On belief and reason: why we should trust the projections of global warming by climate models