In a way google serp, e-mail, twitter already use this simplified UI - now if only we could unify it somehow.
One way of doing this would be by using console browsers to interface the web, or set your browser to use 1 font and style and no images. One of these browsers, Elinks 0.12pre6 has gopher as compile time option as added benefit.
Aonther problem with Gopher is that it doesn't appear to support encryption and compression. I remember the pipe and slash symbols 'spinning' while waiting for a large plain text document to complete downloading in the old days. So the "resource lite" argument doesn't fly completely.
So perhaps the best option is to use simplified and structured gopher like menu-ing system using html, something like markdown did for html as is suggested... but why then not put Gopher to sleep for even and start a www-less project?
There really isn't a significant difference between the two. They both operate over a single TCP channel. HTTP/1.1 has a slightly higher header size, but it's not even half a kilobyte.
The only real difference is that HTTP usually serves up rich HTML, which contains all sorts of graphics and JavaScript toys and other bandwidth-hogging transmissions. If we limited HTTP to using HTML 3.2, no JavaScript, and only the occasional 2kb gif, it'd be resource lite, too.
FYI--I wrote the Apache::GopherHandler Perl module for running a Gopher server under Apache2, which I now consider to be mostly a waste of time. (https://metacpan.org/pod/Apache::GopherHandler)
gopher://gopher.floodgap.com:70/0/gopher/relevance.txt