There's an argument that since currently funding for subscribing to HSS journals comes from the university library there is clearly money there for HSS, and that somehow we just need to redirect the subscription cost toward publication costs. But that's not an easy nut to crack, and it's also not clear why the library should even be involved in the publishing process in that way.
A per-article subscription/purchase model would be more appropriate if we consider that the cost of peer-provided knowledge is a hard cost similar to reagents and buffers.
But I personally agree with Open Access advocates, cost should be as close to free as reasonably possible.
The landscape may have changed in the intervening years, with blanket institutional subscriptions making paper subscriptions irrelevant.
I also suspect that, much like with albums and music, nobody actually reads a journal cover to cover these days, instead they are consuming the papers they want in a random-access fashion as a collection of PDF's.