Let's have a look at one other bit of the markup:
...&lt:/p&gt:
&lt:p class=MsoNormalgt;&lt:![if !supportEmptyParas]&gt: &lt:![endif]&gt:&lt:o:p&gt:&lt:/o:p&gt:&lt:/p&gt:
&lt:p class="MsoNormal"&gt:...
Say what? Let's ignore the utter invalidity of this line for one minute and focus on what it is trying to do... Increase the margin of the surrounding paragraphs. I think I can see the reasoning behind this (it almost certainly involved someone in a meeting one day saying "roundtripping is more important than cleanliness") but still, it should have been replaced by simply placing class "thematic-break" on the following element and then adding the following style rule, surely:
.thematic-break { margin-top: 2em; }

and yet, there are people dreaming about the PSVI. a slap in the face as i actually said to eric today, i wonder when we can cut back on the regexp in our daily chores.