The current failure of the HHS 'Obamacare' website and healthcare portal heathcate.gov and extreme cost of the failure (some site list $634 million) has pointed out once again the government has no business in specifying, building or developing large and complex technology projects.
1. Gov internals
The folks inside (and helping) the government oftentimes aren't clued in with what is 'state of the art' in commercial industry. Not their fault, skills degrade as you work in a large bureaucracy.
Worse: the gov also tells the industry not just what it wants (bad specifying usually) it also tells industry how to build tech, which is really bad because the gov doesn't know how modern tech is built (see DevOps). Adding extra processes that aren't used in commercial practice simply creates a sub-class of specialized companies hose only competitive advantage is understanding gov-speak (CMMI, CONC, CNA, etc.), to say nothing of actually gov contracting.
Also as skills degrade (and this is especially important for higher tech goods that are assembled with other tools) the knowledge of the tools around the tech also degrade. Example: in the lead up to WW2 the US Army was still bolting tanks together using railroad chassis, where by that time private industry had moved to building beefed up truck chassis with newer and faster welding for large load trucks. The US Army's knowledge of the process for how to build large vehicles was way behind industry, because the new, fun, interesting and hard problems to solve were in the private sector. This leads to point 2.
2. Smart and innovative tech people do not want to work for the government, except is very short doses or on really hard and unique problems. No extremely talented tech person wants to work at HHS, HUD, etc.. Reasons: they don't pay well and the problems they have aren't hard and/or require unique skills. The only exception to this are areas where the government is still the leader in a tech: crypto and comm's (NSA), cancer research (NIH), nuclear (DOE) and weapons (DoD). Agencies like NRO and NGA and their contractors (satellites and maps) can't hold onto top tech talent anymore as the commercial industry is more fun and less burdened by regulations and bureaucracy.
3. The acquisitions process.
The process for how the gov acquires tech is slow per 1, but why? Because it grew up to buy hardware (chairs, desks, tanks and planes), not tech that rapidly changes like the more malleable soft tech's (computer hardware / software). In soft-tech since the technologies are so malleable, evolve and morph over time, the tooling and process for how these systems get built is more important than the resulting product. Its the process that matters more for soft-tech than the actually product (because the product is simply a finite snapshot of what the process can deliver at that time).
So having an acquisitions process that is historically geared to buying stuff versus process leads to failure.
Also, hardware buying front-ends design optimization to build many things cheaply per unit (in gov speak 'measure a xxx times, and cut once') whereas in soft-tech the end product is constantly optimized throughout its lifecycle (design, test: repeat).
So where does that lead gov tech and acquiring soft-tech?
Rule 1: The gov should not acquire or pay to build a system that is not core to the Agency mission.
Unless an Agency truly can say it has the best and brightest in the industry working for it (all the way up and down the chain AND is driving that industry) to help it specify what it wants (and can throw the BS flag) it must not be in a role to specify a system to be built. The knowledge just isn't there, its not a fair fight. How can a blind man ask for red hat?
Examples of things that aren't core: travel systems, motor-pools, HR (unless you run under-cover operatives), tax-return software...
Things that are core: see above, but crypto, weapons, tools associated with espionage, research, others….
An example of acquiring non-core: years ago the DoD decided it needed a new travel system. The options were build or buy the service. The travel industry came back and said: we will build you a system and charge you per ticket, just give us the high level needs (security, etc.) and we will charge you per ticket issued (i think it was $64). Morons in DoD said 'no, we can do it more cheaply' and then set out to blow through allot of money (see: http://www.gao.gov/products/GAO-09-577). $ 2 billion? to say nothing of the hassle of actually using it.
Sound a lot like healthcare.gov?
Even the CIA and NS have realized that building cloud computing technologies aren't core to its mission. The CIA is outsourcing its cloud to Amazon (probably) and the NSA has adopted other cloud tech for its use, versus developing it on their own.
Rule 2: If a desired capability isn't part of the Agency core mission, outsource the need.
Example: if industry can meet the high level requirements at a fraction of the cost that the gov can, why wouldn't you outsource it?
In the above Defense Travel Example: setting high requirements service level agreement around security, uptime, response time for buying tickets, help desk, website usage, etc. would be so much better than trying to specify a capability based on old outdated knowledge of tech and tooling. Also in this manner industry can more rapidly upgrade and evolve the capability to take advantage of new tech and ideas. Like adding mobile smartphones to travel, etc.
Whats to be done with Helathcare.gov/ObamaCare portals?
Monday morning QB in full effect…
HHS should put out high level req's around the service it wants for citizens (its customers) to have: response times, help desk, etc. and then put a bounty or fee for any company who could meet it. Some companies might do it stickily over the phone out of Indian call centers, some might use tech. we just don't know since HHS preconfigured and (worse) assumed a solution.
A fee for every person successfully signed up would be better than to blow $634 million in a few years with nothing, but a 404 to show for it (I got 404 problems, but a plan ain't one). $1k per person initially and then declining over time as the tech becoming cheaper, more reliable and more companies enter the marketplace.
If the goal was to sign up as many people as possible for healthcare, why would you assume you know what the answer should be? Why not unleash American ingenuity and business on that problem to solve it in anyway possible.
Frankly: Kathleen Sebelius (the head of HHS) should be held accountable and fired, weaseling out and saying we didn't know, etc. its the process that sucks is a cop out we should not expect as taxpayers. Nothing will change until someone takes responsibility for this, she's in charge otherwise we let mediocrity rule.
Expect more from your Gov, it should suck less.