It should be possible to disable the reducer completely. In full-blown Hadoop, this is often useful when running large jobs where computation is only done in the map phase. Disabling the reducer does, however, bypass the sort phase, so this might cause confusion about the map-reduce model.