dissenting:
Suppose you work for a securities firm. Your boss drafts an email message and tells you to send the email on his .behalf to two clients. You promptly send the emails to the two clients without thinking too much about the contents of the emails. You note in the emails that you are sending the message “at the request” of your boss. It turns out, however, that the message from your boss to the clients is false and defrauds the clients out of a total of $15,000. Your boss is then sanctioned by the Securities and Exchange Commission (as is appropriate) for the improper conduct.
What about you? For sending along those emails at the direct behest of your boss, are you too on the hook for the securities law violation of willfully making a false statement or willfully engaging in a scheme to defraud?
According to the SEC, the answer is yes. And the SEC concludes that your behavior—in essence forwarding emails after being told to do so by your boss— *597warrants a lifetime suspension from the securities profession, on top of a monetary-fine.
That is what happened to Frank Lorenzo in this case. The good news is that the majority opinion vacates the lifetime suspension. The bad news is that the majority opinion—invoking a standard of deference 'that, as applied here, seems akin to a standard of “hold your nose to avoid the stink”—upholds much of the SEC’s decision on liability. I would vacate the SEC’s conclusions as to both sanctions and liability. I therefore respectfully dissent.
[[Image here]]
The SEC initiated an enforcement action against Frank Lorenzo and his boss. The boss eventually reached a settlement agreement with the SEC. Apparently thinking he had done little wrong by merely sending emails to two clients at the request of his boss, Lorenzo did not settle.
The case then proceeded through three stages: a trial before an SEC administrative law judge, review by the Commission itself, and then review by this Court. To understand my disagreement with the majority opinion, it is necessary to describe all three acts in this drama.
Act One: The Administrative Law Judge
The case proceeded to trial before an administrative law judge. This was not your usual trial. Surprisingly, the SEC did not present testimony from Lorenzo’s boss or from anyone else at the securities firm where Lorenzo worked. Instead, only Lorenzo testified about the extent of his involvement in drafting and sending the emails.
After hearing Lorenzo’s testimony and weighing his credibility, the judge concluded that Lorenzo’s boss had “drafted” the emails in question and that Lorenzo’s boss had “asked” Lorenzo to send the emails to two clients. ALJ Op. at 5 (Dec. 31, 2013), J.A. 906. The judge also concluded that Lorenzo did not read the text of the emails and that Lorenzo “sent the emails without even thinking about the contents.” Id. at 7, J.A. 908; see id. at 9, J.A. 910 (“Had he taken a minute to read the text ... ”). Furthermore, the judge noted that the emails themselves expressly stated that they were being sent at “the request” of Lorenzo’s boss. Id. at 5, J.A. 906.
Those factual findings were very favorable to Lorenzo and should have cleared Lorenzo of any serious wrongdoing under the securities laws. At most, the judge’s factual findings may have shown some mild negligence on Lorenzo’s part. The judge, however, went much further than that. The judge somehow concluded that those findings of fact demonstrated that Lorenzo willfully violated the securities laws—meaning that Lorenzo acted with an intent to deceive, manipulate, or defraud. (A finding- of willfulness, as opposed to a finding of negligence, matters because it subjects a defendant to much higher penalties.) As a sanction, the judge not only fined Lorenzo, but also imposed a lifetime suspension that prevents Lorenzo from ever again working in the securities industry.
The administrative law judge’s factual findings and legal conclusions do not square up. If Lorenzo did not draft the emails, did not think about the contents of the emails, and sent the emails only at the behest of his boss, it is impossible to find that Lorenzo acted “willfully.” That is Mens Rea 101. Establishing that a defendant acted willfully in this context requires proof at least of the defendant’s “intent to deceive, manipulate, or defraud.” Dolphin & Bradbury, Inc. v. SEC, 512 F.3d 634, 639 (D.C. Cir. 2008) (internal quotation marks omitted). How could Lorenzo have *598intentionally deceived the clients when he did not draft the emails, did not think about the contents of the emails, and sent the emails only at his boss’s direction?
