I believe in some chemistry fields they do put two levels of uncertainty. If anyone has example code of propagating uncertainty automagically with a normal distribution and correlations I'd be interested. Implementing the output of gum workbench in c++ is a bitch.
I thought about reveal codes as well when I saw this. I miss it sometimes, but it always brought out my OCD side. I was always tempted to spend time erasing codes that didn't actually impact the document.
Reveal codes was awesome. My memory is of when word processors started to be able to apply HTML. An instant shitshow of massive amounts of repeated inline styling for a single character. You could not find the actual text in the mess.
Yeah, I get this anxiety too. And it can actually be an issue. The best case is that at some point in the future you click into that space and start typing, only to find everything in bold. Easily fixed. The worst case is that you find one of the chaotic, baffling stupid dead ends in Microsoft Word's layout engine and lose hours fighting it to get what you're sure should be a relatively simple layout.
In fairness this is exactly the model that Google Plus used, and people hated that. Though maybe with today's machine learning you would have some auto generated sharing groups which would make this easier.