There is a familiar motif among Court-watchers about conservative Justices migrating left over the course of their time on the bench.  This week, the website FiveThirtyEight.com ran a story of this genre under the title Is Chief Justice Roberts a Secret Liberal?  It’s a bad story, and I’ve said so by Tweetstorm.  But I’ve also seen other outlets pick up the FiveThirtyEight story and repeat the idea, so I figure it’s worth laying out the reasons why it’s bad story clearly and in a forum that’s conducive to a little bit of sustained argument.

The FiveThirtyEight story is an examination of what the story takes to be a demonstrated fact: that Roberts has moved to the left.  But the answer to the story’s titular question is no.  Roberts isn’t a liberal, secret or otherwise.  (Except in the classical/market sense, and to the extent that  Roberts is that kind of liberal, it’s never been a secret.)  But the topic may require a longer discussion.  Why do some people think that Roberts is in fact a liberal (in the prevailing political sense), having either secretly been so all along or having moved left since become Chief Justice?  Similarly, whence the more general idea that conservative Justices regularly move left? 

Where Roberts is concerned, a big part of the answer is of course the salience of the Chief Justice’s votes in the two big ACA challenges, NFIB v. Sebelius and King v. Burwell.  Where the more general question is concerned, a big part of the answer is more or less the same thing, generalized: the experience of many conservatives over the past quarter-century of being repeatedly frustrated at the Court’s unwillingness to do certain things that those particular conservatives think the Court should do (like overrule Roe, or end affirmative action, and so on), even though the Court has had a Republican-appointed majority for decades.  In short, there’s a gap between where (many) conservatives are on salient legal issues and where Roberts—or the center of the Court—is.  Roberts or the center of the Court seem, in those salient ways, to be to the left of the relevant conservatives.  And one way to explain that gap would be to say that Roberts, and other Justices who were appointed as ostensible conservatives, moved to the left after being appointed.

That’s the answer that the FiveThirtyEight story takes as factual.  And it purports to do so not just on the basis of anecdote or impression but on the basis of the thing that is the stock-in-trade of FiveThirtyEight: data.  The story presents lots of data about decisions Roberts has made and the voting coalitions he has joined—tables, graphs, and other such stuff—which it says show that Roberts has in fact moved to the left.  More generally, it says that similar data show that several Republican appointees to the Supreme Court moved left over the course of their careers. 

But I don’t think Roberts has moved to the left.  And the data presented do not prove that he has. 

The story’s analysis suffers from a huge and basic flaw. It’s more or less the same flaw that Joseph Fishkin and Cary Franklin went to town on when the New York Times ran a cognate story two years ago about the ostensible liberalism of the Court as a whole.  It’s regrettable that the point needs to be made again.  But apparently it does.  So here goes. 

In brief: Maybe Roberts isn’t moving to the left.  Maybe the questions the Court is being asked are moving farther and farther to the right.

Consider an illustration.  At Time 1, the Court hears a case in which the question presented is “Does the Second Amendment confer an individual right to own firearms?”  Justice X votes yes.  At Time 2, the Court hears a case in which the question presented is “Does the Second Amendment confer an individual right to carry automatic weapons near schools?”  Justice X votes no.  Has Justice X moved left between Time 1 and Time 2?  There’s no reason to think so.  Justice X might hold the ordinary and consistent position that the Second Amendment confers an individual right and that the scope of that individual right does not include the carrying of automatic weapons, or the carrying of weapons near schools, or the combination of the two.  But in the data on which the FiveThirtyEight story relies, the first vote would be coded “conservative” and the second vote “liberal.”  Thus “showing” that Justice X had moved to the left from Time 2 to Time 2.  Even if he hasn’t.

There’s a similar problem with the portion of the data that codes Justices as liberal or conservative based on the frequency with which they vote with other Justices.  Go back to the two hypothetical firearms cases.  In the Time 1 case, Justice X would vote with the most conservative Justices and against the most liberal ones.  In the Time 2 case, the situation would be reversed.  But Justice X hasn’t moved at all.  What’s changed is the question he is answering, not his views on the subject matter. 