The administrative law judge’s decision in this case contravenes basic due process. A finding that a defendant possessed the requisite mens rea is essential to preserving individual liberty. See, e.g., Morissette v. United States, 342 U.S. 246, 250-51, 263, 72 S.Ct. 240, 96 L.Ed. 288 (1952); see also United States v. Burwell, 690 F.3d 500, 527 (D.C. Cir. 2012) (en banc) (Kavanaugh, J., dissenting); United States v. Moore, 612 F.3d 698, 703 (D.C. Cir. 2010) (Kavanaugh, J., concurring); Bluman v. FEC, 800 F.Supp.2d 281, 292 (D.D.C. 2011) (three-judge panel). As Justice Jackson explained: “The contention that an injury can amount to a crime only when inflicted by intention is no provincial or transient notion. It is as universal and persistent in mature systems of law as belief in freedom of the human will and a consequent ability and duty of the normal individual to choose between good and evil. A relation between some mental element and punishment for a harmful act is almost as instinctive as the child’s familiar exculpatory ‘But I didn’t mean to.’ ” Morissette, 342 U.S. at 250-51, 72 S.Ct. 240 (footnote omitted).
The administrative law judge’s opinion in this case did not heed those bedrock mens rea principles. Given the judge’s pro-Lorenzo findings of fact, a legal conclusion that Lorenzo “willfully” violated the securities laws makes a hash of the term “willfully,” and of the deeply rooted principle that punishment must correspond to blameworthiness based on the defendant’s mens rea.
Act Two: The Securities and Exchange Commission
Fast forward to the Securities and Exchange Commission, which heard the appeal of the administrative law judge’s decision. Surely the Commission would realize that the administrative law judge’s factual findings did not support the judge’s legal conclusions and sanctions?
And indeed, the Commission did come to that realization. But instead of vacating the order against Lorenzo, the Commission did something quite different and quite remarkable. In a Houdini-like move, the Commission rewrote the administrative law judge’s factual findings to make those factual findings correspond to the legal conclusion that Lorenzo was guilty and deserving of a lifetime suspension.
Recall what the administrative law judge found: that Lorenzo’s boss “drafted” the emails, that Lorenzo did not think about the contents of the emails, and that Lorenzo sent the emails only after being asked to do so by his boss. ALJ Op. at 5, J.A. 906. The judge reached those conclusions only after hearing Lorenzo testify and assessing his credibility in person.
Without hearing from Lorenzo or any other witnesses, the Commission simply swept the judge’s factual and credibility findings under the rug. The Commission concluded that Lorenzo himself was “responsible” for the emails’ contents. In the Matter of Francis V. Lorenzo, Securities Act Release No. 9762, Exchange Act Release No. 74836 at 16 (Apr. 29, 2015), J.A. 930. How did the Commission magically explain its decision to discard the administrative law judge’s findings of fact? Easy. In a footnote, the Commission said that it did not need to “blindly” accept the administrative law judge’s factual findings and credibility judgments. Id. at 16 n.32, J.A. 930 n.32. Voila.
The Commission’s handiwork in this case is its own debacle. Faced with incon- • venient factual findings that would make it hard to uphold the sanctions against Lorenzo, the Commission—without hearing *599any testimony—simply manufactured a new assessment of Lorenzo’s credibility and rewrote the judge’s factual findings. So much for a fair trial.
Act Three: This Court
Fast forward to this Court. To its credit, the majority opinion rightly concludes that Lorenzo did not “make” the statements in the emails for purposes of Rule 10b-5(b) liability. See Janus Capital Group, Inc. v. First Derivative Traders, 564 U.S. 135, 131 S.Ct. 2296, 180 L.Ed.2d 166 (2011). And the majority opinion, also to its credit, vacates the grossly excessive lifetime suspension of Lorenzo and sends the case back to the SEC for reconsideration of the appropriate penalties.
