Apparently, when I pitched the idea of replacing backtracking with forking (with lazy side effecting of course) a ~year ago, I was reinventing STEPS work from 2008: http://www.vpri.org/pdf/rn2008003_john.pdf