Scribd’s HTML5 is a mess
14 May 2010
Because Mike Cane won’t stop nagging me about it:
What Scribd has implemented isn’t HTML5, it’s barely even HTML 4. It’s a slapdash construction made out of SPANs, absolutely positioned P tags, @font-face rules and a huge, boiling vat of indistinct inline css. All talk of scribd using ‘semantic markup’ is nonsense, its output is about as meaningful and semantic as tag soup. How could it be? It’s a hack used to render PDFs in HTML, it couldn’t be semantic without a generous dollop of telepathy. Then there’s the issue with how they deal with fonts, which is almost certainly in blatant violation of the licensing terms for the original typefaces embedded into the PDFs.
But this issue goes beyond just scribd (and scribd’s problems go way beyond the fact that their main product is a kludge), every major HTML5 app today is built using unreadable markup, obfuscated javascript and inelegant, messy CSS. Try to use ‘view source’, for example, to figure out how Gmail or Facebook work.
Go on, I dare you.
