Here's a sample of terrible MS Word HTML for just just one list item:

<li style="text-align:justify"><span style="font-size:12pt"><span style="font-family:&quot;Times New Roman&quot;,serif"><span style="letter-spacing:0.25pt"><span style="font-family:&quot;Segoe UI&quot;,sans-serif">Use this link to <a href="whatever.htm" target="_blank">Recommend</a></span></span></span></span> <span style="font-size:12pt"><span style="font-family:&quot;Times New Roman&quot;,serif"><span style="letter-spacing:0.25pt"><span style="font-family:&quot;Segoe UI&quot;,sans-serif">Lorem Ipsum</span></span></span></span></li>

Or possible it's a combination of MS Word HTML combined with changing styles in CKEditor.

In English this is saying:

Make a list item...
with justified text
Inside, make the font size 12pt (overriding the site's default font size)
Inside, make the font Times New Roman (overriding the site's default font)
Inside, make the letter spacing 0.25pt (a very esoteric thing to do, override the site's default)
Inside make the font Segoe UI (overriding the font just set)
... ( a bit later ) ...
Close off all those styles
Now, make the font size 12pt (repeating again what was just closed off)
Now, make the font Times New Roman (")
Now, make the letter spacing 0.25pt (")
Now make the font Segoe UI (")

We should be able to know what the default font size and font is, and remove any top level rules for those. It's not 100% perfect as a rule as we don't know the global defaults are in play in context, but >99% of the time they will be.
We should be able to strip any rules that get overriden by lower-level rules without anything being applied to it otherwise.
It needs to be recursive.

Another common problem is empty paragraphs appearing at the end of a document, possibly before a Tempcode comment.

EDIT: Another is setting "font-size:1em". Does absolutely nothing.