Old forms of the web
I came to what I thought was an interesting realisation the other day: the Web is almost of necessity backwards compatible. By this I mean that producers of pages can be sure of getting their message across by sticking to basic HTML because browsers can’t afford to throw away that functionality. As an extreme example of this, just look at the way you can still access GopherSpace. Follow this link (possible in Netscape, Internet Explorer and Lynx, at the very least) and you’re no longer in the world of HTML. You’re navigating through a strange, isolated world with lots of dead links, but still the browsers let you do it.
Introducing backwards incompatibilities in a program that you want other people to use is a generally a bad thing. With something like a compiler or a word processor, the incompatibilities can mean that the user has to alter their files. This can be enough of a pain to give the developer pause. But if their new browser is backwards incompatible in such a way that it fails to display pages that it used to display, and these pages are (naturally) beyond the power of the user to change, then that browser is hardly going to win itself many converts.
While support for the gopher protocol may disappear
because fewer and fewer people are using Gopher servers, HTML seems to
have become seriously entrenched. (So much so that I remember
reading an interesting article a while back about the fact that so
much of the information in URLs is coming to be seen as redundant; how
many company advertisements bother with
http://www.company.com? Most just go for
company.com; everyone knows to add the other stuff, or
their browser does it for them.) This entrenchment means there is a
great inertia protecting all those plain HTML pages. ASCII is
assuredly safer, but HTML is getting there too.