Yeah it does seem to be "pick the artifacts that bother you the least" with PDF conversion. I wish it wasn't such a popular format, hopefully a friendlier format will start to take over as the popularity of e-readers continues to grow.
I just got some insight into my issue as I was writing my reply. I was going to paste in a bit of sample text from a PDF, then convert it and paste in some results. I think my problem is a "carriage return line feed" type of issue. I noticed the text was formatted in such a way that wherever the text jumped to the next line in the PDF, the converted result had a full blank line inserted in. I remember an option about adding a line at paragraph breaks, perhaps I am triggering that inadvertently? I will experiment some more.
The problem I am having with Calibre's convert is it takes input like this small blurb form Brandon Sanderson's Elantris:
"ELANTRIS was beautiful, once. It was called the city of the gods: a place of
power. radiance, and magic. Visitors say that the very stones glowed with an
inner light, and that the city contained wondrous arcane marvels. At night,
Elantris shone like a great silvery fire, visible even from a great distance.
Yet, as magnificent as Elantris was, its inhabitants were more so. Their hair
a brilliant white, their skin an almost metallic silver. "
The same text from the converted Mobi file ends up looking like this on the Kindle:
"ELANTRIS was beautiful, once. It was called the city of the gods: a place of
power. radiance, and magic. Visitors say that the very stones glowed with an
inner light, and that the city contained wondrous arcane marvels. At night, Elantris
shone like a great silvery fire, visible even from a great distance.
Yet, as magnificent as Elantris was, its inhabitants were more so. Their hair
a brilliant white, their skin an almost metallic silver."
Actually it is too neat there still, here is a small jpg that shows the effect a bit better on flickr:
http://www.flickr.com/photos/45598342@N08/