TORTINI

For your delectation and delight, desultory dicta on the law of delicts.

Archives »

The Relative Implausibility of Relative Plausibility

May 26th, 2025

For the moment, in the American legal academy, there seems to be a fair amount of support for the idea that the burden of proof in fact-finding is centered around a vigorous contest between the plausibility of competing stories advanced by the litigants. Professors Ronald Allen and Alex Stein, two well-respected evidence law scholars have written widely about this “relative plausibility” theory of adjudication and the burden of proof.[1] They claim to “demonstrate that factfinders decide cases predominantly by applying the relative plausibility criterion guided by inference to the best explanation … .”[2] As they see American courtroom practice, the norm is “the relative plausibility mode of factfinding involving a rigorous comparison between the parties’ stories about the individual event.”[3] They insist that their “theory aligns with ordinary people’s natural reasoning.”[4]

I am not so sure. Semantically, the authors’ choice of the term “plausibility” is curious. Plausibility in ordinary usage has only a tenuous relationship with epistemic warrant. Allen and Stein acknowledge that in law (as in science and in life), the “coin of the realm is truth.” Plausibility in epistemology and philosophy of science, however, is typically treated as a weak and often irrelevant factor in assessing the correctness of a factual (scientific) claim. A leading textbook of epidemiology, for instance, offers that

“[a] causal explanation is plausible if it appears reasonable or realistic within the context in which the hypothesized cause and its effect occur. *** Plausibility can change as the context evolves and will be misleading when current understanding is misleading or wrong.”[5]

A plausible fact is not the same as a known fact, or a well-established fact. In his oft-cited after-dinner speech, Sir Austin Bradford Hill, for instance, acknowledged that plausibility is a helpful but non-necessary consideration in evaluating an association for causality, but he cautions that plausibility

“is a feature I am convinced we cannot demand. What is biologically plausible depends on the biological knowledge of the day.”

Since Hill, many scientific writers have relegated plausibility to a limited role in assessing the correctness of a causal claim. In the language of a recent effort to modernize Hill’s factors, one author noted that “plausibility and analogy do not work well in most fields of investigation, and their invocation has been mostly detrimental.”[6]

To be sure, Allen and Stein do, in some places, make clear that they take plausibility to mean more than some uninformed Bayesian prior. Relative plausibility in their view thus has some connection to the coin of the realm. At least, fact finders are seen as considering more than the usual extent of plausibility; they must “determine which of the parties’ conflicting stories makes most sense in terms of coherence, consilience, causality, and evidential coverage.”[7]

Relativity of Plausibility

If we are truly concerned with “naturalism” (ordinary reasoning?) then we should give some weight to how people react to a claim that has support. In the real world, real people are implicitly aware of Brandolini’s Law. Inspired by his reading of Daniel Kahneman’s Thinking, Fast and Slow,[8] Alberto Brandolini articulated the “Bullshit Asymmetry Principle,” in a 2013 peer-reviewed tweet. According to Brandolini, The work needed to refute bullshit is [at least] an order of magnitude greater than the work needed to produce it.

Despite its crude name, the Bullshit Asymmetry Principle has a respected intellectual provenance. In 1845, economist Frédéric Bastiat expressed an early notion of the adage:

“We must confess that our adversaries have a marked advantage over us in the discussion. In very few words they can announce a half-truth; and in order to demonstrate that it is incomplete, we are obliged to have recourse to long and dry dissertations.”[9]

Recognition of Bullshit Asymmetry actually goes back to ancient times. The Roman lawyer and teacher of rhetoric, Marcus Fabius Quintilianus, known as Quintilian to his friends, addressed the principle in his Institutio Oratoria:

“The task of the accuser is consequently straightforward and, if I may use the phrase, vociferous; but the defence requires a thousand arts and stratagems.”[10]

Quintilian’s insight explains why most people, without invoking economic efficiency or grand moral theories, believe it is natural to place the burden of proof upon the accuser or the pursuer, as opposed to the defender.

The demands made upon us by claims and “stories,” often frivolous, result in our naturally evaluating the warrant (or plausibility if you insist) for a claim before we become mired down in assessing “relative plausibility.” Courts have developed procedural mechanisms, such as summary adjudication and expert witness gatekeeping, to avoid unnecessary detailed assessments of relative plausibility. Even when cases are submitted to the factfinder, the decision may be made solely on “plausibility” of the story proffered by the party with the burden of proof.

The framing of adjudication and the burden of proof as relative plausibility seems to contradict what often happens in American courtrooms. Frequently, the defense does not put forward a “story,” but attempts to show that the plaintiff’s story is rubbish. Indeed, litigation may end without the plausibility of the defense position ever being considered. In the litigation over health claims involving exposure to Agent Orange, Judge Jack Weinstein granted summary judgment because the plaintiffs’ medical causation case was weak and insufficient.[11] The defense may have been even weaker in terms of its “coherence, consilience, causality, and evidential coverage,” but the party with the burden of proof attracts the first round of critical scrutiny in the summary judgment process. In Agent Orange, Judge Weinstein found the plaintiffs’ “proofs” to be rather crummy, without regard for the strength or weakness of the defendants’ evidence.

