A new informational base for social choice

As soon as Arrow (1951) had shown us the impossibility of designing 'good' social choice procedures based solely on individual ordinal preference rankings, the idea was developed of improving collective choice by gathering and using better information than just those rankings. Cardinal preferences and extended preferences have been favored alternatives. One common thread in this literature is the idea that using only rankings keeps us from employing knowledge about 'intensities' sities' of preference to improve choices. The get-more-information approach has been unhelpful in the sense that you still get impossibility theorems that, though different in detail, still bear a family resemblance to those obtained for procedures based only on rankings of alternatives (see Kelly, 1978). Roughly, the greater information available at each point of the domain is offset by the much larger domain over which the procedure must operate. This paper illustrates these ideas again using a new informational base.