If you took each IPCC report and made every number 20% worse (larger or smaller depending on the context), would you end up with a better set of predictions?