Similarly, Judge Weinstein granted summary judgment on plaintiffs’ claims of systemic disease injury in the silicone gel breast implant litigation. In granting judgment, Judge Weinstein pretermitted the defendants’ motions to exclude plaintiffs’ expert witnesses on grounds of Rules 702 and 703. Again, without comparison with the defendants’ “story,” Judge Weinstein found the plaintiffs’ story to be insufficient.[12]

The situation in Agent Orange and in Silicone Gel often obtains in trial itself. Defendants are often unable to disprove the plaintiffs’ claim, and the law does not require them to do so. When the evidentiary display is insufficient to support a claim, it may well be insufficient to show the claim is false. Defendants may want to be able to establish their own “story,” but the best they may have to offer is a showing that the plaintiffs’ story is not credible. Allen and Stein suggest that “[t]heoretically, a defendant can simply deny the plaintiff’s complaint, … but this virtually never occurs.”[13] They cite to no empirical evidence in claiming that “a rigorous comparison between the parties’ stories about the individual event is the norm in American courtrooms.”[14]

As shown by the summary judgment examples above, the burden of proof means something quite different from Allen and Stein’s contention about relative plausibility. Before the advent of expert witness gatekeeping, one of the few ways that a party could challenge an adversary’s expert witness was to object that the plaintiff’s medical expert witnesses offered conflicting opinions on a key factual issue in dispute. In Pennsylvania, dismissals for inconsistent expert witness opinion testimony is known as the “Mudano rule,” for a 1927 Pennsylvania Supreme Court case that held that “there must be no absolute contradiction in their essential conclusions.”[15] The Mudano rule arises because the plaintiff must furnish consistent evidence on key issues, even though the jury could otherwise freely choose to accept some or all or none of the inconsistent expert witnesses’ testimony. The Mudano rule requires dismissal without regard to the “plausibility” or implausibility of the defense case.[16]

The Mudano rule follows from the plaintiff’s having the burden of proof. When the party with the burden of proof proffers two conflicting opinions, the guesswork is simply too palpable for an appellate court to tolerate. The rule does not apply to the defense case, should the defense mange to proffer two inconsistent expert witnesses on a key issue raised by plaintiff’s case.[17]

The party without the burden of proof on causation or other key issue requiring expert witness testimony need not present any expert testimony. And if the opposing party does present expert witnesses, the law does not require that they be as precise or certain as those presented by the party with the burden. As one work-a-day appellate court put the matter:

“Absent an affirmative defense or a counterclaim, the defendant’s case is usually nothing more than an attempt to rebut or discredit the plaintiff’s case. Evidence that rebuts or discredits is not necessarily proof. It simply vitiates the effect of opposing evidence. Expert opinion evidence, such as that offered by [the defendant] in this case, certainly affords an effective means of rebutting contrary expert opinion evidence, even if the expert rebuttal would not qualify as proof.”[18]

A defendant need not engage in “story telling” at all; it may present an expert witness to testify that the plaintiff’s causation claim is bogus, even if the alternatives are merely possible. This statement of law with respect to the required certitude of expert witnesses and the burden of proof comes from a Pennsylvania case, but it appears to be the majority rule.[19]

The asymmetry created by the epistemic requirements of the burden of proof undermines the simplistic model of a court, or jury, deciding the “relative plausibility” of a claim.

Jury Instructions

The typical jury instruction on expert witness opinion testimony also shows that the burden of proof may operate without the head-to-head comparison of “stories,” as suggested by Allen and Stein. Under the law of most states, the trier of fact is free to accept some, all, or none of an expert witness opinion. In New Jersey, for instance, jurors are instructed that they

“are not bound by the testimony of an expert. You may give it whatever weight you deem is appropriate. You may accept or reject all or part of an expert’s opinion(s).”[20]

The practice of American courts with respect to burden of proof does not support the reductionist formula offered by the evidence law scholars. Burden of proof has implications in terms of summary judgment and directed verdict practice, which seem glossed over by “relative plausibility.” Furthermore, in situations in which the factfinder assesses both parties’ stories for relative plausibility, it must reject the story from the party with the burden of proof, when that party fails to show its story is more likely than not correct, even when the opponent’s story has a lesser plausibility.

Cases almost always involve incomplete evidence, and so we should expect that evidential warrant, or relative plausibility, or posterior probability of both sides’ cases to be less than complete or 100 percent.  If the party with the burden has a story with 40% probability, and then opponent’s story has a 30% probability, the case still results in a non-suit.

[1] Ronald J. Allen & Alex Stein, “Evidence, Probability, and the Burden of Proof,” 55 Ariz. L. Rev. 557 (2013) [cited herein as Allen & Stein].  They are not alone in endorsing relative plausibility, but for now I will key my observations to Allen and Stein’s early paper on relative plausibility.

