Wikisource is still too complicated, and this is one of the reasons we don't have big communities.
IMHO what is really complicated is, the last step of digitalization (OCR review + formatting), it's almost impossible to simplify what is intrinsically complex. Do you know the Distributed Proofreaders approach to split such an intrinsic complexity into many steps?
Nevertheless, there's a wide range of complexity - some text being very simple (i.e. novels), other being extremely difficult (ancient books, theatre, scientific textbooks); perhaps the degree of complexity could be evaluated and explicitely stated, both by automatic scripts (page length + no. of templates + no. of unicode, non-ASCII characters....) and by expert users.