Moreover, I think they're a good thing as a short hand when thinking about problems. You don't have to think about why this is a best practice for every problem you encounter, because if you did, you'd never get anything done. i.e., Why don't we write all of our code in a single main method? Oh yeah, because it becomes a nightmare to find bugs, update to add new features, reuse algorithms inside it...etc...etc. Instead of going through the rationalization every time, the Best Practices rule of thumb generally works out.
Where it doesn't work out is if you've never done the exercise of why they're best practices in the first place. For every best practice I follow, I have understood the justifications and as a result, have developed an intuitive sense for when my situation is the exception to the rule. Or, if my intuition fails me and it gets questioned, I can re-run the exercise and perhaps change my mind.
Related article: http://www.satisfice.com/blog/archives/27
That said, a lot of what I see in web development seems to be churn and features-for-the-sake-of-features bloat, and I think much of it has to do with the fact that the basics of a useful page --- images, text, hyperlinks, and some styling --- were essentially solved problems many years ago, so all web developers are left with is constantly trying to "reinvent" and "innovate" things that were, albeit not perfect, entirely usable. The "do everything in JS" trend is one of the clearest examples of this, as well as the "appification" of sites that are perfectly fine as a collection of statically linked pages. Blogs are the most common example of this that I come across --- instead of displaying the content using HTML, they require the browser to download and execute a huge blob of JS that often exceeds the size of the post content itself, just to show that exact same content, sometimes even making another (AJAX! It's awesome, so let's try to use it everywhere!) request to fetch that content in a different format, parse it, and then render it back into HTML. It sounds very Rube Goldberg, yet a lot of developers think this massive waste of resources is somehow normal.
Thanks to the Internet Archive, you can try visiting the BBC site from (exactly!) 10 years ago on your mobile device:
http://web.archive.org/web/20050521031013/http://www.bbc.co....
I don't have a mobile device with me at the moment, but it's clearly less "fluffy" than the page today, and loads essentially instantly even from IA's not-so-fast servers.