PRECEDENTIAL

      UNITED STATES COURT OF APPEALS
           FOR THE THIRD CIRCUIT

                     __________

                     No. 16-2795
                     __________

          UNITED STATES OF AMERICA

                          v.

               ASKIA WASHINGTON,
                     a/k/a SKI

                       Askia Washington,
                              Appellant
                     __________

    On Appeal from the United States District Court
       for the Eastern District of Pennsylvania
       (D.C. Criminal No. 2:13-cr-00171-002)
      District Judge: Honorable Joel H. Slomsky
                      __________

             Argued on February 7, 2017

Before: MCKEE, COWEN, and FUENTES, Circuit Judges

           (Opinion Filed: August 28, 2017)
Mark S. Greenberg, Esq. [Argued]
920 Lenmar Drive
Blue Bell, PA 19422
             Counsel for Appellant

Eric B. Henson, Esq. [Argued]
Bernadette McKeon, Esq.
Zane David Memeger, Esq.
Office of United States Attorney
615 Chestnut Street
Suite 1250
Philadelphia, PA 19106
              Counsel for the Appellee

                        __________

                OPINION OF THE COURT
                      __________

FUENTES, Circuit Judge.

      Defendant-appellant Askia Washington was ensnared
by a “stash house reverse sting” operation—one which hit
many of the by-now-familiar beats.1 Acting on what
appeared to be insider information from a drug courier,
Washington and his three co-conspirators planned to rob a
Philadelphia property where they thought 10 kilograms of
cocaine were being stored for distribution. But as they
discovered on the day of the robbery, the “stash house” was a

1
  See United States v. Pedrin, 

, 794 (9th Cir.
2015) (explaining the basic framework of stash house reverse
sting operations), cert. denied, 

(2016).




                              2
trap set by law enforcement. Their “courier” was an
undercover federal agent with the Bureau of Alcohol,
Tobacco, Firearms and Explosives (“ATF”), which had
developed the scenario from the ground up. The cocaine did
not exist.

      Under federal law on conspiracy and attempt, the
government could, and did, prosecute the crew as if fantasy
had been reality. Washington, the sole member to take his
chances at trial, was convicted by a jury of two Hobbs Act
robbery charges and two drug charges (18 U.S.C. § 1951(a)
and 21 U.S.C. § 846), although he was acquitted on a gun
charge.

        Developed by the ATF in the 1980s to combat a rise in
professional robbery crews targeting stash houses, reverse
sting operations have grown increasingly controversial over
the years, even as they have grown safer and more refined.
For one, they empower law enforcement to craft offenses out
of whole cloth, often corresponding to statutory offense
thresholds. Here, the entirely fictitious 10 kilograms of
cocaine triggered a very real 20-year mandatory minimum for
Washington, contributing to a total sentence of 264 months in
prison—far more than even the ringleader of the conspiracy
received. For another, and as Washington claimed on
multiple occasions before the District Court—and now again
on appeal—people of color are allegedly swept up in the
stings in disproportionate numbers.

       These elements of controversy are bound up in the
three claims Washington now raises on appeal. Two are
constitutional claims: Washington challenges his conviction
and sentence by arguing that the use of the statutory
mandatory minimum term violated his rights to due process,




                             3
and he also alleges that the attorney who represented him at
trial rendered constitutionally ineffective assistance. While
stash-house reverse stings can raise constitutional concerns,
the use of a mandatory minimum sentence on these particular
facts did not deprive Washington of his right to due process.
And while this is the rare case where a claim of ineffective
assistance of counsel was properly raised on direct appeal
instead of through a collateral attack, Washington has not
shown prejudice sufficient to call into doubt the integrity of
his trial. We thus conclude that both constitutional claims are
without merit.

        The remaining claim challenges the District Court’s
decision to deny Washington pretrial discovery on ATF’s
operations and enforcement statistics. Washington contends
that, in denying his motion, the District Court erroneously
relied on the hard-to-meet test for “selective prosecution”
discovery developed by the Supreme Court in United States v.
Armstrong2 and United States v. Bass3 (which we will refer to
as “Armstrong/Bass”). He encourages us to follow instead
the en banc Seventh Circuit’s recent opinion in United States
v. Davis,4 which distinguished between claims of selective
prosecution and selective law enforcement and appeared to
endorse a relaxed discovery standard for the latter.

