by on 
By Michael J. Hasday*
When the State of California and Planned Parenthood recently sued the Trump Administration over regulations implementing an abortion gag rule,[1] they must have thought they had a good chance before the famously liberal Ninth Circuit Court of Appeals. But lady luck was not on their side. Although the majority of judges on the Ninth Circuit were appointed by Democratic presidents, the random draw turned up three Republican-appointed judges for the panel hearing the case.[2] The three Republican appointees then promptly ruled for Trump.
In an apparent attempt to reverse the rogue decision, the majority of active judges on the Ninth Circuit authorized en banc review.[3] But because the Ninth Circuit has so many judges, it alone among the federal appellate courts employs a limited en banc review, where only eleven of its active judges sit en banc, consisting of the chief judge and ten other active judges who are randomly selected.[4] Accordingly, limited en banc allows for “minority rule” in a subset of cases,[5] such as when the random draw results in six or more judges holding minority views for that particular case.[6] And that is what appears to have happened here: even though 15 of the 26 active judges were Democratic appointees, the limited en banc panel consisted of seven Republican and four Democratic appointees.[7] The limited en banc panel subsequently voted 7-4, upholding the three-judge panel decision, with all votes on “party” lines.[8]
Although a few Ninth Circuit judges have criticized the limited en banc procedure because of the possibility of minority decisions,[9] most of the judges on the Ninth Circuit have apparently accepted the limited en banc procedure based on the “Churchill democracy defense”: it is the worst system, except for all others.[10] For example, Ninth Circuit Chief Judge Sidney Thomas has defended the limited en banc procedure by pointing out that, while the rules allow for a full en banc review after a limited en banc decision, the Ninth Circuit has never exercised this option.[11] In a similar vein, Thomas refers to an experiment the Ninth Circuit conducted with a more robust limited en banc panel, with 15 judges instead of 11, intended to bring more “representativeness” to the limited en banc panel. The experiment proved to be unpopular with both the Ninth Circuit judges and the bar as being too unwieldy.[12] Finally, the nuclear option for replacing the limited en banc procedure—splitting the Ninth Circuit into two or three smaller circuits—is opposed by nearly all the judges on the Ninth Circuit, Democratic and Republican appointees alike.[13] However, there is another way that the Ninth Circuit can fix its limited en banc representativeness problem while retaining the efficiency of the current procedure—a method I introduce here[14] and name the “Grouping Method.”
The Grouping Method
The Grouping Method retains all the key elements of the current limited en banc procedure—eleven active judges on the panel, the chief judge guaranteed placement, and the other active judges having an equal chance to be selected—but slightly tweaks how the random draw occurs. Instead of a true random draw, some person (e.g., a court administrator), persons (e.g., a committee of experts), or thing (e.g., a computer via an algorithm)[15] places all the active judges, with the exception of the chief judge, into two equal groups. Let’s call them Group A and Group B.[16] The limited en banc panel is then composed of the chief judge and five randomly selected judges from Group A and five randomly selected judges from Group B.
The person or persons or thing (hereinafter, the “Grouper”) is instructed/programmed to place the active judges into the two groups based on how it predicts the active judges will rule in the particular case. In other words, if the Ninth Circuit has 28 active judges other than the chief judge, the Grouper should place the 14 judges perceived to be most favorable to Party A into Group A and the 14 judges perceived to be most favorable to Party B into Group B. In making these placements, the Grouper can make individualized assessments based on how the judges have previously ruled in similar cases, or even in the actual case if the judge was on the original three-judge panel. The Grouper need not wholly, or even mostly, rely on the proxy of the appointing president’s political party, although the Grouper could certainly also factor in this proxy, particularly for newly appointed judges.
Playing With The House’s Money
The beauty of the Grouping Method is that there is tremendous upside in using it with effectively no downside. As long as the Grouper makes reasonably accurate groupings, limited en banc panels will be significantly more likely to reflect majority views of the Ninth Circuit. Moreover, even if the Grouper makes mistakes in the groupings, the Grouping Method could hardly do worse than the current procedure. The following example makes this clear. Let’s say that for a given case, the chief judge favors Party A, 15 of the other active judges favor Party A, and 13 of the other active judges favor Party B. Under the current limited en banc procedure, there would be a 24.91 percent chance for a panel with a majority of judges favoring Party B although that’s the minority position for the Ninth Circuit.[17] For the purposes of this Essay, I will define this number as the “error rate.”
