Completely and totally agree.

Here’s another one from Tim Bray that I got over at Yakov’s blog.

1) The consulting firm & different consultants involved recommended going with a blank slate for the system. There is a lot of +/- for doing this but the scale is increased exponentially trying to tackle a project of this size from the ground up. That and starting fresh also means that there could and probably were a lot of hidden issues that were not thought about until they become more apparent.
2) Many of the systems, such as Medicare/Medicaid are legacy systems that had to be integrated into the workflow. Some of these systems are built in Cobol, so I can only imagine how complex/fragile the integration points must be.
3) Many of the States used the federal funding to tackle other outstanding issues & outdated systems at the same time, effectively increasing the complexity and time frame of the project(s)
4) Much of the system is really a bridge that interacts with 3rd party systems, such as listing the current offers from insurers and then transmitting the form back to them. It sounds like a lot of these 3rd party endpoints came online late in the game and having to stitch them all together was probably a last minute scramble.

Fundamentally, this sounds like a nightmare project in complexity and scope without all the other issues such as bureaucracy, red tape and other guidelines that occur when working in governmental projects. I definitely have a lot more leaway & understanding for the people behind the system, but only because I have been there and know how projects like this work.