       Like the Seventh Circuit, we conclude that the
proposed distinction between enforcement and prosecution is
well taken, and that the law supports greater flexibility when

2
    

(1996).
3
    

(2002) (per curiam).
4
    

(7th Cir. 2015) (en banc).




                                4
the discretionary decisions of law enforcement, rather than
those of prosecutors, are targeted by a defendant’s request for
discovery. We therefore hold that a district court may
exercise its discretion to grant limited discovery, or otherwise
to conduct in camera analysis of government data before
deciding whether limited discovery is warranted. A district
court may do so even if a defendant seeking discovery on a
selective enforcement claim has not otherwise met his or her
full burden under Armstrong/Bass. Because the District
Court in this case thought that its discretion was cabined by
Armstrong/Bass, and because we cannot otherwise say that
the same result would have occurred under the standard we
announce today, we will vacate the orders denying discovery
and remand for limited post-judgment proceedings. The
judgment of conviction and sentence are otherwise unaffected
by this remand.




                               5
    I.   BACKGROUND

           A. The Plan5

        Codefendant and ringleader Dwight Berry came to the
attention of the ATF in late 2012, when he made it known
that he was interested in conducting robberies of drug users
and dealers. In the course of asking around, Berry spoke to
an acquaintance who, unbeknownst to him, was an ATF
confidential informant (“CI”). The CI alerted the ATF, which
determined that Berry’s criminal history fit its required
profile for a sting operation and opened an investigation in
February 2013, under the supervision of ATF Special Agent
John Bowman. From here on out, many of the meetings and
phone calls about the developing robbery plan would be
surreptitiously recorded for playback at trial.

       Meanwhile, the CI kept Berry on the line with word of
a connection: a drug-courier friend who frequented a South
Philadelphia stash house on his trips to and from New York.
When Berry and the CI met again, they were joined by the

5
  Our description of the trial and underlying scheme is drawn
primarily from the District Court’s opinion denying
Washington’s motion for a new trial, United States v.
Washington [hereinafter “Washington New Trial”], 184 F.
Supp. 3d 149 (E.D. Pa. 2016). Washington accepts the
factual accuracy of the District Court’s opinion, see
Washington Br. at 7 n.4, and both parties have structured their
briefs around it. As Washington is not challenging the
sufficiency of the evidence, we strive to recite the facts in a
balanced manner. See United States v. Cox, 

,
118 n.1 (1st Cir. 2017).




                              6
supposed drug courier—in reality, undercover ATF Special
Agent Patrick Edwards, a veteran of over a dozen robbery
scenarios. In his role as the courier, Edwards reported seeing
over 10 kilograms of cocaine (in the context of cocaine
“bricks”) inside a cooler during a trip to the stash house.
Berry indicated that he knew of a crew who might be
interested in participating in the robbery and that he was
willing to engage in violence if necessary.

       Washington first entered the picture about a week and
a half after this encounter as one of two members of Berry’s
proposed robbery crew (the other man, never identified,
apparently dropped out of the plan shortly afterwards). At
another meeting in early March 2013 with Berry, Edwards,
and the CI, Washington probed Edwards about the logistics of
the robbery: what level of resistance they could expect,
whether the house would be watched from the outside, and so
on. Prompted by Edwards, the conspirators also discussed
how to move and sell the stolen cocaine.6

6
  As captured by the recording, and as later explained at trial,
the conspirators made frequent reference to “jawns” or
“jauns,” a distinctive Philadelphia regionalism that serves as a
wildcard stand-in for other nouns. See Dan Nosowitz, The
Enduring Mystery Of ‘Jawn’, Philadelphia’s All-Purpose
Noun,                       Atlas                      Obscura,
http://www.atlasobscura.com/articles/the-enduring-mystery-
of-jawn-philadelphias-allpurpose-noun (last visited Aug. 21,
2017; archived at https://perma.cc/6XM6-JQEW); see also
United States v. Gibbs, 

, 200 n.4 (3d Cir. 1999)
(“Apparently, ‘jawn’ is slang for any noun, and throughout
this case it was used variously to describe a car, cocaine, a
nightclub, and a beeper.”).




