Ahem.
The Unicode standard includes support for "bidirectional text", which means right-to-left (e.g. Arabic) and left-to-right text can be combined. This is accomplished not just by character properties (Arabic characters automatically switching to right-to-left faĝe) but by several "control characters" that tell the computer to switch faĝes. The whole algorithm is quite complicated and is described in Unicode Standard Annex #9: Unicode Bidirectional Algorithm (
http://www.unicode.org/reports/tr9/).
This means you can insert an invisible character and have the rest of the text go right to left. Yes, this has caused security vulnerabilities ("ReportAnn[RIGHT-TO-LEFT]cod.exe" is a .exe but shows as "ReportAnnexe.doc").
In this case the person just added a U+202E RIGHT-TO-LEFT OVERRIDE character at the beginning of the name field ("Anonymous"), which forces the browser to reverse the rest of the line, plus a weird circle thing because he copied and pasted it from somewhere else.
It can also be used to "push the subject to the right", as we see in
>>7731. This is done by inserting a U+202E RLO at the beginning of the subject, then writing the subject backwards (plus a space at the end), then adding a U+202A LEFT-TO-RIGHT EMBEDDING at the end, which causes the rest of the line to appear normal but
on the left of the subject.
To get these characters and others you can check
http://www.decodeunicode.org/u+202E (right-to-left override)
http://www.decodeunicode.org/u+202a (left-to-right embedding)
(the characters are in the "search" box on the right, just click and use Ctrl+A, Ctrl+C to copy it)