Under the Grouping Method, if the Grouper is completely accurate in its groupings, it will guarantee a panel with a majority of the judges favoring Party A, i.e., a zero-percent error rate.[18] However, even if the Grouper makes mistakes in the groupings, i.e. by placing one or more Party A judges into Group B, the error rate of the resulting panels will still be better than the current procedure, unless the Grouper completely botches the groupings, which would then result in an error rate roughly the same as the current procedure. The following chart shows the “error rate” for all the possible number of mistakes:
Number of Mistakes Error Rate
Zero
(Group A contains 14 Party A Judges, 0 Party B Judges 0%
Group B contains 1 Party A Judge, 13 Party B Judges)
One
(Group A contains 13 Party A Judges, 1 Party B Judge 14.13%
Group B contains 2 Party A Judges, 12 Party B Judges)
Two
(Group A contains 12 Party A Judges, 2 Party B Judges 19.38%
Group B contains 3 Party A Judges, 11 Party B Judges)
Three
(Group A contains 11 Party A Judges, 3 Party B Judges 22.20%
Group B contains 4 Party A Judges, 10 Party B Judges)
Four
(Group A contains 10 Party A Judges, 4 Party B Judges 23.88%
Group B contains 5 Party A Judges, 9 Party B Judges)
Five
(Group A contains 9 Party A Judges, 5 Party B Judges 24.83%
Group B contains 6 Party A Judges, 8 Party B Judges)
Six
(Group A contains 8 Party A Judges, 6 Party B Judges 25.28%
Group B contains 7 Party A Judges, 7 Party B Judges)
From this chart, we can see that the biggest gains in reduced error rate come when the Grouper makes zero, one, or two mistakes. If the Grouper makes three, four, or five mistakes, the error rate will only be marginally better than the current procedure (24.91%). And if the Grouper makes the maximum number of mistakes—six—the error rate will be only marginally worse, by about one-third of one percent. In sum, no matter how bad a job the Grouper does, it can hardly do worse than the current procedure. Accordingly, any concern about possible strategic manipulation of the limited en banc panel is remote. Because even if the Grouper wanted to manipulate the composition of the limited en banc panels by enhancing the probability of minority views on the panel,[19] the Grouper could effectively do no better than what is currently accepted.
Motive, Means, and Opportunity
The Democratic appointees on the Ninth Circuit should be highly motivated to implement the Grouping Method. At the start of the Trump Administration, Democratic appointees held an 18-7 advantage on the Ninth Circuit.[20] However, the advantage is now only 16-12,[21] and may be cut further to 16-13 if the Trump Administration is able to fill the remaining vacancy. A 16-13 ratio will place a majority of Republican appointees on about 1 in 4 limited en banc panels. This statistical fact has excited one prominent conservative commentator who stated that “the Ninth Circuit’s unique ‘limited en banc panel’ mechanism means that a sizeable increase in the number of Republican appointees can dramatically improve Ninth Circuit law even if that number falls short of a majority.”[22] However, what excites this conservative commentator may alarm liberals. In crude political terms, why should liberals allow conservatives to exercise power via this “statistical loophole” when the whole point of an en banc review is to ensure that the circuit’s majority view prevails in the most important and divisive cases?[23] The political stakes are high considering that decisions from the Ninth Circuit impact about one in five Americans.[24]
The Democratic appointees also have the means to implement the Grouping Method. The Ninth Circuit has full authority to implement this reform itself by local rule,[25] i.e., by majority vote of the active judges.[26] There is no requirement that limited en banc panels be selected in a random fashion. Indeed, the Ninth Circuit’s limited en banc procedure already departs from strict randomness as the chief judge is automatically placed on every limited en banc panel.
Finally, in a political environment where leading Democrats are discussing “court-packing” ideas for the U.S. Supreme Court,[27] this relatively modest proposal may be too good an opportunity to miss.
© 2019, Michael J. Hasday.
* Michael J. Hasday is an attorney practicing in New York.
[1] See California v. Azar, Nos. 19-15974, 19-15979, 19-35386, 19-35394, 2019 U.S. App. LEXIS 18452 (9th Cir. June 20, 2019).
[2] See Ed Whelan, Re: Ninth Circuit Victory for Trump Administration Defunding of Planned Parenthood, Bench Memos (July 15, 2019, 11:28 AM), https://www.nationalreview.com/bench-memos/re-ninth-circuit-victory-for-trump-administrations-defunding-of-planned-parenthood/.
[3] See California v. Azar, Nos. 19-15974, 19-15979, 19-35386, 19-35394, 2019 U.S. App. LEXIS 19971 (9th Cir. July 3, 2019).
[4] LAURAL HOOPER ET AL., FED. JUDICIAL CTR., CASE MANAGEMENT PROCEDURES IN THE FEDERAL COURTS OF APPEALS 36 (2d ed. 2011); Ninth Circuit Rule 35-3.
[5] See Arthur D. Hellman, Getting It Right: Panel Error and the En Banc Process in the Ninth Circuit Court of Appeals, 34 U.C. Davis L. Rev. 425, 462 (2000) (“almost certainly, some departures from majority rule have occurred” in the limited en banc procedure); Pamela Ann Rymer, The “Limited” En Banc: Half Full, or Half Empty?, 48 Ariz. L. Rev. 317, 321 (2006) (“experience does show that anomalies occur in the limited en banc procedure that could not occur in a true en banc”); Written Testimony of Diarmuid F. O’Scannlain, United States Circuit Judge, United States Court of Appeals for the Ninth Circuit, before the United States Senate, Committee on the Judiciary 12 (July 31, 2018) (the limited en banc procedure “presents the quite real possibility that some of the circuit’s most important or divisive issues will be decided by a substantial minority of judges, who do not represent the true views of our court”).
