18 RUCHO v. COMMON CAUSE KAGAN, J., dissenting too far. Consider again Justice Kennedy’s hypothetical of mapmakers who set out to maximally burden (i.e., make count for as little as possible) the votes going to a rival party. See supra, at 12. Does the majority really think that goal is permissible? But why even bother with hypotheticals? Just consider the purposes here. It cannot be permissible and thus irrelevant, as the majority claims, that state officials have as their purpose the kind of grotesquely gerrymandered map that, according to all this Court has ever said, violates the Constitution. See supra, at 13. On to the second step of the analysis, where the plaintiffs must prove that the districting plan substantially dilutes their votes. The majority fails to discuss most of the evidence the District Courts relied on to find that the plaintiffs had done so. See ante, at 23–24. But that evidence—particularly from North Carolina—is the key to understanding both the problem these cases present and the solution to it they offer. The evidence reveals just how bad the two gerrymanders were (in case you had any doubts). And it shows how the same technologies and data that today facilitate extreme partisan gerrymanders also enable courts to discover them, by exposing just how much they dilute votes. See Vieth, 541 U. S., at 312–313 (opinion of Kennedy, J.) (predicting that development). Consider the sort of evidence used in North Carolina first. There, the plaintiffs demonstrated the districting plan’s effects mostly by relying on what might be called the “extreme outlier approach.” (Here’s a spoiler: the State’s plan was one.) The approach—which also has recently been used in Michigan and Ohio litigation— begins by using advanced computing technology to randomly generate a large collection of districting plans that incorporate the State’s physical and political geography and meet its declared districting criteria, except for partisan gain. For each of those maps, the method then uses