
Your confusion probably arises from the fact that /software/ which is designed explicitly to deal with ePub or MOBI files is /also/ designed to display the interior HTML in a way that makes it more "book-like" (e.g. page-oriented, with new pages starting at the beginning of chapters).
My "confusion" is over why anyone would consider reading HTML via any of the major HTML browsers a "book-like" experience. I certainly know at least something about the internal formats of E-book files having made literally about 20,000 such E-books from PG HTML files. I've tried as many ways as I can figure out what HTML browsers can be "adjusted" to give a "book-like reading" experience and have always been stumped. Tried things like playing with the accessibility features, etc. Problems include no ability to set or adjust margins, no easy way to set the reading width, no easy way to choose a sensible reading font or sensible reading size, no way to adjust background vs. foreground contrast ratios to make the contrast less blinding, etc. It's not that all the e-book reader software "works" either, for example Adobe Digital Editions (ADE) ALSO doesn't have any way to set sensible margins. And it's not that the e-book readers are perfect either, for example Kindle doesn't allow sensible margin adjustments on PDF files -- where in effect often one would like to set a NEGATIVE margin allowance to remove some of the margin hard-wired into the PDF file. Kindle DX seems to perform automagic PDF margin trimming in portrait mode, but not in landscape mode. Also when one plays with the HTML browser to make book reading a more "book-like" reading experience, then that in turn screws up one's use of the HTML browser for non-book-reading "normal web site" experiences. Agreed that EPUB and MOBI are basically ENCAPSULATED HTML -- plus "SPINE" information. What then is the difference? HTML isn't encapsulated, is missing spine information, and comes with browsers which "don't get it." On the positive side, HTML browsers can do YouTube which I think is the way the HTML world is heading: namely video. We should be very happy that other people who like to read are willing to buy dedicated hardware to do so when they could instead be buying flip video cameras to film god say what! In short, I think we the entire computer world is actually still very much in the early days of learning what it takes to make reading a pleasant and transparent experience on one-or-another chunk of hardware. The paper book industry has had at least 400 years to figure this out -- too bad the CS types (of which I am one) don't look to the paper book industry to see how it ought to be done!