[6] In this Essay, I will make the reasonable assumption that a panel with a majority of judges holding the minority view will end up reaching the minority-view decision.
[7] See Whelan, supra note 2.
[8] See California v. Azar, Nos. 19-15974, 19-15979, 19-35386, 19-35394, 2019 U.S. App. LEXIS 20587 (9th Cir. July 11, 2019). It did not help matters from the plaintiffs’ perspective that three Democratic appointees had recused themselves from the case.
[9] See, e.g., Rymer, supra note 5; O’Scannlain, supra note 5.
[10] See Rymer, supra note 5, at 317 (noting that “lawyers and judges are willing, as a practical matter, to accept the decisions of a limited en banc panel as authoritative” because there is “no realistic alternative in the Ninth Circuit” due to its size).
[11] See WRITTEN TESTIMONY OF CHIEF JUDGE SIDNEY R. THOMAS, CHIEF JUDGE, UNITED STATES COURT OF APPEALS FOR THE NINTH CIRCUIT, BEFORE THE UNITED STATES SENATE, COMMITTEE ON THE JUDICIARY 51 (Aug. 24, 2017) (“[T]he court has never voted to rehear a case en banc before the full court. This statistic is a testament to the success of the limited en banc court model.”)
[12] See id. at 53-54.
[13] See Letter from 34 Ninth Circuit Judges to the U.S. Senate Committee on the Judiciary (July 27, 2017), https://indivisibleeb.org/wp-content/uploads/2017/08/grassley-feinstein-ltr-7-27-17.pdf
[14] The method is a simplified variation of the “rank-order method” I previously developed. See Michael J. Hasday, The Rank-Order Method for Appellate Subset Selection, 93 Notre Dame Law. Rev. Online 17 (2017). The simplified variation is tailor-made for the Ninth Circuit’s limited en banc procedure because the rule which guarantees placement for the chief judge solves the problem of selecting a final judge for an odd-numbered panel if the full set of judges are divided into two groups. Additionally, the relatively large size of the limited en banc panel makes the individual differences between the judges placed in the same group less important to the overall ideological composition of the panel.
[15] Computers are already performing this type of “judicial prediction” work to advise clients of their “odds of winning” based on the particular three-judge panel assigned at the Federal Circuit. See Roy Strom, The Algorithm Will Hire Your Patent Lawyer Now, Bloomberg Law, Aug. 1, 2019, https://biglawbusiness.com/the-algorithm-will-hire-your-patent-lawyer-now.
[16] If the number of active judges, excluding the chief judge, is odd, then one of the active judges should be eliminated by lot before the sorting process begins.
[17] All statistical calculations for this Essay are on file with the author.
[18] Of course, if the chief judge is in the minority, then the error rates will be higher under both the current limited en banc procedure and the Grouping Method.
[19] This concern would be even more remote if the Grouper is a computer.
[20] See Ed Whelan, Improving the Ninth Circuit, Bench Memos (May 18, 2018, 10:41 AM), https://www.nationalreview.com/bench-memos/improving-the-ninth-circuit/.
[21] See UNITED STATES COURTS FOR THE NINTH CIRCUIT, THE JUDGES OF THIS COURT IN ORDER OF SENIORITY (Updated July 2019), https://www.ca9.uscourts.gov/content/view_seniority_list.php?pk_id=0000000035.
[22] See Ed Whelan, Re: Improving the Ninth Circuit, Bench Memos (Oct. 25, 2018, 2:29 PM), https://www.nationalreview.com/bench-memos/re-improving-the-ninth-circuit/
[23] See United States v. American-Foreign S.S. Corp., 363 U.S. 685, 689-90 (1960) (“The principal utility of determinations by the courts of appeals in banc is to enable the court to maintain its integrity as an institution by making it possible for a majority of its judges always to control and thereby to secure uniformity and continuity in its decisions. . . .”) (quoting Albert Branson Maris, Hearing and Rehearing Cases in Banc, in 14 F.R.D. 91, 96 (1954)).
[24] See Dylan Matthews, How the 9th Circuit Became Conservatives’ Least Favorite Court, Vox (Jan 10, 2018 2:30 PM) (“The Ninth covers a population of 64.3 million people, according to the latest Census data, or nearly 20 percent of the US.”)
[25] The relevant statute authorizing limited en banc panels only states that a federal appellate circuit with more than 15 active judges may “perform its en banc function by such number [of judges] as may be prescribed by rule of the court of appeals.” Pub. L. No. 95-486, §6, 92 Stat. 1629.
[26] See FRAP 47(a)(1) (“Each court of appeals acting by a majority of its judges in regular active service may, after giving appropriate public notice and opportunity for comment, make and amend rules governing its practice.”).
[27] See Philip Elliot, The Next Big Idea in the Democratic Primary: Expanding the Supreme Court? Time, March 13, 2019, https://time.com/5550325/democrats-court-packing/.