That is correct, and in-fact, when generating ePub, I already dump-down the HTML that goes inside them. However, it makes no sense to dump-down a version where everybody has the capability to enjoy the added value you can add with HTML.
OK, how about the ability to provide two versions of HTML to PG: One the "vanity" HTML, and another the "dumbed-down" EPUB/MOBI version? And if a contributor doesn't want to provide the "dumbed-down" version, then let those of us who want to actually be able to read PG books on ebook readers create the "dumbed-down" aka "fixed" version so that PG customers, including ourselves, can read a PG version on our ebook readers which are not "scrambled eggs." And let's be clear, most of the time that which is being done in the "vanity" HTML can be done in a way which is almost identical and yet produce results that actually work on EPUB and MOBI devices -- it's just that the "vanity" HTML has been written using code which is nonconforming and ill-defined in the first place. Again, I don't understand the "vanity" in providing PG with HTML which is simply "broken." Just because HTML happens to display the way that you want it to look on your desktop computer with your display and your browser doesn't mean that code is right.