I viewed the HTML page source just before, and after, the page copied itself (I didn't know if there would be a difference..). They're attached at .txt files.
There's certainly some differences in those files, including invalid HTML. The odd part is that the bad HTML is in
BeforeCopy.txt, rather than after. It may be a red herring, but if you can find out why that is being produced and fixed it it may help.
I cannot easily figure out what commentary system you are using, but certainly something there needs investigating. The following were bad features noticed.
In the javascript function
fetchmorecommentary():
c</brTimeout(commentarytimerid);
which later becomes the correct
clearTimeout(commentarytimerid);
In the outputted
charinfo section (only part of line shown since it is long):
<span class='colLtGreen'>50span>
which later correctly becomes
<span class='colLtGreen'>50</span>
It's hard to imagine what might cause this, but it does suggest some kind of corruption of the output stream.