....Furthermore one has to make assumptions of the true intent of the author!!
I'm not sure what the problem is if one has an <i>tag to indicate the author's intent was rendered in the original book in italic</i> and a <sc>tag to indicate the author's intent in the original book was rendered in small-caps</sc> etc.? On the contrary, the assumptions have to be made when the input markup language and the output rendering file formats are required to be one-and-the-same AND the rendering file format's power is less than that used by real-world printers already 400 years ago. Then the markup transcriber is forced to interpret authors intent and how to compromise that intent in order to make it fit within the constraints of the rendering language -- which is being artificially constrained to be identical to the input markup language. If one had a input markup language that closely follows author's intent as rendered by the original printer then the problem becomes how do you reduce the strength of this markup to match the weaknesses of the output rendering file format, and that in general is an issue of style that can be represented in CSS for example. Or hacked up by hand if and when absolutely necessary. But it still means that the previous round of volunteers efforts are correctly and completely maintained in the input markup language text so that the next round of volunteers can take another shot at the text some time in the future.