A lot of conventional journals have an open access option available to the author for a fee (usually for about $3000). What about offering a no-cost upgrade after reviewing some number of papers for the journal?
Journal generally won't do this so long as researchers continue to do it for free. In addition, even if researcher's followed the OP's advice and the journals adopted this strategy, it get us where (most people think) we should be: all research article available free to everyone.
