Technically that doesn't prove Socrates never said it, he might've, though it would be one hell of a coincidence. It amuses me to think we'll never know.

I'm an Emacs users, so unsurprisingly I use magit, but perhaps surprisingly I use it sparingly, using Emacs's VC most of the time.

Kris Jenkins is a top-notch interviewer! He lets the interviewee talk, really pays attention and asks good follow up questions. I know that sounds like standard things an interviewer should do but at least in tech podcasts few seem to.

i get that from org-ql.

As you say, they are basically just window configurations, so I do use them ocassionally. If, in addition to remembering an Emacs window configuration I also want to remember whether the frame is maximized or not, I will use frames instead of tabs. I used to put window configurations into registers, before tabs existed, but tabs are better because when you put a window configuration into a register it even remembers the location of point in every buffer. This means that when you restore the window configuration from the register, points get restored to where they were when you stored the configuration, not to the last time you were using it. In this sense tabs are like window configuration registers that automatically update every time you switch away from them.

Tabs only have "useless UI elements" if you want them to! This is Emacs, after all. To use tabs without displaying any UI element set tab-bar-show to nil.

I use it in all buffers whose major mode is derived from text-mode.

By default undo does work in the scratch buffer so it is something in your configuration that is keeping this from working. As a quick way to check, try running emacs -q, which skips loading your configuration, and see if you have undo in the scratch buffer there; if so, it's definitely something you have in you configuration.

You can bisect your configuration to figure out how you are deactivating undo. You can do this manually or with the help of the bug-hunter package.