It's actually not a good idea, as it results in inconsistent values being returned by LuceneSearchQuery.getLuceneSort() and LuceneSearchResult.getTopDocs()}: when the sort is non-null but contains only a sort by descending score, we end up returning top docs while "scoreDocs" are {{ScoreDoc instances instead of the expected FieldDoc instances. This in turn complicates the (user) implementation of merging top docs.