Who Changed on the Court
Justices are appointed for life on the assumption that we know what we are getting. Here is how far 27 of them moved — and how often it was the Court that moved instead.
Harry Blackmun changed more than any justice of the modern era — 32.8 points, from Warren Burger’s ally to the Court’s conscience on the death penalty. Hugo Black made the same journey backwards. But the more interesting finding is how many famous “drifters” never drifted: William Brennan’s own voting moved 1.4 points in thirty-four years. He stood still, and four Republican presidents rebuilt the Court to his right.
What is being measured
Counting how often a justice cast a liberal vote tells you very little on its own, because the cases change. A Court hearing mostly criminal-procedure appeals in 1965 and mostly administrative-law disputes in 1995 will produce different numbers from identical judges. So every figure on this page is a gap: the justice’s liberal-vote share minus the share for the other justices deciding the same cases. Same docket, same day, same nine-member sample.
A line above the rule means voting liberal more often than the colleagues; below, less often. Zero is not neutrality — it is agreement with the bench of the day. Drift is how far that gap moved between a justice’s first five terms and the last five.
The gap fixes the docket problem but creates another one, and it is the reason this page is built the way it is: a gap can move because the justice moved or because the bench did. Every figure is therefore reported with both halves — how far the justice’s own share moved, and how far the colleagues’ did. Where they disagree, the page says so.
Six Justices
In every chart the darker horizontal rule is level with the bench. Above it, the justice was voting liberal more often than the colleagues deciding the same cases; below it, less often.
Harry A. Blackmun§98
The one everybody means
Richard Nixon appointed Harry Blackmun in 1970 as Warren Burger’s reliable ally. The two had been friends since childhood in St. Paul, and the press called them the Minnesota Twins. Linda Greenhouse, working from Blackmun’s own papers after his death, found that in his first five terms he voted with Burger 87.5 percent of the time in closely divided cases and with the liberal William Brennan 13 percent. By his final years those figures had very nearly traded places: Brennan 70.6 percent, Burger 32.4.
He left the Court in 1994 having written, dissenting from a denial of certiorari in Callins v. Collins, that “from this day forward, I no longer shall tinker with the machinery of death.” Both measures agree about Blackmun, which is rarer than it sounds: his own voting moved 26.9 points while the bench beneath him moved only 6.0 the other way. He is the case where the man, not the company, changed.
Hugo L. Black§76
The liberal who ended up on the other side
Hugo Black arrived from the Senate in 1937 and spent two decades as the Court’s great liberal. He finished it voting with the conservatives. He dissented in Griswold v. Connecticut (1965), refusing to find a right of privacy the Constitution did not name; dissented in Bell v. Maryland (1964); wrote the 5–4 majority in Adderley v. Florida (1966) upholding trespass convictions of civil-rights demonstrators, over dissents from Warren, Douglas, Brennan and Fortas; and dissented in Tinker v. Des Moines (1969) when the Court protected students wearing armbands.
The philosophy never moved. Black’s First Amendment absolutism — “no law means no law” — covered pure speech and not symbolic conduct, so as the law expanded to protect conduct, a fixed judge produced an increasingly conservative record. Here the decomposition splits the difference honestly: his own share fell 12.3 points, and the bench moved 8.0 points the other way. Both things happened.
John Paul Stevens§101
The Republican who kept saying he had not moved
Gerald Ford named John Paul Stevens in 1975, and in Regents of the University of California v. Bakke (1978) he wrote an opinion joined by the Court’s most conservative members. He retired in 2010 as the Court’s senior liberal, having maintained throughout that his own views had held still while the Court traveled.
The data only half agrees. Stevens’s own liberal share rose 12.5 points between his first five terms and his last five, while the bench moved 1.7 points the other way. On these numbers he did move — and moved further than the Court around him.
William J. Brennan Jr.§90
The one who did not move at all
By the net figure William Brennan is the fourth-largest drifter on the Court, 14.2 points more liberal relative to his colleagues at the end than at the beginning. Almost none of it is him. His own liberal share rose 1.4 points across thirty-four terms — within the noise. The bench moved 12.8 points to his right.
Brennan did not become a liberal lion. He stood exactly where he had always stood while Nixon, Ford, Reagan and Bush rebuilt the Court around him, and the distance between them opened up. The same is true of Thurgood Marshall, whose 15.6-point gap is roughly two parts bench to one part justice. This decomposition is our own; the standard scholarship measures drift differently and classes both men as left-trending.
David H. Souter§105
The stealth nominee
David Souter was George H. W. Bush’s 1990 nominee precisely because he had no paper trail to attack, and conservatives expected a reliable vote. He gave them one for about two terms. His own liberal share then climbed 11.8 points while the bench moved 2.4 points the other way, and he retired in 2009 voting with the Court’s liberals.
Among the scholarship he is sometimes called the new Blackmun. The numbers put him a little short of that — Blackmun moved more than twice as far — but in the same direction, and for a shorter tenure.
Clarence Thomas§106
The immovable object
Thirty-four terms, and the gap between Clarence Thomas and the justices sitting with him is 0.7 points wider at the end than at the beginning — less than the year-to-year noise in any single term. His own share moved 0.6 points. The bench moved 0.1.
He is not alone: Lewis Powell (−0.3) and the second Justice Harlan (−1.0) are just as fixed. The scholarship agrees about Thomas, listing him among the handful of justices who show no meaningful change at all across a long tenure.
