Skip to main content


Reset A A Font size: Print

United States Court of Appeals, Eleventh Circuit.

UNITED STATES of America, Plaintiff-Appellant, v. Latecia WATKINS, Defendant-Appellee.

No. 18-14336

Decided: August 20, 2021

Before WILLIAM PRYOR, Chief Judge, WILSON, MARTIN, JORDAN, ROSENBAUM, JILL PRYOR, NEWSOM, BRANCH, GRANT, LUCK, LAGOA, BRASHER, ED CARNES, and MARCUS,* Circuit Judges. Brandy Brentari Galler, U.S. Attorney's Office, West Palm Beach, FL, Alexandra Chase, U.S. Attorney Service - Southern District of Florida, U.S. Attorney Service - SFL, Miami, FL, Emily M. Smachetti, U.S. Attorney's Office, Miami, FL, William T. Zloch, U.S. Attorney's Office, Fort Lauderdale, FL, for Plaintiff - Appellant Michael Caruso, M. Caroline McCrae, Federal Public Defender, Federal Public Defender's Office, West Palm Beach, FL, for Defendant - Appellee

Rehearing en banc is granted. The rules are suspended, see Fed. R. App. P. 2; 11th Cir. R. 2-1, and the case is submitted on the Appellee's petition for rehearing and the Appellant's response. This is the decision of the en banc Court vacating the panel opinion and remanding the case to the panel for further proceedings.


The exclusionary rule bars admission of evidence resulting from a Fourth Amendment violation, unless an exception applies. See Nix v. Williams, 467 U.S. 431, 442–43, 104 S.Ct. 2501, 81 L.Ed.2d 377 (1984). One of the exceptions is when the unconstitutionally obtained evidence would ultimately have been discovered through lawful means had there been no constitutional violation. Id. at 443–44, 104 S.Ct. 2501. The government has the burden of proving that exception applies. Id. at 444, 104 S.Ct. 2501. We granted rehearing en banc in this case to decide what standard of proof the government must satisfy to show that the evidence would ultimately have been discovered through lawful means without the constitutional violation. Must it show there was a reasonable probability of ultimate discovery, or show by a preponderance of the evidence that the evidence ultimately would have been discovered?

Forty-one years ago our predecessor court held that the proper standard of proof for determining if the evidence would ultimately have been discovered through lawful means is reasonable probability. United States v. Brookins, 614 F.2d 1037, 1042 n.2, 1044–48 (5th Cir. 1980). Ever since then the Brookins reasonable probability standard of proof has been reiterated in decisions of this Court involving the ultimate discovery exception. See United States v. Wilson, 671 F.2d 1291, 1293–94 (11th Cir. 1982); United States v. Roper, 681 F.2d 1354, 1358 (11th Cir. 1982); Jefferson v. Fountain, 382 F.3d 1286, 1296 (11th Cir. 2004); United States v. Johnson, 777 F.3d 1270, 1274 (11th Cir. 2015). Except for the occasional lapse. See McKathan v. United States, 969 F.3d 1213, 1232 (11th Cir. 2020) (“The ‘inevitable discovery’ doctrine applies when the government can show by a preponderance of the evidence that it would have discovered the evidence by some other lawful means.”).

Four years after our Brookins decision, the Supreme Court held that for the ultimate discovery exception, proof by a preponderance of the evidence was at least a permissible proof standard, if not the required one. Nix, 467 U.S. at 444, 104 S.Ct. 2501 (“If the prosecution can establish by a preponderance of the evidence that the information ultimately or inevitably would have been discovered by lawful means ․ the evidence should be received. Anything less would reject logic, experience, and common sense.”) (footnote omitted).

Because Nix was decided after our Brookins decision, one might have expected that it would prompt this Court to replace our reasonable probability standard with the plain old preponderance of the evidence, more-likely-than-not standard that the Supreme Court approved in Nix. Instead, we re-pledged allegiance to the Brookins reasonable probability standard and went on applying that standard, holding it was not inconsistent with the Nix decision. See Jefferson, 382 F.3d at 1296 (citing the Brookins decision for the reasonable probability standard and stating that “[s]ince the [Supreme Court's] Nix decision, we have continued to follow the Brookins decision, which is entirely consistent with it”).

