IIRC they were random letters that do not occur naturally in English words, in order to serve as easily searchable annotation/footnote identifiers. He somewhat already knew they were extra details that would not make it to the main body. Later these get changed [1], [2], etc but you don't want to use these at the outset due to possible reordering/deleting.

Source: I was one of the founders of Stypi and we asked him when porting the data from Etherpad to Stypi.

https://techcrunch.com/2011/08/09/yc-funded-stypi-is-etherpa...