The proposal by Adam Back does reduce the size of ring signature. In case of large ringsig it maybe roughly two times smaller. However, from asymptotic point of view there is no difference from CryptoNote implementation: it is still a linear dependence O(n).