I live in Washington state, which had one of the cleaner Affordable Care Act website implementations in the country. I had insurance in literally less than 10 minutes, and an ID card in the mail a week later. There were stumbles, but living on the border with Oregon—and being subject to its public radio network—really puts those stumbles in perspective. Last fall, as states (and the feds) around the country sweated and pleaded for patience, the governer of Oregon was only petulant, saying simply that the site didn’t work and that he had no idea when it would work. That’s about it, and on top of things, Cover Oregon, the state agency in charge of the insurance exchange and its rollout, has largely kept quiet about the problems and how they might in the future be resolved. Right now, the state has only the slightest promise of resolution at all. 

The current situaton is that Oregon is at the decision point between attempting to fix the current site—whose bugs have expanded from 13 in January to 300 in April—or ditch the project entirely, throwing away the $140 million in federal dollars already spent on the site, not including the wages of a small army of workers hired to do manual (paper) insurance sign-ups. From the start of the open enrollment period that began last October, the state has ranked dead last in terms of exchange-based enrollment and is currently the only state where residents are still unable to sign up online. The current chatter is that it may not even be possible to have a functional enrollment website by the opening of the next enrollment period, in fall of 2014. That’s insane. 

Implementing the federal exchange system in Oregon could take up to eight months, according to Portland consulting firm Deloitte, which has been in closed (or at least very quiet) negotiations with the state to take the project over after the original contractor, Oracle, departs this month. The nature of that would-be handoff remains a mystery, like most things to do with the site since October. According to the Oregonian:

Officials for the state, Cover Oregon and Kitzhaber’s office declined to answer repeated questions about the exchange and related topics. For example, officials declined to answer questions concerning Deloitte’s billings, Oracle’s departure date and whether Kitzhaber plans to have a competitive bidding process to replace Oracle.

The state blames Oracle for the problems with the site; we know at least that. But it’s not nearly so easy. For one, Oregon picked Oracle recklessly, ditching the state’s normal competitive bidding process midstream, essentially taking a blind shortcut. Two, that shortcut shouldn’t have been blind at all. In the adult world, it’s not nearly enough to ask Can you do this? Any contracting process that doesn’t involve someone with the ability to look at a contractor’s proposal and say it’s busted and next please is fundamentally broken, or corrupt.

Information Week explains the disaster that came next:

Extricating the state from dependency on Oracle won’t be easy. One of the audit’s primary findings was that the system architecture was overloaded with Oracle software. In an abbreviation-packed critique of Cover Oregon (CO) and Oracle Consulting Services (OCS), the report places ultimate blame on CO for failing to have adequate contracting processes in place but also faults Oracle for poor performance and transparency, meaning that the company failed to provide information needed to remedy issues with website and contract performance.

So, every nook and cranny of the Cover Oregon website is polluted with Oracle, a popular provider of propietary or closed-source database software and middleware. If you’re a government or business that doesn’t want to deal with technology at the ground level, you might buy its products for out-of-the-box ease. The result was a project “highly dependent on Oracle consultants,” according to IW. This dependency on Oracle brand-name products poses problems for whoever takes over the project from Oracle. In particular, Siebel, the customer-relationship management element of the system, isn’t even open to revision/code branching. It just has to be built again from scratch.

A federal report issued last winter found that, in addition, “Oracle has often refused to turn over requested information about programming changes, system performance, and testing to the state or to other contractors working on the project.” Meanwhile, Oracle was overloading the project with its software, overcomplicating the site with CPU and memory taxing applications. Servers were constantly needing to be restarted. The report notes that any replacement for the current software will need to be vastly more simple. 

Simplicity is far easier when you’re actually building a software project, rather than pasting a bunch of existing brand-name products together. Kentucky, yes Kentucky, had one of the most successful exchange rollouts in the country, thanks largely to its bare-bones approach to building the project—no interactive bells and whistles—and also state employees/managers having had recent experience building and implementing large software/website projects, like a prescription drug-monitoring database and a document management database.   

That’s a strength you’d imagine the state of Oregon to have as well, being a generally technology-friendly and highly technology-concentrated state, not to mention a state very in love with its own programs and would-be ability to proactively boost its resident’s lives through those programs. I guess that might explain why state leaders are being so raw about the whole thing, if not childish. Hubris.