There is the excellently accurate sentence attributed to Tom Cargill that goes:
The first 90% of the code accounts for the first 90% of the development time. The remaining 10% of the code accounts for the other 90% of the development time.
This (fairly humorous) statement is a dig at all those projects that are developed and that overrun, but with a more than significant amount of truth in it. There can be a number of reasons why the project may not appear on time. Often, it's because someone is trying really hard to avoid accepting the fact that the Project Triangle exists for their solution. The triangle states that there are always three options: Do it Fast, Do it Cheap & Do it Good - but of the three options, you can only have two. You as project manager have to negotiate with your stakeholders, which of the three are the two most important ones - because there is flying-pig chance that you'll be able to get all three of them. They aren't mutually exclusive - but achieving all three is certainly something that is seldom heard of for a project.
I believe these two things are linked. I think that projects overrun because the project early on was "Do it fast & do it cheap", but near the end and as people start to inspect the system, people (whether it be the Stakeholders, End users or development team) decide that they need to do it well at the sacrifice of doing it fast. This could be argued as bad project management - but in reality, it's bad Stakeholders. I believe they were correct to concentrate on quality However, they were grossly wrong to have considered leaving it as the third priority in the first place - it should have always been first! Take your pick as to which of the other two is important to you, but unless your project is technically fantastic and the attention to detail is there - then you might not have bothered doing it at all!
Jeff Atwood has eluded to this in the past in response to comments that Stack Overflow would be an easy system to build. He implies that yes, while the functionality would be really simple to replicate, all the little details would be missing, all the little improvements for both technical speed boosts / efficiency savings and for ascetics and intuitiveness for the user. You can spend your time doing 90% of Stack Overflow in 1 week, but then you'll spend the remaining 90% of time faffing with the little details that will make your site stand out.
These little things can totally ruin your project / website / company, if you do not spend the time bowing before them in homage. There are plenty of website's out there which don't do justice to their potential. Websites that for some reason, got to the 90% of the way into their project and implement 90% of the features and functions - but then were put off by the long hard slog of the home straight which is the last 90%. The details beat them into submission! I'm not saying you have to be a designer to produce a good looking site, (it helps and I'd recommend that people get someone on board who specialises in functional website UI design), but if you don't set your sights too high, then patience and an understanding of CSS will normally do. If you spend this time sorting out all the little things (the font size, the position of your error messages, your link positioning etc), your project will find its bounce-rate significantly reduced (a good thing!) and the general happiness of frequent users will be improved as they don't have to work as hard to get the same amount of work done as your system is more intuitive.
Strive for perfection, excellence will be tolerated. If you are passionate about your project and you feel that you want to produce the best "whatever" in the world - then make sure you spend your time on the details! It's sometimes a long hard slog, but it is invariably worth it in the end.