[2] Id. at 4. Allen and Stein cite the classical proponents of inference to the best explanation, but they do not in this 2013 article describe or defend such inferences in detail. See Peter Lipton, Inference to the Best Explanation (2d ed. 2004); Gilbert H. Harman, “The Inference to the Best Explanation,” 74 Philosophical Rev. 88 (1965).

[3] Allen & Stein at 14.

[4] Allen & Stein at 15.

[5] Tyler J. VenderWeele, Timothy L. Lash & Kenneth J. Rothman, “Causal Inference and Scientific Reasoning,” chap. 2, in Timothy L. Lash, et al., Modern Epidemiology 17, 20 (4th ed. 2021).

[6] Louis Anthony Cox, Jr., “Modernizing the Bradford Hill criteria for assessing causal relationships in observational data,”  48 Crit. Rev. Toxicol. 682, 684 (2018).

[7] Allen & Stein at 1.

[8] Daniel Kahneman, Thinking, Fast and Slow (2011).

[9] Frédéric Bastiat, Economic Sophisms (1845), in The Bastiat Collection vol. 1, t 172 (2007).

[10] Quintilian, Institutio Oratoria, book V, chapters 13-14 (Butler transl. 1920).

[11] In re Agent Orange Product Liab. Litig., 597 F. Supp. 740, 785 (E.D.N.Y. 1984), aff’d 818 F.2d 145, 150-51 (2d Cir. 1987) (approving district court’s analysis), cert. denied sub nom. Pinkney v. Dow Chemical Co., 487 U.S. 1234 (1988); In re “Agent Orange” Prod. Liab. Litig., 611 F. Supp. 1223 (E.D.N.Y. 1985), aff’d, 818 F.2d 187 (2d Cir. 1987), cert. denied, 487 U.S. 1234 (1988). See Peter H. Schuck, Agent Orange on Trial: Mass Toxic Disasters in the Courts (1987).

[12] In re Breast Implant Cases, 942 F. Supp. 958 (E.& S.D.N.Y. 1996).

[13] Allen & Stein at 12.

[14] Allen & Stein at 14.

[15] Mudano v. Philadelphia Rapid Transit Co., 289 Pa. 51, 60, 137 A. 104, 107 (1927).

[16] See Daniel E. Cummins, “The ‘Mudano’ Rule: Conflicting Expert Opinions Often Prove Fatal,” The Legal Intelligencer (Mar. 16, 2017). See also Brannan v. Lankenau Hospital, 490 Pa. 588, 596, 417 A.2d 196 (1980) (“a plaintiff’s case will fail when the testimony of his two expert witnesses is so contradictory that the jury is left with no guidance on the issue”); Menarde v. Philadelphia Transportation Co., 376 Pa. 497, 501, 103 A.2d 681 (1954). See also Halper v. Jewish Family & Children Services of Great of Philadelphia, 600 Pa. 145, 963 A.2d 1282, 1287-88 (2009).

[17] See Kennedy v. Sell, 816 A.2d 1153, 1159 (Pa. Super. 2003).

[18] Neal v. Lu, 365 Pa. Super. 464, 530 A.2d 103, 109-110 (1987); see also Jacobs v. Chatwani, 2007 Pa. Super. 102, 922 A.2d 950, 958-960 (2007) (holding that defense expert witnesses are not required to opinion to reasonable medical certainty). See generally James Beck, “Reasonable Certainty and Defense Experts,” Drug & Device Law (Aug. 4, 2011).

[19] Jordan v. Pinamont, 2007 WL 4440900, at *2 (E.D. Pa. May 8, 2007) (“Defendants are entitled to inform the jury of other medical conditions which reasonably could have caused Plaintiff’s complaints, even if it cannot be stated to a reasonable degree of medical certainty that Defendants’ proffered alternatives were, in fact, the cause”); Johnesee v. The Stop & Shop Co., 174 N.J. Super. 426, 416 A.2d 956, 959 (N.J. Super. App. Div. 1980) (holding that defense expert witness may criticize plaintiff’s expert witness’s opinion as unfounded even though he can offer only possible alternative causes); Holbrook v. Lykes Bros. Steamship Co., 80 F.3d 777, 786 (3d Cir. 1996) (affirming admission of defense expert testimony that plaintiff had failed to exclude radiation as a possible cause of his mesothelioma, but reversing judgment for the defense on other grounds); Wilder v. Eberhart, 977 F.2d 673, 676-77 (1st Cir. 1992) (applying New Hampshire law); Allen v. Brown Clinic, P.L.L.P., 531 F.3d 568, 574-75 (8th Cir. 2008) (applying South Dakota law).

[20] N.J. CHARGE 1.13, citing State v. Spann, 236 N.J. Super. 13, 21 (App Div. 1989). See Pennsylvania’s Suggested Standard Civil Jury Instructions. PA. SSJI (Civ), § 4.100, § 4.80 (2013) (providing that the jury is not required to accept an expert witness’s testimony); Nina Chernoff, Standard jury instruction in New York on expert testimony (2023) (“You may accept or reject such testimony, in whole or in part, just as you may with respect to the testimony of any other witness.”).