The panel in this case was bound by the prior panel precedent rule to follow the holdings of our post-Nix decisions that reiterated the Brookins reasonable probability standard. See United States v. Tellis, 748 F.3d 1305, 1309–10 (11th Cir. 2014); Smith v. GTE Corp., 236 F.3d 1292, 1302–03 (11th Cir. 2001). But, sitting en banc, we are not bound to do so but may decide anew whether Supreme Court precedent simply permits the preponderance of the evidence standard of predictive proof in ultimate discovery exception cases, or instead requires its use. We hold that Supreme Court precedent requires the use of the preponderance standard for these purposes. And even if that were not so, we would hold that the preponderance standard must be used because of the advantages it has over the reasonable probability standard when it comes to finding whether evidence ultimately would have been discovered through lawful means if the constitutional violation had not taken place.


In Nix the Supreme Court explicitly rejected the defendant's argument that a standard of proof higher than a preponderance should be required for application of the ultimate discovery exception. 467 U.S. at 444 n.5, 104 S.Ct. 2501. But, as the panel opinion in this case pointed out, the Court did not explicitly hold in Nix that a lesser standard of proof was insufficient. See United States v. Watkins, 981 F.3d 1224, 1234 n.3 (11th Cir. 2020). Apparently for that reason, most post-Nix decisions of this Court continued to follow the Brookins decision and its reasonable probability standard of proof. But none of those decisions took note of what the Supreme Court said about its 1984 Nix decision three years later in Bourjaily v. United States, 483 U.S. 171, 107 S.Ct. 2775, 97 L.Ed.2d 144 (1987).

Bourjaily involved the Federal Rule of Evidence 801(d)(2)(E) co-conspirator exception to the hearsay rule. Id. at 173, 107 S.Ct. 2775. More specifically, the issue was what standard of proof the offering party must meet on the preliminary factual issues of whether “there was a conspiracy involving the declarant and the nonoffering party, and [whether] the statement was made during the course and in furtherance of the conspiracy.” Id. at 175, 107 S.Ct. 2775 (quotation marks omitted). The Court stated that its decision of the issue was “guided by our prior decisions regarding admissibility determinations that hinge on preliminary factual questions,” which had “traditionally required that these matters be established by a preponderance of proof.” Id.

Most importantly for our purposes, the Court explained in Bourjaily that “[w]e think that our previous decisions in this area resolve this matter,” and it listed four of them. Id. at 176, 107 S.Ct. 2775. The second one the Court listed, and the Court's parenthetical summary of that decision, was: “Nix v. Williams, 467 U.S. 431, 444 n.5 [104 S.Ct. 2501, 81 L.Ed.2d 377] (1984) (inevitable discovery of illegally seized evidence must be shown to have been more likely than not).” Bourjaily, 483 U.S. at 176, 107 S.Ct. 2775 (emphasis added).

So just three years after it had issued the Nix decision, the Supreme Court told us what that decision requires. The Court did not tell us in Bourjaily that Nix allows a reasonable probability standard or anything less than a preponderance of the evidence more-likely-than-not standard. Instead, it told us that Nix requires that the offering party must –– not may, but must –– prove that ultimate discovery through lawful means was more likely than not, which is the preponderance of the evidence standard, see infra at 1184-85. “Must” is a term of requirement. Cf. Burban v. City of Neptune Beach, 920 F.3d 1274, 1279 (11th Cir. 2019) (pointing out that in statutory construction “must,” like “shall,” is a mandatory term that connotes a requirement).

Some may argue that the Court's statement in Bourjaily about Nix is dicta, but we need not decide whether it is. As we have stated before, “there is dicta and then there is dicta, and then there is Supreme Court dicta.” Schwab v. Crosby, 451 F.3d 1308, 1325 (11th Cir. 2006). Lacking the temerity to tell the Supreme Court that it was wrong in Bourjaily about what its holding in Nix was, we will realign our circuit law about Nix’s holding to square with what the Supreme Court in Bourjaily said Nix held.


Even if the Bourjaily opinion had not provided us with the Supreme Court's understanding of its Nix decision, we would still hold that preponderance of the evidence is a better standard of proof than reasonable probability is for the ultimate discovery exception to the exclusionary rule. Here's why.

The primary problem with using the reasonable probability standard of proof to predict whether evidence discovered through a constitutional violation ultimately would have been discovered anyway is that no one knows exactly what reasonable probability means in this context. The words are plain enough separately, but their combined meaning is anything but plain. The term “reasonable probability” implies there must be an unreasonable probability, just as darkness must exist for light to have meaning. Otherwise, why put the limiting adjective “reasonable” in front of the noun “probability” –– what work does “reasonable” do? But how can a probability be unreasonable? How does a reasonable probability differ from an unreasonable one? In the 41 years that the term “reasonable probability” has been the guiding standard in this circuit for an important exception to the exclusionary rule, it has never been defined in this context. Probably because no one knows exactly what it means here.

