The Failure of HealthCare.gov Exposes Silicon Valley Secrets

Y our jaw will drop when you see how the " tech surge " team fixed HealthCare.gov! Well, okay, now that I've got your click, I'll open with the lead that HealthCare.gov didn't really get fixed. And it's even worse than it appeared. Here's the story from my good friend Robert Kennedy, who was part of the team that made this critical, Web-based registration service work well enough that the enrollment period ended successfully, after starting from a complete debacle. You can see Kennedy's talk on his experience at www.youtube.com/watch?v=GLQyj-kBRdo. He shared some more details of the shockingly ill-constructed system with me over drinks one evening, but let me convey here his high-level points and my take on them. I'll just mention one or two anecdotes that motivate these. One you can see on the video, but it's just so good I have to repeat it here, as a teaser. Kennedy met with an employee of one of the site contractors who was in charge of software " security. " When he asked her which penetration threats she had prepared for, she responded that she didn't think about such things: she was in charge of " risk assessment. " In contractor terms, that meant the risk of government contract nonfulfillment. Her job, as she saw it, was to ensure that a software audit confirmed that the software was " secure " as defined in the contract. If you're a Silicon Valley engineer (or any other good engineer), your jaw really did just drop. To be fair, perhaps someone else was in charge of penetration testing, but the overall security objective seemed to be ensuring that the contractual terms were met, rather than that the system was actually difficult to hack. Then, there was the firewall in the wrong place. Oh wait, I can't write about that because the fix would be worse than the problem. I think I can say that something bad probably happened because all of the IP addresses were hard-coded and there was no DNS. That should be enough for you. But there were a lot of not just badly designed but guaranteed to fail subsystems, often at critical points. Here's a nice, top-level piece of brokenness: until the tech surge, no monitoring was in place for this nationwide critical distributed system. After the tech surge, there was an operations room for contractor representatives, each with his …