The Washington Post has referenced a useful graphic from Xerox on how the web application was designed to work, and the many components the system interfaces with. While the graphic is intended to illustrate the system's complexity, it also easily illustrates the biggest flaw in its design and architecture.
Like many websites, Healthcare.gov buries its useful functionality behind a registration gate prioritizing the collection of consumer information over the ability to browse and shop for plans. Web applications are often designed to require emails, Facebook logins, etc. in order to facilitate the development of email lists and capturing of consumer information. Users are essentially asked to give up a little bit about themselves to use the features they like.
Unlike many consumer websites, however, instead of a simple email, the federal exchange requires a much larger amount of consumer data as well as a number of identity verification steps. Not only is this bad for user experience, but is also the likely cause behind the federal exchange being overwhelmed and unavailable.
Instead, comparing plans is buried behind a step called "Eligibility". A quick look at the diagram demonstrates that eligibility determination is among the more complicated processes in the entire workflow, touching no less than 6 different components in different areas of the government. This account creation, data collection and identity verification accounts for the vast majority of errors users have thus far faced.
Ironically, one of the major changes the ACA made to health insurance was simplifying the way in which premiums were calculated, factoring in only age, tobacco and area of residence. Even the calculation of subsidies is straightforward, asking only household size and income. Returning an accurate and anonymous representation of plans and premiums in a manner similar to our exchange tools should have been easy to do. All the major carriers and eHealthInsurance have the same feature. Instead, all users are funneled through the most complex part of the system. Is it a surprise then that the system failed to have the necessary capacity?
Numbers from state-run exchanges, as well as launches of similar programs, indicate that the vast majority of users would simply be browsing in the first few days, with only a minority going through the entire enrollment process. It is completely inexplicable that the exchange's designers would require every potential user to undergo its most complex task first. Why anyone trying to create a user-friendly marketplace would organize a system in such a way is confusing at best.
One of the most important strategies when designing scalable software is to ask the user and your application to do only that which is required. The federal exchange was clearly not designed with this in mind. Rather than allowing users to shop immediately the system and its users are forced to jump through a number of hoops for seemingly no purpose at all. Unless of course, that purpose is to collect as much data on as many people as possible.