Posted in Scientific Evidence, Trial Procedure | No Comments »

SCOTUS Embraces Totalitarian Version of Unitary Executive Doctrine

May 25th, 2025

Earlier this year, President Felonious Trump fired Gwynne Wilcox, a member of the National Labor Relations Board (NLRB), and Cathy A. Harris, the chair of the Federal Merit Systems Protection Board (MSPB). Wilcox sued, and the district court in the District of Columbia enjoined the President from removing her. Federal statutory law prohibits the president from removing the plaintiff officers except for cause. See 29 U. S. C. § 153(a); 5 U. S. C. § 1202(d). Felonious Trump identified no such cause.

Felonious Trump took the case up to the Supreme Court, not on the merits, but on the propriety of the interlocutory stay. Last week, the Court reversed the stay, and remanded the case to be heard on its merits.[1]  But in addressing the stay, the Court necessarily addressed the merits, as it was required to do, in assessing the likelihood of success. The net result was that the Court signaled its embrace of an imagined unitary and totalitarian conception of the president’s power.

In an unsigned opinion, the Court went beyond its limited role in addressing the interlocutory injunction, when it declared:

“Because the Constitution vests the executive power in the President, see Art. II, § 1, cl. 1, he may remove without cause executive officers who exercise that power on his behalf.”[2]

This assertion begged the question whether these officers were the sort of executive officers whom the president can remove at will. The Court’s reasoning was unusually anemic. The Court assumed that the two fired officials exercise “considerable executive power,” which thus puts them at the mercy of the president. Of course, the whole point of independent agencies is that they exercise hybrid executive-legislative-judicial power. The Court briefly acknowledged that there are “exceptions,” but left for the litigation’s outcome to address them.[3]

Wilcox and Harris invoked the specter of Trump’s removing, without cause, members of the Federal Reserve’s Board of Governors. Felonious Trump has intimated he wanted to fire Jerome Powell, but has for the while backed off his threats, after the stock market tanked in response to his palavering. The Court went out of its way to distinguish the Federal Reserve as “a uniquely structured, quasi-private entity that follows in the distinct historical tradition of the First and Second Banks of the United States.”[4] All the independent agencies, however, are uniquely structured and designed so as not to be under the complete control of the president.

In dissent, Justice Elena Kagan, joined by Justices Sotomayor and Jackson, blasted her colleagues for using the court’s emergency docket to make a significant statement about an issue with implications for the fundamental structure of the federal government. The order, she wrote, essentially allows Trump to overrule the court’s 1935 decision in Humphrey’s Executor [5] “by fiat,” at least until the justices eventually take up the case. In the meanwhile, the Court takes up the surrender monkey posture of some of our leading law firms.

Justice Kagan understandably emphasized the abandonment of precedent in obeisance to Trump, as well as its ignoring and abandoning the legitimate interests of Congress in fashioning an independent agency. Justice Kagan also noted that the Court’s ruling on the interlocutory stay threw out “the presumption of constitutionality which attaches to every Act of Congress,” as somehow unworthy of consideration in an equitable analysis of the stay.

Although Justice Kagan enjoyed the better grounded argument, the dissent did not go far enough. Her dissent never actually identified the full range of executive branch interests at stake. The NLRB and the MSPB were created by statutes, passed by Congress, but which presidents signed.

In 1935, Congress passed the National Labor Relations Act to advance a policy encouraging collective bargaining. The Act was signed into law by President Franklin Roosevelt. The NLRB was created by the Act, passed by Congress, and made law by the executive.

Similarly, the Merit Systems Protection Board is an independent, quasi-judicial agency that protects the federal merit system. The Board was created by statute, and was codified by the Civil Service Reform Act of 1978, Public Law No. 95-454. This legislation was signed into law by President Jimmy Carter.

Executive power was involved in creating both boards, and in specifying the protections that board members would enjoy in doing their jobs. The firings by Felonious Trump were thus in derogation of executive power, as much as they were of legislative power. So once again, the Court has allowed Felonious Trump to ignore law that previous presidents, who exercised their executive power to create the for-cause protections. The executive may be unitary, but it is bound by laws that it enacts in conjunction with the legislature. The logic of the Court’s position implies that there is no limit to a president’s power to ignore statutory law.

[1] Trump v. Wilcox, ___ S.Ct.___, 605 U. S. ____ (May 22, 2025) [Slip op.]. See Nina Totenberg, “Supreme Court allows Trump to fire members of independent agency boards — for now,” Nat’l Public Radio (May 22, 2025).

[2] Slip op. at 1.

[3] Id. (“Because the Constitution vests the executive power in the President, he may remove without cause executive officers who exercise that power on his behalf, subject to narrow exceptions recognized by our precedents.”).

[4] Slip at 2.

