It is natural for a user, after researching with our system, to desire to print selected texts and fragments for personal - or group - study away from a computer. Such printouts are one-use artifacts. It is clear that ability to produce such printouts must be present in the system from the beginning. The question is: how good the typographically does it need to be?
We need to format a tree of texts: main one, commentaries of it, commentaries on commentaries etc. It is known about each piece of commentaries what is it commenting on. All the font metrics are also known: glyph sizes, what is haging how low and what is sticking up and how high. Result needs to be readable and (is it a separate requirement?) beatyfull.
To format "like in a book", we need to optimize the following contradicting constraints (the list is probably incomplete):
the page must be fully covered with print
comment must start on the same page where what it comments on is
comment must end on the page it started
Koritz says that we do not need to print books, but "leaflets" instead: text with comments that fit on one page. In the "forum format", whatever that means.
Gorkin says that printing "like in the book" of the multi-layered text is extremely challenging typographically, and thus very interesting, but design of the overall interface of the system is even more interesting - and difficult. And more importants. Also, what exactly are the requirements for the printing facility, and what is their order of importance, will become clear only in the process of using the system. So, initially printing needs to be acceptable, but primitive - we do not have resources to do fancy stuff from the beginning.
Dubinsky says that the format that will "grow" from the use of the system, will turn out to be a familiar to us all format "like in the book", or so close to it, that a solution for one will fit the other; that good leaflet is not easier to print than a book; and that ability to print familiar "book-like" format is neccessary for the psychological comfort of the users. But he also agrees that features and interface of the system are more important.
Thus, everybody agrees that initial printing facility will be "primitive". Gorkin does not want to expend any effort to even find out how primitive. Dubinsky would like to see something acceptable. Nothing of the sort has been found so far. XSL-FO  is insufficiently expressive for our problem - even version 1.1, it seems.
Beyond Pretty-Printing: Galley Concepts in Document Formatting Combinators