MELISSA BLOCK, HOST:
That million-dollar question referred to in Eric's story is more specifically a $400 million question. That's the total cost to create the online health insurance marketplace. Some 55 separate government contractors worked on it. And to computer programmer Rusty Foster, that number adds up to a problem.
RUSTY FOSTER: Once you hear sort of the details of how this was intended to be done and rolled out, everybody just kind of nods their head and goes, yeah, that's not going to work. You have a human services agency managing 55 different government contractors, trying to produce one complex piece of software. Like, that's crazy.
FOSTER: I can't think of any way that could work. I'm sort of amazed at how well it does work, actually, which is, you know, where it kind of - it could've been worse.
BLOCK: And that leads to the title of Foster's blog post for The New Yorker, "Healthcare.gov: It Could've Been Worse." I asked Foster where he thinks the project went wrong.
FOSTER: Overall, I think my sense is that they tried to do too much in too short a time. The project was managed the way that sort of government software projects have been managed. That's kind of what I talked about a lot in my blog post, was I went back to the FBI's rollout of their computer upgrade, which took more than 10 years and cost $1 billion, and there was one entire failed project involved. And this kind of went much the same way. So if anything, you know, the title "It Could Be Worse" means it took, you know, only a couple of years, and there is a site that exists.
BLOCK: Well, you mentioned that it was done in too short a time. And you mentioned a New York Times story that claims that the government was very slow in writing specifications for the contract, so that the biggest contractor didn't start actually coding software until this past spring. Would that be too short a window to do this the right way?
FOSTER: Yeah. I mean, I think absolutely, it was. Given the short window that they had, one approach they could have taken would have been more of a phased rollout, where perhaps the first launch of the site - I mean, they were on track to do this; the first launch of the site was just informational. Maybe the second phase of the site could've been you can make an account and sign in. And then after they had people making accounts and that was all working, they could've rolled out - now you can provide some information and we'll start looking up, you know, what kind of health insurance we can offer you.
BLOCK: Based on what you've read - I know you're on the outside - but based on what you've read, does it seem to you that there was adequate testing of this system before it was launched?
FOSTER: To an extent, there isn't any point in testing until you sort of feel like you're definitely finished. And I doubt many of the contractors felt like they were even finished. So, when somebody's giving you a deadline and saying it has to launch October 1st, come hell or high water, there isn't that much point running tests because they're going to push the code out anyway, and you're going to have a bunch of people using it regardless of whether your tests fail or not.
BLOCK: Well, the Obama administration has announced what they're calling the tech surge, right, the best and brightest being brought in to work on this.
BLOCK: Does having more bodies thrown at the problem really help? Is it a numerical thing? If you have more people, you can fix problems faster?
FOSTER: Not really. The conventional wisdom kind of comes from Brooks' - it's called Brooks' law, usually stated as adding programmers to a late software project makes it later. Because on the one hand, new programmers have to get up to speed on the system and some things, especially like this, is probably very complicated. And then at the same time, the people who are already working on the project have to sort of stop and train the new people in what they were doing. So there's a big slowdown across the board as soon as you try to add people to a project.
BLOCK: So what would the solution be if not more bodies?
FOSTER: That is hard to say. I mean, when you get yourself into kind of a bind with a software project, there aren't a whole lot of good options other than take the time and do the work.
BLOCK: Well, Rusty Foster, thanks for talking with us today.
FOSTER: Thank you for having me.
BLOCK: Rusty Foster is a computer programmer in Peaks Island, Maine. His blog post on The New Yorker website is called "Healthcare.gov: It Could Be Worse." Transcript provided by NPR, Copyright NPR.