I too agree that a model with author fees is restrictive for the
publication of research. I would even go further and suggest that any
"paper based" journal model stiffles research publication. In paper
based journals, the marginal cost for every page printed is considered
and so an artifical scarcity is produced, resulting in a "pipeline"
and publication delays.

An open research model should better align itself with the publication
models of slashdot.org or kuro5hin.org, both of which have peer
review, yet fast publication, and large readership. Both are
electronic publication models.

I would consider an alternate paper based model, where the publication
is essentially electronic, but authors could request (at their cost) a
"bound journal issue" (printing on demand) so that they can give it to
their Dean at the time of tenure decisions.

If you have not seen it, the University of British Columbia offers a
free software for the management of electronic journals. You can find
it at http://www.pkp.ubc.ca/ and it runs on a PhP/MySQL platform
(either with Linux/Apache or with Windows + web server).