First let me say that I like some of what Obamacare is about. Pre-existing condition coverage is a big winner in my book. But there are few things
that annoy me more than utter incompetence.
I was trying to determine what might be an appropriate phrase to describe the sheer magnitude of the failure. So I put together a few choice
descriptions offered by some of the earlier posters:
"disastrous technological failures"
"This is a tech / IT failure of comical proportions"
"Geez epic fail is right.""
"I find this hilarious"
So how about this: An epic disastrous technological F-A-I-L-U-R-E of completely hilarious proportions.
When I read the Ars
article, I didn't know whether to laugh at these poor slobs, or sit down and cry with them.
Overall I felt the article was written by an apologist who provided a rather thoughtful series of excuses for this gargantuan failure. However, he
also provided a nice description of some of the better examples of the contractor's and the government's utter incompetence:
The result of the headlong rush to October 1 was a system that had never been tested at anything like the load it experienced on its first day
of operation (if it was tested with loads at all). Those looking for a reason for the site's horrible performance on its first day had plenty of
things to choose from.
A few years ago, I used to build in-house performance testing systems. I pushed databases and business process management systems to their maximum
load. Then I used that data to create documents detailing loading and performance characteristics for several different classes of server and
I would get final releases of software that were being approved by QA that would fly apart at the seams with the least load. But product had to ship
by the end of the quarter so the company could make their guidance for Wall Street. Quality scrificed in pursuit of the artificial deadline.
It was a common inside joke that any customer that was crazy enough to go live with that version of the program was in for a wild ride. A month or so
later I would get a workable version of the program that could actually stand up to the load.
Now in the case of the Obamacare website, the developers made the artificial deadline of October 1st. Just like my old company made the
end-of-the-quarter deadline. But the product was beta quality at best, and certainly not ready for deployment.
First of all, there's the front-end site itself. The first page of the registration process (once you get to it) has 2,099 lines of HTML
For a fully operational system that's an obscene amount of (probably)redundant data to transfer. But this would be acceptable for a beta-test
product. However under load it ties up network bandwidth, server time, and disk throughput. And remember, all of this code has to be processed by
the client's browser. So, now this bloat of a system is also churning cycles on every one of the applicant's computer systems. This epic fail
actually started the instant the applicant selected the link to the government website. At that point, the failure had already started to seep into
the applicant's computer, like a piece of viscious malware.
Navigating the site once you get past registration is something of a cheese chase through the rat-maze. "It's like a bad, boring video game
where you try to grunt and hack your way through to the next step," one site user told Ars.
Okay, there's no excuse for a poorly designed "user experience." There are engineers whose speciality and whole career are based on designing
user-friendly interfaces. Not to mention the fact that this should have been done in the early design phase.
Once you get through all that, it’s not clear that it's going to do you any good. Underlying problems in the back-end code—including the
data hub built by QSSI—have been causing errors in determining whether individuals are eligible for subsidized plans under the program.
Here is some more information about the data hub built by QSSI:
In June, the Government Accountability Office, the nonpartisan auditing body that provides oversight reports to Congress, said that it was
still a crapshoot as to whether the system would work on time. This uncertainty persisted because the hub being built by QSSI still hadn't been
completely tested (the hub is responsible for making automated decisions about eligibility). While the policies to govern how the hub works—and how
various state systems were supposed to work—had been completed, there was still a lot of code to be written to make those policies into an actual
So, less that 4 months before the system goes live, and they're still writing some of the trickiest code in the whole project! If the project
manager for this fiasco worked for Darth Vader, he would have had some trouble breathing come October 1st.
Even without knowing all of the specific details of how this project came together, I can discern enough to see that this whole thing has a very
amateurish feel to it. It kind of reminds me of some code that a bunch of undergraduates would build for a class project.
This project failed along so many different paths in so many different dimensions that something occured to me: The probability of achieving an epic
failure of this magnitude has to be almost as low as the probability of pulling off a perfect implementation without any hitches at all. So that
leads me to one conclusion: It was sabotage.