To be sure, the Supreme Court has crafted “reasonable probability” as a relatively new term of art for use in other areas of the law. See generally Borden v. United States, ––– U.S. ––––, 141 S. Ct. 1817, 1828, 210 L.Ed.2d 63 (2021) (“[T]erms of art depart from ordinary meaning.”) (quotation marks omitted). The most famous example is use of “reasonable probability” as the standard of prejudice in ineffective assistance of counsel cases. In Strickland the Court undertook the task of deciding the appropriate measure of prejudice where counsel's performance fell outside the broad scope of reasonable professional assistance. Strickland v. Washington, 466 U.S. 668, 691–96, 104 S.Ct. 2052, 80 L.Ed.2d 674 (1984). The guiding star for the Court's analysis was that “[t]he purpose of the Sixth Amendment guarantee of counsel is to ensure that a defendant has the assistance necessary to justify reliance on the outcome of the proceeding.” Id. at 691–92, 104 S.Ct. 2052. Given that purpose, the Court reasoned that “not every error [of counsel] ․ undermines the reliability of the result of the proceeding.” Id. at 693, 104 S.Ct. 2052. It considered making the prejudice standard whether counsel's deficient conduct “more likely than not altered the outcome in the case,” which would have required a defendant to show by a preponderance of the evidence that counsel's errors had determined the outcome. Id. at 693–94, 104 S.Ct. 2052. But the Court rejected that “outcome-determinative standard” as “not quite appropriate.” Id.

Instead, the Strickland Court decided on a lesser showing, holding that: “The defendant must show that there is a reasonable probability that, but for counsel's unprofessional errors, the result of the proceeding would have been different.” Id. at 694, 104 S.Ct. 2052. And in the very next sentence, the Court helpfully provided this definition for its term of art phrase: “A reasonable probability [of a different result] is a probability sufficient to undermine confidence in the outcome.” Id. That definition dovetails nicely with the purpose of the Sixth Amendment guarantee of effective assistance of counsel, which is to ensure that the outcome of the criminal proceeding will be reliable. But Strickland doesn't tell us what the reasonable probability standard means when used in connection with the ultimate discovery exception to the exclusionary rule.

Unlike the Sixth Amendment, the purpose of the Fourth Amendment and the exclusionary rule is not to increase the reliability of criminal proceedings or bolster our confidence in their outcome. Nor is that their effect. Just the opposite. The primary effect of using the exclusionary rule to enforce the Fourth Amendment is to exclude from the trier of fact some relevant and probative evidence, which could decrease the reliability of the outcome of a criminal proceeding. See Lange v. California, ––– U.S. ––––, 141 S. Ct. 2011, 2027, 210 L.Ed.2d 486 (2021) (Thomas, J., concurring in part and in the judgment, joined by Kavanaugh, J.) (One “especially salient” cost of the exclusionary rule is that “excluding evidence under the Fourth Amendment always obstructs the truth-finding functions of judge and jury.”) (quotation marks omitted); id. (The exclusionary rule's “interference with the purpose of the judicial system also creates a downstream risk that some guilty defendants may go free or receive reduced sentences.”) (quotation marks omitted).

We are not criticizing the Fourth Amendment and the important constitutional values it embodies. Not at all. Nor are we criticizing the exclusionary rule and the important purposes it serves. Not at all. Instead, we are simply explaining why the Strickland definition of “reasonable probability” as a term of art, which serves its function well in ineffective assistance prejudice inquiries, won't work as a standard of proof for the ultimate discovery exception to the exclusionary rule. We “do not force term-of-art definitions into contexts where they plainly do not fit and produce nonsense.” Johnson v. United States, 559 U.S. 133, 139–40, 130 S.Ct. 1265, 176 L.Ed.2d 1 (2010).

The Strickland reasonable probability definition plainly does not fit into the analysis of whether an exception to the exclusionary rule applies, and it would be nonsensical to force it in. To the question of whether applying an exception to the exclusionary rule would increase our confidence in the accuracy of the outcome of a trial, the answer invariably will be yes. The application of an exception to exclusion results in inclusion. Instead of being kept from the trier of fact, more evidence that is relevant and probative will be admitted. (If the evidence were not relevant and probative, it would not be admitted anyway.) The more relevant and probative evidence put before the trier of fact, the more reliable the verdict is likely to be. And the more reliable the verdict is likely to be, the greater our confidence in the outcome will be. That is why using the Strickland “reasonable probability” definition or standard, which focuses entirely on confidence in the reliability of the trial's outcome, would always lead to application of the ultimate discovery exception to the exclusionary rule. A standard or test that always produces the same result is not a standard or test. A box that will always be checked is not useful in an analysis.

