If you read through old HIGs, they generally have actual concrete reasoning behind the choices they made. They're pretty interesting to read because back then they did scientific user studies to try to make computers more usable.

For example, Microsoft didn't make the UI gray because of some aesthetic preference. They chose gray because black-on-gray text is easy to read, and because it makes it immediately obvious what's UI vs. what's content (vs. if they made the background white or something). They made all the buttons three-dimensional because it makes it immediately obvious that something is clickable. Getting rid of stuff like this makes the UI prettier, but it also gets rid of hints for how to use the software.

It’s hard to determine what was aesthetically good from a different time because tastes change.

But there’s nothing wrong with small buttons when your only input tool is a pixel accurate mouse. Arguably what’s worse is too large a button.

But what’s inarguably worse are things like window corners where the click zone extends far beyond the actual window, and parts of the window are not click zones.

As for inconsistent - what!? Every app looked approximately the same, and behaved in a predictable manner.