Behavioral economics shows that people don't always have consistent preferences. Can we still do welfare economics without being too controlling? The 'New Consensus' approach has problems, but a new way of thinking about welfare measurement might help. Bernheim and Rangel's framework could work well with this new approach.