Daniel Kokotajlo comments on Some thoughts on David Roodman’s GWP model and its relation to AI timelines