Where this number lies
Three justices come out of the arithmetic pointing the wrong way, and they are worth naming rather than burying.
His own liberal share barely moved — down 3.1 points across sixteen terms — while the bench moved 4.7 points to his left as Goldberg, Fortas and Marshall arrived. The net figure reads as a turn to the right. What actually happened is that his colleagues caught up with him. The standard scholarship classes Warren as left-trending.
The largest personal move on the whole page: Clark’s own liberal share rose 14.5 points. But the Warren Court he sat on moved 17.8 points in the same direction and faster, so the gap closed and the net figure comes out negative. Read the decomposition, not the headline.
A net figure of +5.4 reads as a drift left. His own share did rise 3.6 points — but the comparison group was being rebuilt around him by Thomas, Roberts and Alito, which drags a relative measure. Epstein and colleagues, using a different method, place Scalia among the justices trending right.
In each case the net figure is arithmetically correct and substantively backwards. That is a property of any relative measure, and it is why the decomposition sits beside every number here rather than in a footnote.
Every Justice, Most Moved to Least
All 27 justices with at least 12 terms on the modern record. Every chart shares the same vertical scale, ±40 points, so they can be read against one another. “Him” is the change in the justice’s own liberal share; “bench” is the change in the colleagues’.
13 more justices appear in the database with tenures too short to measure this way — among them Goldberg, Fortas and, for now, Gorsuch, Kavanaugh, Barrett and Jackson. Five terms at each end of a career needs a career.
Method, and its limits
Built from the Supreme Court Database (2025 Release 01, justice-centered), which covers the 1946 through 2024 terms and classifies every vote as liberal, conservative, or not classifiable. One vote per justice per case. Cases the database declines to code, and cases where it records a justice on both sides of different issues, are excluded.
The labels are the database’s, not ours. “Liberal” is a term of art defined issue by issue — pro-union in labor cases, anti-business in economic cases, pro-United-States in tax cases — and the six areas covering criminal procedure, civil rights, the First Amendment, due process, privacy and attorneys share a single combined definition. The full definitions are on the database’s codebook page.
Lines are smoothed with a centered five-term window, pooling cases rather than averaging rates. That window is clipped at the start and end of a tenure, so the first and last points on any line rest on fewer cases than the middle — Douglas’s final point on 283 cases against 693 mid-career. The vertical scale is set at a round ±40 points rather than at the observed maximum, so no single thin endpoint sets the scale the others are drawn against.
The strongest objection to all of this comes from Michael Bailey, who argues that scores built out of case outcomes are partly circular — votes are used to build a measure of ideology, which is then used to explain votes — and that apparent movement can come from changes in which cases reached the Court rather than from any change in a justice. He also notes that the database’s own coding can cut against political intuition: a vote to uphold campaign-spending limits is recorded as conservative because it restricts speech. Readers who want the numbers taken further should start with Epstein, Martin, Quinn and Segal, Ideological Drift Among Supreme Court Justices: Who, When, and How Important?, 101 Northwestern University Law Review 1483 (2007), which finds drift to be the norm rather than the exception and reaches its conclusions with a different method (Martin-Quinn scores) that agrees with this page about Blackmun, Black and Thomas and disagrees about Warren and Scalia. Linda Greenhouse published a reply in the same review’s Colloquy.
Blackmun’s agreement figures are from Linda Greenhouse, Becoming Justice Blackmun (2005), drawn from his papers. Case citations: Griswold v. Connecticut, 381 U.S. 479 (1965); Bell v. Maryland, 378 U.S. 226 (1964); Adderley v. Florida, 385 U.S. 39 (1966); Tinker v. Des Moines, 393 U.S. 503 (1969); Regents of the University of California v. Bakke, 438 U.S. 265 (1978); Callins v. Collins, 510 U.S. 1141 (1994).
Questions
Do Supreme Court justices become more liberal over time?
More often than not, on this measure. Of the 27 justices since 1946 with a long enough record to measure, 17 ended further to the left of their colleagues than they began and 10 ended further to the right. But the pattern is weaker than the famous cases suggest: only 7 moved more than ten points, and 6 moved less than four, which is close to the year-to-year noise in a single term.
Which Supreme Court justice changed the most?
Harry Blackmun, by a wide margin. Appointed in 1970 as Warren Burger’s ally, he ended his tenure 32.8 points further to the left of his colleagues than he began — more than twice the shift of the next-largest mover, and unusual in that almost all of it was his own voting rather than a change in the bench around him.
Did Hugo Black become conservative?
On the record, yes, relative to the justices sitting with him. Black arrived in 1937 as the Court’s great liberal and finished it dissenting in Griswold and Tinker and writing the majority against civil-rights demonstrators in Adderley v. Florida. Scholars generally attribute this to a fixed judicial philosophy rather than a change of politics: his First Amendment absolutism protected pure speech but not symbolic conduct, so as the law expanded, a stationary judge produced an increasingly conservative voting record.
Do justices drift because of the job or because the Court changes around them?
Both, and this page separates them. Blackmun, Stevens, Souter, Kennedy and Sotomayor moved mostly on their own. Brennan, Rehnquist, Burger and Frankfurter barely moved at all — the bench moved around them, opening or closing the distance without any change of heart. Reporting only the net figure credits the justice for what was really a change in his colleagues.
Every justice’s full voting record, broken out by issue area, sits on their own profile.