                               7
       In a subsequent phone discussion, Edwards pressed
Berry on the professionalism of his crew. Berry, in an
attempt to reassure, told Edwards that “[t]his is what [our
crew] do[es].”7 When Edwards singled out Washington for
concern over a perceived lack of robbery experience, Berry
said that Washington “rock[ed] out” and “put work in,” which
Edwards interpreted to mean that Washington was some sort
of shooter or enforcer.8

        On the day of the robbery, Washington and Berry met
at Berry’s mother’s house, where Berry picked up two guns
and hid them in an Eggo Waffles box. The group, which had
added two new members—codefendants Antonio Ellis and
Jermau Johnston—then gathered in the parking lot of the
Philadelphia Airport Hilton to review its plan. (Washington’s
girlfriend was also present, although she did not participate
and remained in her parked car.) Edwards went over the
salient details once more, emphasizing the 10 kilograms of
cocaine and explaining that no money would be found in the
house.

      In three cars—Berry, Ellis, and Johnston in a minivan;
Washington and his girlfriend (the latter driving) following
behind in a Chrysler 300; and Agent Edwards bringing up the
rear—the crew made its way to the chosen address on
Passyunk Avenue in southwest Philadelphia. As the caravan
moved in, agents swooped down. All but Berry surrendered;
Berry fled on foot but was apprehended shortly afterwards.
From the minivan, law enforcement recovered two guns,
ammo, gloves, and zip-ties. From Washington’s Chrysler

7
    Supplemental Appendix (S.A.) 55.
8
    S.A. 60–61.




                              8
300, they recovered a backpack, gloves, a mask, a lighter, and
lighter fluid.

          B. Procedural History

        What follows is an abbreviated summary of the
criminal proceedings, setting up the claims that Washington
now raises on appeal. We will return in greater detail to the
salient parts later, in the Analysis section of this opinion.

             1. Indictment; Codefendants Plead
                Guilty

       In April 2013, the four men were indicted in the
Eastern District of Pennsylvania. Counts 1 and 2 of the
indictment charged attempt/conspiracy to commit Hobbs Act
robbery (18 U.S.C. § 1951(a)), while counts 3 and 4 charged
attempt/conspiracy to possess with intent to distribute five
kilograms or more of cocaine (21 U.S.C. § 846 through 21
U.S.C. § 841(a)(1), (b)(1)(A)). Count 5 charged all of the
defendants with carrying a firearm during a crime of violence
(18 U.S.C. § 924(c)) and count 6 charged all but Johnston
with being felons in possession of a firearm (18 U.S.C.
§ 922(g)(1)).9

       Washington’s codefendants eventually pleaded guilty.
Johnson and Ellis received 27-month and 46-month
sentences, respectively. Although Berry, the ringleader, faced


9
  The government later obtained a superseding indictment
against Washington. A minor modification of the original, it
focused on Washington as a defendant and amplified a few of
the factual allegations.




                              9
a Guidelines range of 270–322 months, his binding plea
agreement reflected a 180-month sentence,10 and the
government did not seek to formally introduce his previous,
eligible convictions to secure an enhanced mandatory
minimum penalty. Berry ultimately received the agreed-upon
180-month custodial sentence.

       Unlike his codefendants, Washington pleaded not
guilty and prepared for trial. He was assigned a Criminal
Justice Act attorney, whom we will refer to as the “defense
counsel” or “trial counsel.”

               2. Motion for Discovery

       During the pretrial phase, Washington moved (both
pro se and through trial counsel) for discovery relating to
sting operations and related prosecutions, which he claimed to
be racially motivated. Trial counsel’s filing cited three prior
federal prosecutions in which all of the defendants were
African American. The moving papers also clarified that the
discovery was sought not for trial defense, but rather to
support a motion to dismiss the indictment on the basis of
“racial profiling or selective prosecution . . . by the
Philadelphia District Office of [ATF] . . . in complicity with”
the U.S. Attorney’s Office.11

      After oral argument, and as set forth in a thoughtful
opinion, the District Court denied Washington’s motion for
discovery. Finding the Armstrong/Bass standard to control,

10
  Plea agreements under Fed. R. Crim. P. 11(c)(1)(C) “bind[]
the court once the court accepts the plea agreement.”
11
     Discovery Motion at 1, ECF No. 126.




                              10
the District Court held that Washington failed to meet this
“rigorous standard to obtain discovery,”12 and later denied
Washington’s requests for reconsideration.

             3. Recordings Deemed Admissible;
                Government Seeks Enhanced
                Mandatory Penalties

       With discovery denied, Washington did not file an
actual motion to dismiss the indictment, and the parties
otherwise prepared for trial. In an important ruling, the
District Court decided that the government could use the
audio and video recordings and related transcripts at trial.
Meanwhile, the government filed a 21 U.S.C. § 851
information stating that Washington had a prior Pennsylvania
drug felony conviction from 2004—a prerequisite to
enhanced mandatory minimum penalties at sentencing.

