[skreutzer]

Well, such explanation might sound pragmatic, but it is part of the problem instead of the solution. There's off-the-shelf EPUB validation (based upon off-the-shelf XHTML validation according to W3C's specification) according to IDPF's specification, and if some software creates invalid EPUB output, it will deliberately break all reading/processing software which benefits from those off-the-shelf solutions. Further, the goal of creating an EPUB initially is to be compliant to an industry standard, because otherwise why not just create plain HTML output and maybe zip it? Oh, maybe because you have to be HTML compliant if you want browsers and other HTML software to work properly. One might question the value of standards in general, so please go ahead and specify your own binary data format which nobody then will be able to read.

Especially in the context of the future of the web, we can't afford to repeat the mess of the past with browser-specific HTML rendering, which hurts us badly with all kinds of incompatibilities up to the present day.