Each animal microRNA (miRNA) targets many genes for repression. Down-regulation of most of these targets is weak and has no detectable individual phenotypic effect. Whether this extensive weak repression is biologically relevant is a central issue in the debate on miRNA functionality. In the "small (target) pool" view, weak repression is nonfunctional and should be gradually removed during evolution. However, since the selective advantage of removing individual targets is small, testing this hypothesis is a challenge. We propose a novel approach by using miRNAs we call twin-miRs, which produce two mature products from the hairpin of the same miRNA precursor. Loss of the minor miR partner would affect all its targets and thus could be visible to selection. Since the minor miRs repress all their targets weakly, the "small pool" hypothesis would predict the elimination of twin-miRs over time. Surveying and sequencing 45 small RNA libraries in Drosophila, we found that nearly 40% of miRNAs produce twin-miRs. The minor forms are expressed in nontrivial abundance and repress their targets weakly. Interestingly, twin-miRs are often evolutionarily old, highly conserved, and comparable to solo-miRs in expression. Since there is no measurable trend toward reduction in target pool size, we conclude that at least some of the weak repression interactions are functional. A companion study using the May-Wigner theory of network stability suggests that distributed weak repression cumulatively contributes to stability of gene regulatory networks.