Rajvik Wrote:what makes this even more funny is that 3 guys in their "basement" built a working version inside a week. Stable, doesnt give anyone any trouble in their tests, the only problem, it gives you the tiers and prices up front, not at the end or after you've entered all your "private" data
Because A) The system doesn't need to deal with the kind of scale, B) it probably does not reflect the various state law ideosynchrasies, C) It doesn't have the anti-fraud systems the official one has to have, D) They don't need to deal with large organisation beurocracy.
It is certainly possible to roll out a system of this scale in the time they had, but that would have required strong leadership and management, and a willingness to tell states that since they didn't want to implement their own exchange any changes they want in the exchange... well too bad, it gets put in the after roll-out task-list, this includes changes in statelaws regarding healthcare, etc (ah I can hear the states that didn't want to cooperate howl). That way you are working towards a fixed set of objectives not a constantly changing set. I do note requirements where still being changed until just before the rollout. Garanteed disaster that, you need a fixed set of requirements to work towards if you are going to do it quickly and well.
Second you should probably rely on an internal team, not contractors, and if you are relying on contractors go hire someone that knows how to do this, talk to google, or perhaps microsoft or... basicly any of the providers that know how to build scalable systems. Hell the NIH could just look at who has done well in projects like CaGRID, which is fairly large scale and hire those contractors instead of the set they hired that is more adept at navigating the federal beurocracy and at leaching money than doing their task. (Ah I can hear congress howl about their pork providers not being treated fairly.)
And finally The order of priorities should be; Works, Scalable, Complies With Rules, Pretty. Instead the Priority was pretty clearly, Looks Pretty, Complies With Rules, Works, Scalable. Yes purple text on a black background looks very nice, until you try to read it, and while the sins of the website where not that bad it was very clearly designed to look pretty first and foremost, but then that is often all that non-technical management sees, and even technical management won't see more unless they dig in. This is a very common problem. (Ah I can hear upper management howl about the looks. Get it working as black text on a white page before adding images, css and everything else.)
You don't need to do all three, but you better do at least two of them, and do them well if you are skipping one.
E: "Did they... did they just endorse the combination of the JSDF and US Army by showing them as two lesbian lolicons moving in together and holding hands and talking about how 'intimate' they were?"
B: "Have you forgotten so soon? They're phasing out Don't Ask, Don't Tell."