I'd argue it's because they're universally rather dull.
On a technical level, demos are highly impressive and how they manage to overcome the limitations imposed by the scene to create the works that they do is ingenious, but it doesn't make the resulting product interesting from an artistic viewpoint.
For me at least, the glaring weakness is the lack of narrative - there's no stories being told, there's nothing that engages or challenges the viewer. They're akin to videos that show off the features of a game engine where they engender the "that's pretty" reaction, but little else.
To my mind, this is caused by two things - primarily, it's a side effect of the restrictions imposed by the artform. With the limitations of the scene in play, there simply might not exist the scope to create enough content for an engaging narrative structure to be based around and procedural generation only takes you so far when it comes to creating assets in a resource limited environment.
Secondly, I'd argue that the types of people who are primarily interested in creating narrative are going to gravitate to different creative areas - creating short films, or animations for example. As a result, the demoscene is likely to be made up of people who mostly interesting in the question of how pretty the output is, rather than how interesting it is. This is of course also related to the fact that people who 'grew up' in the scene are likely to have a relatively narrow view of what the scene looks like and are unlikely to buck that trend and break out into doing something completely different.
The above is naturally just my two euroyencent of course.