So far, so good. But applying what it calls “very deferential” review, the majority opinion upholds the finding of liability against Lorenzo under Section 10(b), Rule 10b-5(a) and (c), and Section 17(a). Maj. Op. 583, 588-92. The majority opinion does so on the ground that Lorenzo willfully engaged in a scheme to defraud even though he did not “make” the statements in the emails.
I disagree on two alternative and independent grounds with the majority opinion’s merits analysis.
First, the majority opinion does not heed the administrative law judge’s factual conclusions, which were based on the judge’s in-person assessment of Lorenzo’s testimony at trial. Those factual conclusions demonstrate that Lorenzo lacked the necessary mens rea of willfulness.
To show that Lorenzo willfully engaged in a scheme to defraud, the SEC had to prove that Lorenzo acted with an intent to deceive, manipulate, or defraud. But recall that, as findings of fact, the administrative law judge concluded (after hearing Lorenzo testify) that Lorenzo did not draft the emails, did not think about the contents of the emails, and sent the emails only at the behest of his boss.
In light of the administrative law judge’s factual findings, how can Lorenzo be deemed to have willfully engaged in a scheme to defraud? The majority opinion says that the facts found by the administrative law judge are not the right facts. Instead, in reaching its conclusion, the majority opinion relies on the SEC’s alternative facts, which the SEC devised on its own without hearing from any witnesses. See Maj. Op. 589-90, 592-93 (adopting the SEC’s view of the facts over the administrative law judge’s view).1
It is true that, under certain circumstances, an agency such as the SEC may re-examine and overturn an administrative law judge’s factual findings. See Universal Camera Corp. v. NLRB, 340 U.S. 474, 492, 71 S.Ct. 456, 95 L.Ed. 456 (1951). But an *600agency does not have carte blanche to rewrite an administrative law judge’s factual determinations. Rather, an agency must act reasonably when it disregards an administrative law judge’s factual findings, a point the SEC’s attorney expressly acknowledged at oral argument. See Tr. of Oral Arg. at 28. It is black-letter law, therefore, that “a contrary initial decision” by an administrative law judge “may undermine the support for the agency’s ultimate determination.” Ronald M. Levin & Jeffrey S. Lubbers, Administrative Law and Process 101 (6th ed. 2017). And here is the key principle that speaks directly to this case: “When the case turns on eyewitness testimony ... the initial decision should be given considerable weight: the ALJ was able to observe the demeanor of the witnesses and assess their credibility and veracity first hand.” Id.
In my view, the majority opinion misapplies those black-letter principles. Contrary to the majority opinion’s acceptance of the SEC’s reconstruction of the facts in this case, I would conclude that the SEC’s rewriting of the administrative law judge’s findings of fact was utterly unreasonable and should not be sustained or countenanced by this Court. Given that Lorenzo was the only relevant witness at trial (dwell again on that point for a few moments) and given that his credibility was central to the case, the SEC had no reasonable basis to run roughshod over the administrative law judge’s findings of fact and credibility assessments. In short, the SEC’s rewriting of the findings of fact deserves judicial repudiation, not judicial deference or respect.
Instead of deferring to the SEC’s creation of an alternative factual record, as the majority opinion does, we should examine the administrative law judge’s underlying findings of fact and ask whether those findings suffice to support the conclusion that Lorenzo willfully engaged in a scheme to defraud. The answer to that question, as explained above, is a clear no.2
Second, put that aside. Even if I am wrong about the first point, the majority opinion still suffers from a separate flaw, in my view.
