There are so many websites out there that do far more complex operation, and they seem to have very little problem.
Not really at least not that worked at this scale from day one. The closest you're going to get to needing to support millions of unique users on the first day, and hundreds of thousands simultaneously are things like MMO launches and WoW expansion packs or something like google+. And most of those can scale by replication and sectioning people off so it's highly parallel, or are built on already substantial infrastructure. If you crunch the math, there were only 90 days from launch to end date, and you need to enrol about 25 million people or something in that time (the uninsured who don't live in states with their own exchanges), so the daily load is actually quite high, particularly with a large number of people hitting the site to browse and decide. It's also quite likely that they gambled on more states setting up their own exchanges... and lost.
The backend of games and google+ of those is trivial compared to healthcare.gov, which not only needs to talk to databases from federal agencies, but it needs to connect to dozens of insurance companies with multiple sets of rules and regulations. Sure an MMO needs to do math, but one designer with no technical training can decide what equations to use and if they get it wrong no big deal. When you're dealing with money - and we're talking about healthcare that's going to be worth a couple of hundred billion dollars bought through this site, even a 1% error rate is going to cause no end of problems.
is that it's a simple matter of input from the user, and then a matter of storage of that input, and maybe some calculations along the way - all very basic stuff for today's world.
Input from the user that needs to be checked against multiple databases that aren't yours, that have private information in them. Then talking to multiple insurance companies in multiple jurisdictions with slightly different rules etc.
I'm not saying that excuses about 2 months of failure, but one should not assume this is a simple project, that they somehow did not realize that this would require probably 10x the server capacity they had is a complete failure. But other projects that are huge and stable have spent a lot more than 500 million dollars to get to that point, over a lot of years. These guys were trying to solve a problem no one else has ever had to solve on this scale. That they didn't recognize that is pathetic, but we shouldn't suppose this is an easy project.