[5] Humphrey’s Executor v. United States, 295 U. S. 602 (1935).

Posted in Uncategorized | No Comments »

Judging Science Symposium

May 25th, 2025

While waiting for the much delayed fourth edition of the Reference Manual on Scientific Evidence, you may want to take a look at a recent law review issue on expert witnesses issues. Back in November 2024, the Columbia Science & Technology Law Review held its symposium at the Columbia Law Review on “Judging Science.” The symposium explored current judicial practice for, and treatment of, scientific expert witness testimony in the United States. Because the symposium took place at Columbia, we can expect any number of antic proposals for reform, as well.

Among the commentators on the presentations were Hon. Jed S. Rakoff, Judge on the Southern District of New York,[1] and the notorious Provost David Madigan, from Northeastern University.[2]

The current issue (vol. 26, no.2) of the Columbia Science and Technology Law Review, released on May 23, 2025, contains papers originally presented at the symposium:

Edith Beerdsen, “Unsticking Litigation Science.”

Edward Cheng, “Expert Histories.”

Shari Seidman Diamond & Richard Lempert, “How Experts View the Legal System’s Use of Scientific Evidence.”

David Faigman, “Overcoming Judicial Innumeracy.”

Maura Grossman & Paul Grimm, “Judicial Approaches to Acknowledged and Unacknowledged AI-Generated Evidence.”

Valerie Hans, “Juries Judging Science.”

Enjoy the beach reading!

[1] See Schachtman, “Scientific illiteracy among the judiciary,” Tortini (Feb. 29, 2012).

[2] See, e.g., In re Accutane Litig., No. 271(MCL), 2015 WL 753674 (N.J. Super., Law Div., Atlantic Cty., Feb. 20, 2015) (excluding plaintiffs’ expert witness David Madigan); In re Incretin-Based Therapies Prods. Liab. Litig., 524 F. Supp. 3d 1007 (S.D. Cal. 2021), aff’d, No. 21-55342, 2022 WL 898595 (9th Cir. Mar. 28, 2022) (per curiam). Provost Madigan is stepping down from his position next month. Sonel Cutler, Zoe MacDiarmid & Kate Armanini, “Northeastern Provost David Madigan to step down in June,” The Huntington News (Jan. 16, 2025).

Posted in Expert Witnesses, Rule 702, Scientific Evidence, statistical evidence | No Comments »

Can Lawyers Sink Lower Than Plagiarizing a Bot?

May 16th, 2025

Several years ago, I submitted a brief, which I had written, in a New York case. When a co-defendant’s counsel filed the same brief, without acknowledging that it was plagiarised, I was annoyed. It seemed to me that such plagiarism clearly has professional and general ethical implications, especially if the plagiarists charged clients for writing something that they stole from another person.[1]

Legal culture does to some extent encourage plagiarism. Law students work as research assistants for professors and often write segments of legal treatises and hornbooks that are published under their professors’ names. Presumably the law students are satisfied that their work was used and that they received strong recommendations in future job searches. When bright young law school graduates accept judicial clerkships, they understand that their writing for draft opinions or memoranda will be “cannibalized” by their judges at will. Similarly, young law firm associates know that much of their writing may be used in briefs without attribution or signature lines on the brief. This practice goes too far, in my view, when partners require associates to draft articles for publication without making them authors, and with at most an anemic note of gratitude for “assistance.”

When courts use lawyers’ arguments and their actual language advanced in briefs, lawyers rarely complain. At least there are no complaints from the plagiarized lawyers who prevailed.

Sometimes the plagiarized argument reveals the source of a court’s error. In an opinion issued over Justice Sotomayor’s name, the Supreme Court adopted wholesale an argument advanced by the Solicitor General. The origin of the argument was unmistakable because the claim was so egregiously wrong. In its amicus brief, the Solicitor General argued that statistical significance was unnecessary for reaching a conclusion of causation between the use of Zicam and anosmia. In support of its argument, the Solicitor General’s amicus brief cited three cases: Best, Westberry, and Ferebee.[2] The three cited cases all involved disputes over specific causation, whereas the case sub judice purported to involve an issue of general causation. (The Court correctly decided that the corporate disclosure issue under the securities laws did not actually require general causation.) Nevertheless, the Solicitor General’s sloppy legal scholarship was incorporated into the Supreme Court’s opinion. Justice Sotomayor repeated the citation to the first two cases, dropped the reference to Ferebee, but curiously added an even more bizarre third case to the argument by citing the infamous Wells case.[3] Remarkably, as notorious and poorly reasoned as the Wells case was,[4] it involved plaintiffs’ expert witnesses’ reliance upon at least one poorly conducted study that reported nominal statistical significance. And thus the Supreme Court produced the following text, with three inapposite cases:

“We note that courts frequently permit expert testimony on causation based on evidence other than statistical significance. See, e.g., Best v. Lowe’s Home Centers, Inc., 563 F. 3d 171, 178 (6th Cir 2009); Westberry v. Gislaved Gummi AB, 178 F. 3d 257, 263–264 (4th Cir. 1999) (citing cases); Wells v. Ortho Pharmaceutical Corp., 788 F. 2d 741, 744–745 (11th Cir. 1986). We need not consider whether the expert testimony was properly admitted in those cases, and we do not attempt to define here what constitutes reliable evidence of causation.”[5]

I suspect that a law clerk acted as an intermediary, or a vcctor, in the plagiarism by incorporating the Solicitor General’s argument into a draft or a memorandum, which became part of the opinion that none of the justices considered carefully. At least, the three cases cited by the Solicitor General and the cases cited by Justice Sotomayor actually existed.

