500m Lines of Code Suggests Healthcare.gov Will Not Be Fixed Anytime Soon | Hacker News Reader

500m Lines of Code Suggests Healthcare.gov Will Not Be Fixed Anytime Soon

slate.com

Come on guys. At this point in time, I'm pretty surprised HN isn't completely outraged by now by this whole ordeal. We all in our right minds know a website like this shouldn't cost 1/100 of it's $90 million price tag nor be this bloated...

Even if this information isn't accurate, I'd argue the figures fall somewhere near what is being said, and still way more than it should for line count and price! That's simply a colossal, and pathetic failure, no matter which way or how you look at it. Why is everyone being so lax on all this? I don't want to say it....but if this was Bush's legislation and administration....yeah well I said it

It would be nice if they named the source. Perhaps it was some non-technical person who heard one of the junior developers lamenting about what seemed like a project with 500M lines of code? It would be really hard to believe that a website consisting of a registration form and a few API calls to get pricing data would ever get to that many lines of code. Also, keep in mind that the site was reported to have cost $600 million, but actually only ran $90 million. Given the political climate surround the legislation, it seems we must take any reporting on it with a whole bag of salt.

You still don't think even $90 million for a website that mostly performs CRUD operations is absurd?

I definitely agree that it seems like an awful lot. Any idea what a standard hourly bill-rate is for DC? In Arizona, the agency I work for bills at $125/hour.

However, not knowing the exact requirements for the project, we can only speculate as to what the true complexity is. Perhaps the insurance companies in each state did not provide well-documented REST APIs and you had hundreds of custom screen-scrapers pulling the data. Or, perhaps $30 million went to paying all of those insurance companies to build the required APIs? I cannot imagine how difficult and frustrating it would be to interface with those systems. Plus, all of the privacy and data safeguarding practices, and then arranging hosting, etc, etc, etc. Was there waste? Most likely. Was there waste to the tune of tens of millions? Couldn't say. However, to dismiss it as a simple CRUD app might be belying the complexity underneath.

I don't know man, while those possible sources of complexity and complications may be the case, to me they're still inexcusable. Many companies each year manage to successfully implement and release systems of ungodly complexity compared to whatever had to be dealt with to make some website like this.

All those sources of inefficiency are, to me, consequential of the fact that the people doing this work don't have proper incentives. $90M for a failed website that has to be redone? Well it's just tax dollars, let's suck in some more.

If any private company were to perform like this, they would be obliterated and their work would be done by people who are competent.

If you will recall, Twitter in its early days was essentially unusable due to high load. That was a private company that ultimately found its way.

However, that is really here nor there. I am always hesitant to make an armchair judgement about the execution of a development project without all of the facts because there are so many places where things can go pear-shaped. That is basically my point.

I wonder if the reporter has not mixed up lines of code with database records. Perhaps the database has a record for each and every American and 5 million of those records have incorrect data and need to be fixed up so that those people can use the website?

I don't buy it - I don't know where the 500M lines of code came from but it seems absurdly high. Those are 33 linux kernels and a change.

I am not sure if so many LoC is even possible to be produced at such short time.

It's complete guess work. The only mention of 500M lines of code is from the last paragraph of The Times article: "According to one specialist, the Web site contains about 500 million lines of software code."

This "one specialist" is obviously a fraud.

The specialist was talking to a reporter, who was writing for the general public, about how serious, complex, and justifiably expensive their work is. They might be including the linux kernels, dbms, etc in their line count.

Are they rewriting the entire linux kernel from scratch? There's no way that much code has to be re-written.

Who built this website? What stack is it running on?

Healthcare.gov is all opensource: https://www.healthcare.gov/developers/

Some chunks are made in node.js: https://github.com/STRML/Healthcare.gov-Marketplace

They've taken down the source on github, though it used to be all there. The server they're running it on is AkamaiGHost, and likely linux (although I've heard they're running some on it on windows)

It's an odd choice of technologies. For scalability I would have chosen something that scales across the cloud easy. I'd put my money on the different calls between organisations being at fault. There seems to be many API calls going on. It would take only one server to be slow to cause a cascading affect.

Reply on news.ycombinator.com