So now we need to bring in a big fact that the FiveThirtyEight story (and others like it) seem not to have on screen at all.  It’s this: The questions presented in the cases that the Supreme Court decides move through ideological space over time, tracking the ideological direction of the Court itself.  As the median Justice moves right or left (that’s oversimplified, but it’ll do for schematic purposes) on a particular issue, questions that used to be contestable become settled, and other questions that used to be settled become contestable, and litigators know it, and cases are brought at the new frontier—cases that the Supreme Court would never have bothered to hear before, if anyone had even tried to litigate them.  In terms of the firearms example above: ten years ago, it wasn’t yet settled that the Second Amendment conferred an individual right at all.  We were still litigating that question.  Only after that Time 1 question is settled do the courts move on to confront questions like the Time 2 question.

In recent decades, in most (but not all) issue spaces, the questions in the Court’s cases have moved farther and farther to the right.  The Second Amendment context I’ve been using is one example.  See also affirmative action, criminal procedure, federalism, campaign finance, and takings—and that’s just in constitutional law.  (A significant exception, within constitutional law, is LGBT+ issues: more on that later.) This general rightward movement is not mysterious.  Except for the parts of the 2015 and 2016 terms when the Garland stonewall kept the Supreme Court at eight members, the Court has had a GOP-appointed majority continuously since 1970. The law has evolved as shaped by those majorities.  And as the law moves farther and farther to the right, the questions that the Court considers also move farther and farther to the right.  So Justices who are “to the right of where the law is” at Time 1 might come to straddle more of the questions at Time 2.

Put more concretely: Sure, Chief Justice Roberts didn’t read the ACA to require its own self-destruction in King v. Burwell.  But the legal theory he rejected was a highly aggressive theory.  One can be conservative and still not think THAT.  Or, to use an imprecise but serviceable analogy: Senator Jeffrey Flake isn’t a secret liberal just because he’s against Roy Moore.  The more Roy Moores the GOP nominates, the more such people Flake will oppose (I hope).  Flake still won’t be a liberal.

In sum: The data in the FiveThirtyEight story might show that Roberts is not as far right as he once was relative to the questions he’s being asked.  But those questions are moving right.  Without reading the underlying studies closely, I can’t be sure whether what I’ve said here explains all of what’s being reported about those studies or just most of it.  But when you read studies and stories like these, keep this dynamic in mind.

In my conversations on this topic, one question I’ve been asked is about Justice Alito.  If what I’m saying is right, the question goes, shouldn’t other conservative Justices also seem to be moving left?  Alito, appointed by the same President as Roberts, doesn’t seem to be moving that way.  It’s a good question, but the answer is pretty simple.  On a lot of issues, Alito is farther right than Roberts, even though both Roberts and Alito are within the ideological space that was attractive to the George W. Bush Administration when it decided whom to appoint.  A bunch of of litigated issues have now reached the zone where Roberts might come down on either side, and fewer have reached the analogous zone with Alito, because the law has to go a little farther before it gets to him.  But—the big and simple point—anyone can be the median Justice, or the median voter in any constituency, regardless of the substance of that person’s views.  All that matters is who else is in the group.  If the next two Supreme Court appointments replace Democratic appointees with people to Alito’s right, then Alito will start to look like a liberal, too—at least to the people who read data in the way the FiveThirtyEight story does.  And though these data sets don’t show Alito moving right, they do show apparent rightward movement for several other Republican appointees in the last half-century.  John Paul Stevens began as a moderate conservative and left the Court at its left-wing edge not because he radically changed his worldview but because the world, and the institution, changed around him.  (Which isn’t to say that in thirty-three years he never rethought an idea.)