The Supreme Court has also used the same reasonable probability term of art it used in Strickland to describe the measure of prejudice in Brady violation cases. See Kyles v. Whitley, 514 U.S. 419, 434, 115 S.Ct. 1555, 131 L.Ed.2d 490 (1995) (The materiality of suppressed evidence does not require a preponderance showing of a different result but a “reasonable probability” of one, which as in Strickland is measured by whether the suppression “undermines confidence in the outcome of the trial.”); accord Banks v. Dretke, 540 U.S. 668, 698–99, 124 S.Ct. 1256, 157 L.Ed.2d 1166 (2004); United States v. Bagley, 473 U.S. 667, 682, 105 S.Ct. 3375, 87 L.Ed.2d 481 (1985) (“The evidence is material only if there is a reasonable probability that, had the evidence been disclosed to the defense, the result of the proceeding would have been different. A ‘reasonable probability’ is a probability sufficient to undermine confidence in the outcome.”). That term of art definition wouldn't work any better in ultimate discovery exception cases if we borrowed it from the Brady materiality decisions than it would if we borrowed the same definition from the Strickland decision.

So what standard of proof should be used to determine if the ultimate discovery exception applies? We think the wisest course is to follow the Supreme Court's lead in Nix and use the preponderance of the evidence standard. Even putting aside –– as we are here –– what Bourjaily says about it, Nix at a minimum holds that preponderance of the evidence is a permissible standard of proof here. See Nix, 467 U.S. at 444 & n.5, 104 S.Ct. 2501. It provides a green light for use of that standard of proof of ultimate discovery. No Supreme Court decision green lights use of the reasonable probability standard for ultimate discovery purposes.

The preponderance standard of proof is one of widespread and longstanding use. See Concrete Pipe & Prods. of Cal., Inc. v. Constr. Laborers Pension Tr. for S. Cal., 508 U.S. 602, 622, 113 S.Ct. 2264, 124 L.Ed.2d 539 (1993) (noting that the burden of showing something by a preponderance of the evidence is “the most common standard in the civil law”); United States v. Trainor, 376 F.3d 1325, 1331 (11th Cir. 2004) (“[T]he term ‘preponderance of the evidence’ is a common law term of longstanding use.”). Its meaning is simple, straightforward, and clear. “A preponderance of the evidence is evidence which is more convincing than the evidence offered in opposition to it.” Metro. Stevedore Co. v. Rambo, 521 U.S. 121, 137 n.9, 117 S.Ct. 1953, 138 L.Ed.2d 327 (1997) (quoting Greenwich Collieries v. Dir., OWCP, 990 F.2d 730, 736 (3d Cir. 1993)) (cleaned up). It “simply requires the trier of fact to believe that the existence of a fact is more probable than its nonexistence.” Concrete Pipe, 508 U.S. at 622, 113 S.Ct. 2264 (quotation marks omitted) (quoting In re Winship, 397 U.S. 358, 371–72, 90 S.Ct. 1068, 25 L.Ed.2d 368 (1970) (Harlan, J., concurring)). Or phrased in a slightly different fashion, it is proof that persuades the trier of fact that a proposition “is more likely true than not true.” United States v. Deleveaux, 205 F.3d 1292, 1296 n.3 (11th Cir. 2000) (quoting a jury instruction that was upheld); see also 11th Cir. Pattern Civ. Jury Instr. 1.1 (2020) (stating that the standard of proof by a preponderance of the evidence means the party with the burden “must prove that, in light of all the evidence, what [that party] claims is more likely true than not”).

The preponderance standard is well-defined; the reasonable probability standard is undefined in our case law for use in this context. The preponderance standard is unambiguous and clear; the reasonable probability standard is ambiguous and vague in this context. The preponderance standard is straightforward and simple to apply; the reasonable probability standard is not in this context. Use of the preponderance standard in this context has the Supreme Court's good judging seal of approval; use of the reasonable probability standard does not.

Even if Nix does not mandate use of the preponderance standard, we hold that the standard of predictive proof the government must satisfy in order to establish the proper application of the ultimate discovery exception is preponderance of the evidence, not reasonable probability. All of our decisions holding to the contrary are overruled.

The panel opinion is VACATED and the case is REMANDED to it for further proceedings consistent with this opinion.

MARTIN and ED CARNES, Circuit Judges:

Copied to clipboard