I wouldn't necessarily agree with that, but I will say that vague rules can often make for a better simulation than concrete ones.
Using stealth as an example, the existing rules are that the DM tells you whether it's possible to hide, based on all the factors they understand about the scenario. If we wanted to model that with concrete rules, we'd have to introduce things like facing (which would be a pain); and even if we did implement facing (in four or six directions), no amount of granularity would ever encompass all of the relevant factors to the same degree as the DM's adjudication.
A big problem with a lot of highly-granular systems is that they introduce rules which cause worse problems than the ones they solve. TV Tropes refers to this as "Misaimed Realism", and gives an example of Opportunity Attacks in 3E. It makes sense that you can't just run past someone with a sword who wants to stop you, but over-codification of the rules ended up with the famous Bag of Rats. If the burden was instead on the DM, to adjudicate whether or not an opportunity attack was appropriate at the moment, then that never would have happened.
Edit: Argh. I just noticed that I was responding to a post from five years ago. I kinda wish that these threads could be color-coded by year :-/
