I get that as the motivation 20 years ago, I just think it's irrelevant when I have a 4k monitor. I use a 2px border and a small line of text for a title at the top of parent nodes. I have seen the complaint that this reduces the accuracy of the area-representation, but it's a minor and very worthwhile tradeoff.
From experience, it may be minor or it may be quite major depending on the shape of your data. If you have a whole bunch of small items that, combined, only account for say 1% of the total, then adding borders and labels to all of them may significantly skew that proportion, and the proportions of its parent/ancestors
Yes, that's true. It depends on your use case. I'm almost always focused on the largest items, e.g. for space recovery. But also, being able to zoom in and out negates that issue in every case for me personally.
