I work for a defense contractor, so I have some experience here. There are at least two obstacles to making good looking software/sites paid for by the government:
1. Nobody wants to pay for "Looks pretty." You have a list of requirements to satisfy. Each feature costs a certain amount of money. If it comes down to a feature or a pretty UI, it's going to be pretty hard to justify to someone that you spend money making it look good when it otherwise works perfectly well.
2. "Looks pretty" is a testable requirement. Sure, you can have focus groups say what they like (but that costs money, see #1), and you can have requirements like "no functionality shall be more than 2 mouse clicks away from the main screen"--but then you get a screen with 300 buttons on it. There are heuristics you can use to do good design, sure--but there's very few ways to create general, testable requirements that reflect good design.
3. Sometimes there's the problem that the entity paying for the software is not the entity using the software. To clarify: this situation is usually the case, it's just that sometimes the people with the money care about the users, and sometimes they don't or have their own agenda.
I challenge you to find a site for a city, state, or municipal site with more accessible data or more interactive functionality.