The introduction of large language models in artificial intelligence (A.I.) has taken plagiarism to a new level.[6] Lawyers can now prompt a machine to review the case law on a specified issue and to create an output of analysis in favor of an identified litigation position. And then the lawyers can submit the output to a court as their own work, and charge their clients for the brief writing. The submission of a brief or memorandum to a court, however, implies that the facts are supported, and that the legal precedents cited exist and are pertinent authorities for the court to consider.

The problem with the use of AI, however, is that AI “hallucinates” non-existent cases, all with citations, case specifics, and sometimes fanciful quotes that seem pertinent. It sucks to be a bot, but it is even worse for lawyers who take AI output, and present it as their own, without further research, in hopes of persuading a court.

Last week, Special Master Magistrate Judge Michael R. Wilner (retired) imposed a monetary sanction of $31,100 against the lawfirm K&L Gates for failing to fact check a brief prepared in substantial part by AI, as well as failing to disclose its use of AI, or to correct errors after notified of the Special Master’s concerns.[7] The Special Master noted that “no reasonably competent attorney should out-source research and writing to this technology – particularly without any attempt to verify the accuracy of that material.”[8] Ouch. K&L Gates is not the first firm, and it sadly will not be the last firm to be sanctioned for the improvident use of AI, and the submission of fraudulent authorities to a court.[9]

The Special Master’s outrage was generated by his discovery that one third of the legal citations were incorrect, and that two of them were non-existent. Several quotations from cases presented in support of the plaintiffs’ argument were entirely bogus.[10] Although the Special Master freely described the lawyers’ actions as “misconduct,” he concluded that disciplinary sanctions against the lawyers involved was unwarranted.[11]

The Special Master’s opinion elided the ethical significance of the plaintiffs’ lawyers’ conduct. The opinion never mentions the Rules of Professional Responsibility; nor does it suggest that a reference to the California State Bar’s Office of Chief Trial Counsel was in order. Plagiarism is, after all, research misconduct in most other professional domains. Under the regulations of the Office of Research Integrity (ORI), plagiarism is “the appropriation of another person’s ideas, processes, results, or words without giving appropriate credit.”[12] An AI model may not be eligible for copyright, or the moral rights of authors, but it seems that the spirit of the prohibition against plagiarism requires disclosure and credit to AI, however bogus AI’s contribution may have been.

Although I tend to think of plagiarism in briefs and other court submissions as a violation of a lawyers’ professional responsibility, the issue actually is not clear cut. Plagiarism would seem to violate core professional responsibilities of honesty and integrity. In disparaging plagiarism, courts have occasionally invoked Model Rule 8.4(c), which prohibits engaging in “dishonesty, fraud, deceit or misrepresentation.”[13] Legal commentators have divided over the propriety of recycling legal arguments and verbatim language without acknowledgments.[14] Context also matters. The legal analysis in a routine motion in a mass tort litigation, say for dismissal for lack of diversity, or for change of venue, probably should not be re-invented. Copying another lawyer’s appellate brief and and its consideration of a legal issue, without acknowledgment, seems ethically dodgy.

Of course, copying, whether permissible or not, does not mean that lawyers are freed from their ethical responsibility of ensuring that citations and interpretations of authorities are correct. Without having taken steps to ensure the relevance and correctness of cited authorities, lawyers cannot represent to the court that their argument has a good-faith basis in law and fact.

Hallucinations versus Delusions

Philosophy professor Hilarius Bookbinder (not Søren Kierkegaard) points out that describing bogus AI output as “hallucinations” is euphemistic and erroneous. When AI makes stuff up, the output is not really an hallucination, but delusional.[15]

Bookbinder channels the insight of William James, who gave extensive consideration to the phenomenon of hallucinations, and ultimately characterized them as correct reports of altered consciousness.[16] In a footnote to The Principles of Psychology, James offered the following helpful distinction:

“Illusions and hallucinations must both be distinguished from delusions. A delusion is a false opinion about a matter of fact, which need not necessarily involve, though it often does involve, false perceptions of sensible things. We may, for example, have religions delusions, medical delusions, delusions about our own importance, about other peoples’ characters, etc., ad libitum.”[17]

Bookbinder has a point about how we talk about large language models of AI. In James’ parlance, AI does not really hallucinate, but it clearly suffers delusions; or perhaps it simply fabricates unwittingly. This feature, or flaw, of AI makes lawyers’ uncritical reliance upon AI for legal research and writing not only unethical, not merely for plagiarizing, but for violating their professional duties of competence and candor to the tribunal. Outsourcing thinking to a machine seems unbecoming for a profession that is built upon careful, independent analysis.

