At least regarding the US federal government, it was a known issue that (a) the government wasn't willing to spend enough to hire quality software developers and (b) the government bureaucracy lacked a process to determine whether web software was "good." Specifically, no effective methods of software validation based on modern best practices (the processes they used were descended from the validation processes for material acquisition, which look extremely different from software engineering).

I'm not sure if the problems have been fixed, but they were both recognized and a process was put in place to address them after the healthcare.gov debacle.