@jonny well the first one seems to be arguing for longer spent before doing a hypothesis test so arguably that's an even lower information rate overall. 😉 The second one seems closer but I haven't read past first page yet. Does it talk about how we could have a richer error signal?