Web sites are easy. Data is hard.
In our last chapter, we discussed the Not Fun At All launch of HealthCare.GOV, complete with spectacular crashes, copious complaints, and a general lack of amusement.
Except among the enterprise IT community – the people who build Very Large Things – who have been laughing about this project since it was proposed. Why?
Well, past performance is no guarantee of future results, but it’s the smart way to bet your rent money. So let’s look at past performance: the decade-long effort for a single federal department to integrate its own systems. The Federal Bureau of Investigation spent years and $170 million on the Virtual Case File system, and…then trashed that whole thing and started again in 2005. It took the FBI another six years and a half-billion dollars to finally get to the point where they could declare victory.
Then remember that the complexity factor is exponential. That is, you need one connector to meld two systems; three connectors to meld three systems; six connectors to meld four systems; 10 connectors to meld five systems… Then imagine the complexity of melding all the systems necessary for this project from the Internal Revenue Service, Health and Human Services, a gazillion insurance companies, state agencies, health systems and Babbage knows what else.
That would be ambitious under the best of circumstances. Given that the design was still being redesigned 90 days before launch…it was not the best of circumstances.
And how does it work now, more than a year latter? Note the spate of recent news about the millions in over/under/we’re not really sure subsidies. Yep-slapped some spit and duct tape on it so the web site would look like it was working…
(Note to my enterprise architecture friends: yes, you’d be better off with a data mart than umpteen connectors. You’d be even better off with a proper middleware tier. Now go architect something or other.)