             4. The District Court Revisits Discovery on
                the Eve of Trial

       In June 2015, prior to opening statements, the District
Court revisited the matter of discovery in the context of trial
defenses. Referring back to United States v. Alexander,13 a
Northern District of Illinois opinion cited in the earlier
decision denying discovery, the District Court ordered the
government to release redacted portions of an ATF policy

12
   United States v. Washington [hereinafter “Washington
Discovery”], No. 13-171-2, 

, at *7 (E.D.
Pa. June 30, 2014).
13
  No. 11 CR 148-1, 

(N.D. Ill. Dec. 10,
2013).




                              11
manual on stash house sting operations—patterned after the
disclosures ordered in Alexander. The District Court then
issued a protective order restricting defense counsel’s use of
the disclosed material.

              5. Washington’s Trial

       Over the five-day trial, defense counsel used the ATF
disclosures to advance his theory of the case: Washington did
not have the requisite intent to commit a dubious,
discriminatory “conspiracy” that ATF had designed from the
ground up.14 For instance, counsel pointed to Washington’s
use of a separate vehicle and the presence of his girlfriend on
the day of the robbery to suggest that he was cautious and not
fully committed. Counsel also utilized the disclosed ATF
materials to cross-examine supervising ATF Agent Bowman.

        But during that cross-examination, trial counsel
appeared to fumble. He was attempting to show that, as
Agent Bowman would later admit, the only person “targeted”
by the ATF prior to the arrest was Berry, and that the ATF
knew nothing about the other conspirators and could not have
ensured that they fit its target profile, which required (in part)
a violent criminal history. But in addition to asking whether
Washington had a prior robbery arrest (which he did not),
trial counsel also asked Agent Bowman whether Washington
had a drug arrest. This question effectively allowed the
prosecution to bring out Washington’s prior drug conviction
on redirect.



14
  We note that entrapment was not raised as a defense and is
not now at issue on appeal.




                               12
               6. The Jury Verdict

         The jury returned a guilty verdict on counts one
through four of the superseding indictment: the drug and
Hobbs Act robbery charges. It returned a not-guilty verdict
on firearm count five; firearm count six was dismissed on the
government’s motion.15 The jury specifically found that the
government proved beyond a reasonable doubt that the
(fictitious) cocaine at the center of the conspiracy was five
kilograms or more.

               7. Pre-Sentencing Investigation into Trial
                  Counsel’s Constitutional Effectiveness

       Shortly after the trial, Washington wrote a letter to the
District Court requesting a substitution of attorney. He
alleged, in part, that trial counsel had been under the
influence of alcohol throughout the trial.

        The District Court swiftly reacted, appointing a new
Criminal Justice Act attorney, Mark Greenberg—who has
represented Washington ever since—in what became, in
effect, a pre-sentencing investigation of trial counsel’s
performance. After the District Court held an evidentiary
hearing, Attorney Greenberg filed a formal motion for new
trial predicated on the alleged ineffective assistance of trial
counsel. This motion included an attack on trial counsel’s
questions during cross-examination of Agent Bowman that
opened the door to the introduction of Washington’s drug
conviction. The motion was ultimately denied, with the
District Court finding in part that the “mountain” of evidence


15
     See Order, ECF No. 219.




                               13
against Washington forestalled a showing of prejudice under
the two-part Strickland v. Washington16 test for ineffective
assistance of counsel.17

                8. Sentencing Proceedings

       The ineffectiveness question resolved for the time
being, the parties and District Court prepared for sentencing.
Because of his criminal history, Washington was classified as
a “Career Offender” under the Sentencing Guidelines. As a
result of Guidelines calculations we need not delve into, that
Career Offender status overrode the lower Guidelines level
derived from quantity of drugs, yielding a sentencing range of
360 months to life in prison.18

       In his sentencing memoranda, Washington challenged
the proposed sentencing range, emphasizing the troubling
nature of the sting operation and requesting that the District
Court take into account the sentences of his co-conspirators.
He also asked the District Court to disregard the mandatory
minimum sentence of 20 years; if “the reverse sting in this
case involved 0.9 kilograms of non-existent cocaine,” he
argued, “Mr. Washington would not be facing a mandatory
minimum sentence.”19         In response, the government
emphasized that the mandatory minimum penalty was just
that: mandatory.     Evincing some discomfort with the 20-
year mandatory minimum, the District Court nevertheless

16
     

(1984).
17
     See Washington New 

–62.
18
     See U.S.S.G. § 4B1.1(b).
19
     Supplemental Sentencing Memorandum at 2, ECF No. 275.




