We have recently being doing some work with a number of university sites. We have been usability testing their sites and doing attitudinal research with users to understand better their goals and circumstances.
Pick one at random and try answering a few basic questions that potential undergraduates have. Start with the the single most important thing their website can do things that are really important to them like 'What is the accommodation like?' (It's the first time they have been away from home); 'How many girls are there at the university?' (Let's assume this is a boy - need you ask?); 'What does the beer cost?' (Perhaps more important than the previous question!).
Does it really answer your questions? Can you see a picture of the accommodation, of the room, of the bar? How about the course information? 'What courses do they offer that include English?'. Then try and select one from the list you are offered. Can you choose between them? Do you know the difference between English literature, Comparative literature and Global literature? Does it explain this without having to go and look at each one in detail? And don't forget you are 16, never been to University and don't understand the language they use.
We had a bright 16 year old looking to do a Psychology degree at a good university. He hadn't looked at any University sites before, his parents hadn't been to university, and it was still early days. His search returned a range of results including a 'BSc honours Psychology', and a 'Graduate Diploma in Psychology BSc' No other explanation and he had to choose between them. Well, he chose the Graduate Diploma because he thought the Honours course was a higher type of degree and he needed a 'less difficult' course. He was completely unaware that the Graduate Diploma was aimed at graduates not undergraduates; the content of the course was then so badly laid out it then took him some minutes to realise his mistake.
Well there are a whole pile of reasons but the common thread seems to be the structure of the organisation. There appear to be two warring camps: the academics and the administrators. The first lot think they know what they are doing (they are all experts in their areas, after all), and they certainly don't want to do what the second lot want them to do!
Well, the problems are primarily organisational, not technical. Unless the organisational politics can be addressed it's tough to see how a fundamentally user-centred site can be developed: the internal stakeholders have to collectively agree that the the single most important thing their website can so is to allow users to achieve their goals quickly and easily - whatever that may mean for their fiefdoms!