The majority opinion creates a circuit split by holding that mere misstatements, standing alone, may constitute the basis for so-called scheme liability under the securities laws—that is, willful participation in a scheme to defraud—even if the defendant did not make the misstatements.3 No other court of appeals has adopted the approach that the majority opinion adopts here. Other courts have instead concluded that scheme liability must be based on conduct that goes beyond a defendant’s role in preparing mere misstatements or omissions made by others. See, e.g., Public Pension Fund Group v. KV Pharmaceutical Co., 679 F.3d 972, 987 (8th Cir. 2012); WPP Luxembourg Gamma Three Sarl v. Spot Runner, Inc., 655 F.3d 1039, 1057 (9th Cir. 2011); Len*601tell v. Merrill Lynch & Co., 396 F.3d 161, 177 (2d Cir. 2005); see also SEC v. Kelly, 817 F.Supp.2d 340, 343-44 (S.D.N.Y. 2011). Otherwise, the SEC would be able to evade the important statutory distinction between primary liability and secondary (aiding and abetting) liability. After all, if those who aid and abet a misstatement are themselves primary violators for engaging in a scheme to defraud, what would be the point of the distinction between primary and secondary liability?
The distinction between primary and secondary liability matters, particularly for private securities lawsuits. For decades, however, the SEC has tried to erase that distinction so as to expand the scope of primary liability under the securities laws. For decades, the Supreme Court has pushed back hard against the SEC’s attempts to unilaterally rewrite the law. See Janus, 564 U.S. 135, 131 S.Ct. 2296; Stoneridge Investment Partners, LLC v. Scientific-Atlanta, Inc. 552 U.S. 148, 128 S.Ct. 761, 169 L.Ed.2d 627 (2008); Central Bank of Denver, N.A v. First Interstate Bank of Denver, N.A., 511 U.S. 164, 114 S.Ct. 1439, 128 L.Ed.2d 119 (1994). Still undeterred in the wake of that body of Supreme Court precedent, the SEC has continued to push the envelope and has tried to circumvent those Supreme Court decisions. See, e.g., In the Matter of John P. Flannery & James D. Hopkins, Release No. 3981 (Dec. 15, 2014). This case is merely the latest example.4
I agree with the other courts that have rejected the SEC’s persistent efforts to end-run the Supreme Court. I therefore respectfully disagree with the majority opinion that Lorenzo’s role in forwarding the alleged misstatements made by Lorenzo’s boss can be the basis for scheme liability against Lorenzo.
Taking a step back on the scheme liability point, moreover, think about the oddity of the majority opinion’s combined legal rulings today. The majority opinion emphatically holds that Lorenzo did not “make” the statements in the emails. In reaching that conclusion, the majority opinion accurately says that “Lorenzo transmitted statements devised by” Lorenzo’s boss at his boss’s “direction.” Maj. Op. 587. The majority opinion also correctly notes that Lorenzo’s boss “asked Lorenzo to send the emails, supplied the central content, and approved the messages for distribution.” Maj. Op. 588. At the same time, however, the majority opinion emphatically- holds that Lorenzo nonetheless willfully engaged in a scheme to defraud solely because of the statements made by his boss. That combined holding makes little sense (at least to me) under the facts of this particular case. Nor does it make much sense under the law, which is presumably why the other courts of appeals have rejected that kind of legal jujitsu. In these circumstances, perhaps the alleged offender (here, Lorenzo) could have been charged with aiding and abetting, if the *602relevant mens rea requirements for aiding and abetting liability were met. But Lorenzo may not be held liable as a primary violator, in my view.
* * *
Administrative adjudication of individual disputes is usually accompanied by deferential review from the Article III Judiciary. That agency-centric process is in some tension with Article III of the Constitution, the Due Process Clause of the Fifth Amendment, and the Seventh Amendment right to a jury trial in civil cases. See generally Philip Hamburger, Is Administrative Law Unlawful? 227-57 (2014). That tension is exacerbated when, as here, the agency’s political appointees—without hearing from any witnesses—disregard an administrative law judge’s factual findings. That said, the Supreme Court has allowed administrative adjudication ever since Crowell v. Benson, 285 U.S. 22, 52 S.Ct. 285, 76 L.Ed. 598 (1932). But the premise of Crowell v. Benson is that, putting aside any formal constitutional problems with the notion of administrative adjudication, the administrative adjudication process will at least operate with efficiency and with fairness to-the parties involved. This case, among others, casts substantial doubt on that premise.
