“…The best hypothesis is the one that minimizes the length of the description of the hypothesis plus the length of the description of the data when encoded or compressed with the help of that hypothesis. Scientific inference based on some version of this idea has been applied to many biological problems, including sequence alignment (Allison et al, 1999(Allison et al, , 2000Allison and Yee, 1990), phylogeny reconstruction (Cheeseman and Kanefsky, 1993;Li et al, 2001;Milosavljevic et al, 1990;Otu and Sayood, 2003;Ren et al, 1995), and comparison of classifications (Day, 1983), though not to collections of trees constructed from separate data sets. In this framework, a hypothesis is either a tree or a collection of trees and the best hypothesis is the one that permits the maximum compression of that hypothesis plus the sequence alignment.…”