov/. The process isn’t perfect, but it is working and people are finding and fixing issues.
From Github you can download and test the Healthcare.gov site. Or, some of it …
The Marketplace area of the site has been removed. It was the Marketplace section that was apparently causing problems for visitors. People were able to browse the site successfully, but problems only occurred when trying to login
According to Reddit:
“The marketplace (closed source) and the website frontend (open source) were built in connected but separate processes by two different entities. The code for the marketplace … was never published.”
If the Marketplace code had been open sourced and placed on Github, maybe these problems would have been spotted.
Maybe the lesson of healthcare.gov is that not enough open source was used.
Hopefully this doesn’t give open source a black eye. Seems to me the problems are the 100 bajillion legacy systems, like the IRS and SSA, that the system has to pull data from.
I did notice, though, that the Github code was all removed. There were a few participants a little steamed about this – apparently there was no advance notice.
Yes, you’re right Dan. It does seem to have been a very tentative, IT-manager approach to using open source and certainly now with errors.
However, that still seems to have been the most successful side of the project.
True – in the sense that the lifeboats on the Titanic all held water…
Java is open source, you idiot.