Hallucinations and delusions are both distinguishable from a third phenomenon, bullshit – or willful indifference to the truth.[18] For bull shitters, the assertion is more important than the truth value of the statement. When Felonious Trump claimed a Civil War battle took place on one of his golf courses, and even went so far as to identify the site with a plague. Several historians pushed back, and pointed out, his mistake, to which Trump asked “How would they know that?” Were they there?”[19] Trumpian lies and bullshit have now begun to infiltrate into the judicial system, so we must sort delusions, bullshit, and lies among the pathology of lawyering. We can probably say fairly that AI lacks the intentionality to deceive, or the psychopathology that confuses assertion with fact.

There may well be conduct worse than plagiarizing a bot, but that is hardly a recommendation.

[1] Schachtman, “Copycat – Further Thoughts on Plagiarism in the Law,” Tortini (Oct. 24, 2010); “Plagiarism in the Law,” Tortini (Oct. 16, 2010).

[2] Brief for the United States as Amicus Curiae Supporting Respondents, in Matrixx Initiatives, Inc. v. Siracusano, 2010 WL 4624148, at *14-16 (Nov. 12, 2010) (“Best v. Lowe’s Home Centers, Inc., 563 F.3d 171, 178 (6th Cir. 2009) (“an ‘overwhelming majority of the courts of appeals’ agree” that differential diagnosis, a process for medical diagnosis that does not entail statistical significance tests, informs causation) (quoting Westberry v. Gislaved Gummi AB, 178 F.3d 257, 263 (4th Cir. 1999)),” and “Ferebee v. Chevron Chem. Co., 736 F.2d 1529, 1536 (D.C. Cir.) (‘[P]roducts liability law does not preclude recovery until a “statistically significant” number of people have been injured’.), cert. denied, 469 U.S. 1062 (1984)).

[3] Wells v. Ortho Pharmaceutical Corp., 615 F. Supp. 262 (N.D. Ga. 1985), aff’d in relevant part, 788 F.2d 741 (11th Cir.), cert. denied, 479 U.S.950 (1986).

[4] See, e.g., James L. Mills and Duane Alexander, “Teratogens and ‘Litogens’,” 15 New Engl. J. Med. 1234 (1986); Samuel R. Gross, “Expert Evidence,” 1991 Wis. L. Rev. 1113, 1121-24 (1991) (“Unfortunately, Judge Shoob’s decision is absolutely wrong. There is no scientifically credible evidence that Ortho-Gynol Contraceptive Jelly ever causes birth defects.”). See also Editorial, “Federal Judges v. Science,” N.Y. Times, December 27, 1986, at A22 (unsigned editorial) (“That Judge Shoob and the appellate judges ignored the best scientific evidence is an intellectual embarrassment.”);  David E. Bernstein, “Junk Science in the Courtroom,” Wall St. J. at A 15 (Mar. 24,1993) (pointing to Wells as a prominent example of how the federal judiciary had embarrassed American judicial system with its careless, non-evidence based approach to scientific evidence); Bert Black, Francisco J. Ayala & Carol Saffran-Brinks, “Science and the Law in the Wake of Daubert: A New Search for Scientific Knowledge,” 72 Texas L. Rev. 715, 733-34 (1994) (lawyers and leading scientist noting that the district judge “found that the scientific studies relied upon by the plaintiffs’ expert were inconclusive, but nonetheless held his testimony sufficient to support a plaintiffs’ verdict. *** [T]he court explicitly based its decision on the demeanor, tone, motives, biases, and interests that might have influenced each expert’s opinion. Scientific validity apparently did not matter at all.”) (internal citations omitted); Troyen A. Brennan, “Untangling Causation Issues in Law and Medicine: Hazardous Substance Litigation,” 107 Ann. Intern. Med. 741, 744-45 (1987) (describing the result in Wells as arising from the difficulties created by the Ferebee case; “[t]he Wells case can be characterized as the court embracing the hypothesis when the epidemiologic study fails to show any effect”).  Kenneth R. Foster, David E. Bernstein, and Peter W. Huber, eds., Phantom Risk: Scientific Inference and the Law 28-29, 138-39 (MIT Press 1993) (criticizing Wells decision); Hans Zeisel & David Kaye, Prove It With Figures: Empirical Methods in Law and Litigation § 6.5, at 93(1997) (noting the multiple comparisons in studies of birth defects among women who used spermicides, based upon the many reported categories of birth malformations, and the large potential for even more unreported categories); id. at § 6.5 n.3, at 271 (characterizing Wells as “notorious,” and noting that the case became a “lightning rod for the legal system’s ability to handle expert evidence.”).

[5] Matrixx Initiatives, Inc. v. Siracusano, 131 S.Ct. 1309, 1319 (2011).

