You can actually embed fragment identifiers inside PDFs, but I don't know if any web browsers can understand them. Your other points are of course known to everybody - I even know these things myself, and yet I maintain that "PDF is excellent for presenting information on the web". Especially PDFs designed for delivery on the web, rather than 20 MB files designed for printing. HTML solves problems, like reflow, that are unsolved by PDF, but PDF is still the best choice if precise control over typography and layout are needed, as they are for some documents. The article you link to was right on target in 2003, when it appeared, but is obsolete now. Browsers no longer crash when asked to render a PDF; no external readers are needed, if you use the right browser. In Chrome I can navigate into and out of PDF documents completely seemlessly; if used carefully they can be an integral part of the web.