Securities brokers such as Frank Lorenzo obviously do not tug at\the judicial heartstrings. And maybe Lorenzo really is guilty of negligence (or worse). But before the SEC reaches such a conclusion, Lorenzo is entitled to a fair process just like everyone else. Cf. United States v. Burwell, 690 F.3d 500, 527 (D.C. Cir. 2012) (en banc) (Kavanaugh, J., dissenting). He has not received a fair process in this case.
I hope that the SEC on remand pays attention, comes to its senses,- and (at a minimum) dramatically scales back the sanctions in this' case. Indeed, notwithstanding the majority opinion, I hope that the SEC, on its own motion, goes further than that: The SEC should vacate the order against Lorenzo in its entirety and either end this case altogether or (if appropriate and permissible) fairly start the process anew before the administrative law judge.
I firmly disagree with the majority opinion’s decision to sustain the SEC’s findings of liability under Section 10(b), Rule 10b-5(a) and (c), and Section 17(a). I respectfully dissent.
. The majority opinion also says that Lorenzo, in his briefing here, does not describe his own state of mind in the way that the administrative law judge did. In other words, the majority opinion says that Lorenzo accepts the SEC’s reconstruction of the facts. I disagree. To be sure, Lorenzo advances the alternative argument that he should prevail even if the SEC’s reconstruction of the facts is correct. But Lorenzo certainly does not agree with or accept the SEC’s reconstruction.
Moreover, in making this point, the majority opinion draws a dichotomy between Lorenzo's good-faith belief (as noted in his briefs) in the accuracy of the emails and Lorenzo’s statement that he did not think about the contents of the email. That is a false dichotomy. When forwarding an email on behalf of your boss, you could have a good-faith belief in its accuracy because you trust your boss, or at least have no reason to delve deeply into the particulars of the email’s contents, not because you have necessarily read or independently verified the contents of the email. The majority opinion notes that Lorenzo, "as of November 2009,” did not trust his boss. Maj. Op. 593. But that date is of course after the events at issue in this case.
. At oral argument, counsel for the SEC actually stated that it would have been "more difficult” for the SEC to find Lorenzo liable if Lorenzo's email had said that it was being sent “on behalf of” his boss instead of "at the request of” his boss. Counsel for the SEC asserted that those two phrases were "meaningfully different.” Tr. of Oral Arg. at 30. With respect, I find that argument absurd and an illustration of how the Commission jumped the rails in this case. It is startling that the SEC thinks such a wafer-thin semantic distinction can make the difference between (i) a lifetime suspension from your chosen profession and (ii) no penalty at all.
. On page 595, the majority opinion ultimately appears to acknowledge the circuit split: "Insofar as those courts of appeals would find Lorenzo’s actions to lie beyond the reach of those provisions, we read the provisions differently.”
. In this case, the SEC relied on its prior decision in Flannery. But as one respected commentator persuasively explained, the SEC's Flannery decision is wrong. "The substantive concern is that the Commission defined primary liability under portions of the major anti-fraud provisions, in expansive ways that disregarded the reasoning and rationale of the Supreme Court and some courts of appeals. The Supreme Court has sought to clarify the distinction between primary and secondary liability under Rule 1 Ob-5, yet the Commission’s Flannery decision all but eradicated the distinction and committed the same • error with Section 17(a). It sought to regain the ground on primary liability that was lost in Stoneridge and Janus and then went further with novel constructions of primary liability based on lawful, non-deceptive actions or exorbitant doctrines of but-for causality.” Andrew N. Vollmer, SEC Revanchism and the Expansion of Primary Liability Under Section 17(a) and Rule 1 Ob-5, 10 VA. L. & Bus. Rev. 273, 340 (2016).