This same analysis explains why we should be skeptical that Roberts’s appointment is a case of failed GOP vetting.  Vetting failures occur when something about the candidate would disqualify him if it were known to the people responsible for selection at the time of the selection decision.  In 2005, when Roberts was appointed, there was nothing wrong with him from the point of view of most committed conservatives in the American legal elite.  His commitments on the salient issues were just right.  (This is a statement made roughly: I don’t mean to be saying that his precise views fit a particular conservative template exactly, in part because there is no such precise template.)  In 2005, that group of elite conservative lawyers had no clue that seven years later they would be converging on a particular set of legal arguments which, if accepted, would destroy a statute not yet drafted by attacking particular features of that statute which were not yet objects of conservative obloquoy.  For the most part, they developed those views (in good faith, I stipulate) after Roberts was on the bench—and in significant part, Roberts developed right along with them.  He agreed with the commerce argument in NFIB, after all.  To be sure, he didn’t go all the way with the ACA challengers, and that was a big deal.  But how he would come out on the ACA as a tax wasn’t a vettable question in 2005. The vetters of 2005 couldn’t possibly have posed it: the relevant features of the law are too complex and contingent to have been imagined in the abstract.  And even if the question could have been posed in the abstract, the vetters would probably not have regarded it as an important question for a nominee to answer, unless they knew that the legal question they were asking would be presented in the context of the ACA (or some similarly opprobrious statute).  In this context, what is called a failure of vetting is really a failure to predict whether a nominee will change in the future, in a certain direction and at a certain rate that matches the direction and rate of change of a certain set of elite opinions in the party that nominated him.  Ascribing the phenomenon to vetting misses that fact and also lays blame where it doesn’t belong: it suggests that the fault lies with some combination of the vetters’ incompetence and the nominee’s sneakiness, rather than the nominee’s failure to change in the future in a way that the vetters could not possibly completely specify. (If your spouse doesn’t share a new passion that you develop after years of marriage, you’re on shaky ground asserting that he or she isn’t the person you thought he or she was when you got married.)

Finally, consider an area that I marked above as an exception to the law’s general rightward trend: LGBT+ issues.  The Court as a whole has moved dramatically to the left in that space during the last thirty years.  But the exception is itself instructive. Yes, the Court moved left in this domain in absolute terms.  But relative to mainstream opinion, it stayed pretty much the same, which is to say that the Court’s movement largely mirrored that of the country.  Leftward movement on this issue was possible in an era when the Supreme Court had Republican-appointed majorities precisely because an important slice of Republican opinion was moving left on the issue: Dick Cheney endorsed same-sex marriage six years before Anthony Kennedy did.  And as for vetting?  Kennedy wasn’t vetted for his views on same-sex marriage in 1988, and if he had been, he’d have passed with flying colors, because he had no idea that one day he’d have the view that he turned out to have in 2015. 

None of this is to say that Republican appointees never move to the left.  Over time, people sometimes rethink things.  Harry Blackmun moved left on capital punishment, and Anthony Kennedy has plausibly (if not uncontrovertibly) moved left on affirmative action.  But the big mass of data that some take to show several Republican appointees trending left over the course of a few decades would be pretty well explained by noting that during those decades, the contested space of the legal questions the Court confronts has, on most issues, moved farther and farther to the right.  Which it has. 

This observation might not ease the frustration of conservatives who would like Roberts to be as conservative as they are, or to develop farther toward the right as fast as they do.  Nor will such conservatives necessarily be comforted by the knowledge that liberals are no strangers to this kind of frustration.  A great many liberals find the Democratic appointees on the Court maddeningly tepid in their liberalism.  And if that problem stings liberals less than the parallel problem stings conservatives, it might be only because the liberals know they wouldn’t win cases even if the Justices they’re frustrated with moved farther left: a majority requires persuading at least one Republican appointee, and that’s been true since before Justice Gorsuch entered kindergarten.  Which is why the law has moved steadily to the right.  Which is why it can look like Justice Roberts is moving left.  Which I think he isn’t doing.