[6] Schachtman, “Artificial Intelligence May Be Worse Than None At All,” Tortini (Feb. 2, 2025); “Hallucinations in Law and in Government,” Tortini (Feb. 19, 2025).

[7] Lacey v. State Farm General Insurance Co., Case 2:24-cv-05205-FMO-MAA, doc. 119 (C.D. Calif. May 6, 2025).

[8] Id. at 7. The Special Master ducked the more interesting counter-factual question: what if AI had generated a perfect brief, with exactly the right citations, properly cited, with all inferences and conclusions proper and correct? Would the outsourcing of human intelligence be acceptable to us, with all the requisite disclosures? See Hilarius Bookbinder, “Why AI is Destroying Academic Integrity: It’s because students prefer The Experience Machine,” Scriptorium Philosophia (Jan. 02, 2025).

[9] See note 1, supra. See also Mata v. Avianca, 678 F. Supp. 3d 443 (S.D.N.Y. 2023) (ordering sanctions against lawyers who submitted briefs with six fabricated judicial opinions and fake quotes); Gauthier v. Goodyear Tire & Rubber Co., civil action no. 1:23-CV-281 (E.D. Tex Nov. 25, 2024) (imposing sanctions for citing non-existing cases with fabricated quotations). Canadian lawyers have also been seduced by the prospect of outsourcing their thinking, researching, analyzing, and writing to a bot. Ko v. Li, CV-25-00736891-00ES, 2025 Ontario Super. Ct. Justice 2766 (May 1, 2025) (ordering lawyer to show cause why she should not be held in contempt). See Bernise Carolino, “Ko v. Li, Ontario Superior Court, 2 nonexistent case citations, attorney referred for potential contempt proceedings,” Canadian Lawyering (May 13, 2025).

[10] Kat Black, “‘A Collective Debacle’: Ellis George and K&L Gates Ordered to Pay $31,000 after Using AI to Write Brief in Insurance Case,” Legal Intelligencer (May 13, 2025).

[11] Lacey, supra, at 7; Eugene Volokh, “AI Hallucination in Filings Involving 14th-Largest U.S. Law Firm Lead to $31K in Sanctions,” The Volokh Conspiracy, Reason (May 13, 2025).

[12] 42 CFR 93.103 (c) Department of Health and Human Services, Office of Research Integrity. See, e.g., Sena Chang, “Secretary of Defense Pete Hegseth ’03 ‘plagiarized’ small portions of his senior thesis, experts say. But how serious is it?” Daily Princetonian (May 10, 2025).

[13] See In re Mundie, 453 Fed.Appx. 9 (2d Cir. 2011) (publicly reprimanding lawyer for various acts, including copying extensively from another lawyer’s brief in a different case). See also In re Summit Financial, Inc., 2021 WL 5173331 (Bankr. C.D. Cal. Nov. 5, 2021); Lohan v. Perez, 924 F.Supp. 2d 447 (E.D.N.Y. 2013). Compare New York City Bar Formal Opinion 2018-3 (disapproving “extensive” copying, while noting that copying source material without attribution is not “per se” an ethical violation) with North Carolina State Bar Formal Ethics Opinion 2008-14 (acknowledging that lawyers may copy language used in other briefs without attribution).

[14] Dennis A. Rendleman, “Copy That!: What is plagiarism in the practice of law?,” YourABA (Mar. 2020) (arguing that unacknowledged copying in legal filings is different from such conduct in scholarly publications); Andrew M. Carter, “The Case for Plagiarism,” 9 U. Calif. Irvine L. Rev. 531 (2019); Carol M. Bast & Linda B. Samuels, “Plagiarism and legal scholarship in the age of information sharing: the need for intellectual honesty,” 57 Catholic Univ. L. Rev. 777 (2008); Peter A. Joy & Kevin C. McMunigal, “The Problems of Plagiarism as an Ethics Offense,” ABA Criminal Justice 56 (Summer 2011).

[15] Hilarius Bookbinder, “Hallucination, bullshit, confabulation: AI and the outsourcing of thinking,” Scriptorium Philosophia (May 15, 2025).

[16] William James, The Principles of Psychology, vol. 2 (ch. 18-19) (1918).

[17] Id. at ch. 19, n. 41.

[18] Harry Frankfurt, On Bullshit 63 (2005) (“Bullshit is unavoidable whenever circumstances require someone to talk without knowing what he is talking about.  Thus the production of bullshit is stimulated whenever a person’s obligations or opportunities to speak about some topic exceed his knowledge of the facts that are relevant to that topic.”).

[19] Joyce Chen, “Donald Trump’s Golf Course Plaque Honors Fake Civil War Battle,” Rollingstone (Aug. 17, 2017); Nicholas Fandos, “In Renovation of Golf Club, Donald Trump Also Dressed Up History,” N.Y. Times (Nov. 24, 2015).

Posted in Scientific Misconduct, Trial Procedure | No Comments »