Why the Obamacare Website Sucks | Hacker News Reader

Why the Obamacare Website Sucks

tbray.org

6 points__chrismc6 comments

Not impressed. E.g. his knock on the Strategic Defense Initiative ("Star Wars" per its detractors) is on a architectural strawman constructed by its enemies, the people involved planned on sane, somewhat independent defense in depth systems, where one failure of any sort would hopefully be compensated for by the other systems. E.g. "Brilliant Pebbles" were entirely unnetworked except for the necessary "GO!" signal, each would look at its view of the the attack and decide which booster it should try to take out.

And it ignores its stunning political success: the Soviets, after Brezhnev pretty much bankrupted the country with a return to Stalinist repression and all the money spent on the military, especially the 3 complete armies of armor on down supplied to North Vietnam (one used up piecemeal, one destroyed in the first post-"peace" attack, the final succeeded because the Democrats stopped supplying the South with ammo), and the Strategic Rocket Forces.

The latter of which SDI was going to entirely obsolete in one "generation"; faced with the expense of replacing all that investment, and the loss of first strike capability (you have to count on your enemies defenses working somewhat, and you can't pick which warheads make it), threw in the towel. We won the decades long Protracted Conflict/Cold War "without firing a shot", one of the greatest diplomatic successes in history. But it's an Idée fixe among the ignorant like Bray that it was an impossible failure, heck, he doesn't even consider people like me to be sane, "the foaming-at-the-mouth right wing".

Oh, yeah, the embedded systems contractors who do this sort of work have a much higher rate of success, and frequently not one you can fake in peacetime, either that fly-by-wire plane flies or crashes. The success of e.g. the F-16, F-117, F-18, B-2, F-22 speak for themselves. And the ABM Standard Missile 3 and it's supporting AEGIS system sure seem to be able to blow up things....

So if he's this stunningly ignorant or biased about a proven governmental success, how much is the rest of his judgement to be trusted?

Like this ludicrous analysis, either massively ignorant of the basic facts, or trying to protect the political and civil service types responsible for this mess. Seeing as how those masters of the universe in the HHS's CMMS decided to handle the integration job of coordinating 50+ contractors, including integration testing per the AP. And made it impossible for them to succeed: per the NYT "In the last 10 months alone, government documents show, officials modified hardware and software requirements for the exchange seven times." (https://news.ycombinator.com/item?id=6583327), those changes continued through the last week before the launch, and full testing obviously was delayed until that last week.

"So if he's this stunningly ignorant or biased about a proven governmental success, how much is the rest of his judgement to be trusted?"

This is Tim "XML" Bray. So, yeah.

XML???

I rest my case ^_^!

Seriously, I know an equivalent S-expression system wouldn't have been accepted, but it's nonetheless an over-complicated nothingburger to us Lispers.

More seriously, XML is a micro level thing, it says nothing about his high macro level political and systems development knowledge (proven to be sorely lacking as I demonstrate), goodwill (ditto), or expertise (seems to be sorely lacking, but that could be politics interfering).

To make one point perhaps more clear, he's talking about a inextricably political thing and can't help but to totally denigrate the other side on false grounds. Doesn't even have the self-discipline to hide it. So he impeaches his comments, showing he's not to be taken any more seriously than any random Joe programmer on the street. Or rather, even less seriously.

I love that definition. Enterprise Software means "Doing It Wrong". So many management folks just roll over and play dead when an IT guy says "This is an Enterprise class package" not realizing that the IT guy has just said "I don't have a clue what I am doing. Sadly, business schools try to teach managers how to make hard decisions, but in the IT realm, they seem to forget everything they have learned.

And it's not as if there are no examples of how to manage technology projects around. Look at an aircraft manufacturer or NASA and see how much they rely on incremental change followed by lots of testing.

> "The chances that the most elite squad imaginable of Googlers, Facebookers, NSA geeks, and Government-of-China attack hackers all laid end to end could have made this work at startup? Zero."

Yes they could. Look, one of my good friends works at Duke, and they're constantly pulling from legacy, disparate, the most xml you've ever seen, medical systems all the time. The amount of J2EE -0.35 code would make you cry. Sure this is definitely on a larger scale, but it can be done. By scale I mean the amount of disparate systems, not the comical amount of traffic the site got.

Healthcare.gov is an oversized wizard sign up form. Sure there could be bugs later in the process, but signing up a user shouldn't have these problems.

I'm hoping that as a developers everybody can see that.

No way of knowing the amount of people that are signed up? No, that's either some crap the government is feeding us, or a sign that this is an over engineered pile of dog turd.

> "Could Healthcare.gov work? Sure. It probably will, eventually." Eventually is a good word, but throwing more programmers at the problem will only make it worse.

All correct points. And I myself have interacted with lots of older (pre-J2EE) systems, they tend to be characterized by stability and sanity, however weird. Grok the formats, write adapter code, you can get the job done as long as you don't exceed their duty cycles.

You're right, those going on the record are feeding us crap, although the site is a bit more complicated because it wants to confirm identity to reduce fraud (although a lot of that was punted; this is a separate system done by another contractor using Oracle's identity system; CMMS panicked and proposed to replace it 3 days after launch), it must get FICO scores (from Experian as it turns out) because that's a part of insurance policy pricing, and it and only it (or a subsystem below it) is the only allowed source for all sites' subsidy calculations, which are utterly important for Obamacare to work (otherwise a lot of less wealthy people and families simply won't be able to pay for policies, since there's only one gold-plated type now legal, the old major medical ones are outlawed come Jan. 1).

Except perhaps for getting the FICO scores, all of the above is being observed to frequently fail, as is the interfacing to the insurance companies (which ought to be de novo), they're getting self-evidently bogus data including multiple enrollment and cancellations of the same person.

Who know's who's responsible for the latter, but CMMS is unquestionably responsible for making sure that worked.

"I'm hoping that as a developers everybody can see that."

For far too many, like Mr. Bray, revolutionary truth beats bourgeois truth ... pity that attitude is slamming into the brick wall of reality, that computers do exactly what we tell them to do, not what we want them to do. Which of course is delaying any steps that would get this fixed in time, e.g. the White House and CMMS have been consumed with the launch mess for 3 solid weeks and hope to have a plan come Thursday, vs. e.g. the White House realizing CMMS as the integrator, including integration testing per the AP, and HHS/CMMS as requirements specifier are part of the problem, not part of the solution to any likely fix before things get really ugly.

And as you note, ignoring the lessons of The Mythical Man Month" and throwing more programmers at it, calling it a "tech 'surge'" (an odd call out to Bush competency or a precursor to blaming it all on Bush? :-)

Reply on news.ycombinator.com