                                14
ruled that he was “bound to follow the law,”20 imposing a 24-
month sentence on the Hobbs Act robbery charges and a 240-
month consecutive sentence on the drug charges for a total
term of 264 months’ imprisonment. Washington timely
appealed.21

 II.     Analysis

      Washington’s constitutional challenges, which directly
attack the judgment of conviction and sentence, are
considered first. We will then turn to his Armstrong/Bass
discovery claim.

            A. Ineffective Assistance of Counsel

        Although he again invokes trial counsel’s alcohol use,
Washington otherwise limits his ineffectiveness claim on
appeal to the incident where trial counsel opened the door to
testimony about his drug conviction. He attacks the District
Court’s determination that the “overwhelming” evidence at
trial precluded a showing of prejudice, and emphasizes, in
particular, the jury’s acquittal on the firearm count and an
alleged conflation of the prejudicial impact of the admission
on the robbery counts with the far-greater impact on the drug
counts.




20
     S.A. 211.
21
  We have appellate jurisdiction through 18 U.S.C. § 3742(a)
and 28 U.S.C. § 1291.




                              15
              1. Ineffectiveness Claims on Direct
                 Appeal

       We open with the observation that ineffective
assistance of counsel claims are generally not considered on
direct appeal. Instead, they are more commonly brought in a
collateral proceeding, such as through a post-conviction 28
U.S.C. § 2255 motion to vacate.22

       Our “general aversion”23 to reaching ineffectiveness
claims on direct appeal derives in part from their inherently
collateral nature. The trial record, concerned as it is with the
defendant’s guilt or innocence, will not in most instances be
“developed precisely for the object of litigating or preserving
the [ineffective assistance] claim and thus [will] often [be]
incomplete or inadequate for this purpose.”24 Deferring the
question of ineffectiveness to collateral review also protects




22
  United States v. Hankerson, 

, 310 (3d Cir.
2007).
23
  Gov’t of the V.I. v. Vanterpool, 

, 164 (3d Cir.
2014).
24
   Massaro v. United States, 

, 504–05 (2003);
see also United States v. McLaughlin, 

, 556 (3d
Cir. 2004) (“[T]he lack of a fully developed record often
precludes a comprehensive inquiry into the elements of
strategy or tactics that may have entered into defense
counsel’s challenged decision.”).




                              16
criminal defendants from the consequences of resolving the
claims prematurely.25

       While cautioning that we will not “open[] the door to
ineffective assistance of counsel claims on direct appeal as a
matter of course,” we have nevertheless recognized an
exception to the rule when the trial record “is sufficient to
allow determination of ineffective assistance of counsel.”26
Determining sufficiency is case- and claim-dependent.

        We think that Washington’s is the uncommon case
where resolving an ineffectiveness claim on direct appeal is
both feasible and efficient. Strictly speaking, he is not raising
ineffectiveness for “the first time” on appeal. Rather,
ineffectiveness was invoked in and resolved by the District
Court, which held a post-trial, pre-sentencing hearing at
which Washington and the AUSA both testified (trial counsel
was invited to testify, but declined). The District Court—the
trial judge—then denied the claim against the backdrop of the
recently concluded trial.27 This development of the record


25
   See, e.g., United States v. Brown, 

, 90 n.5 (3d
Cir. 2017) (“To spare Brown from having res judicata attach
to the ineffective assistance claim, we decline to address it
here.” (internal quotation marks and citation omitted)).
26
  United States v. Polk, 

, 520 & n.2 (quoting
United States v. Headley, 

, 1083 (3d Cir.
1991)).
27
   See 

(“[T]he § 2255 motion often
will be ruled upon by the same district judge who presided at
trial. The judge, having observed the earlier trial, should have
an advantageous perspective for determining the effectiveness




                               17
amounted to, in effect, a mini collateral proceeding, akin to
what is ordinarily expected under § 2255. It provides us with
a sufficient foundation for direct appellate review.28 We
therefore exercise our discretion to reach the ineffectiveness
claim.29



of counsel’s conduct and whether any deficiencies were
prejudicial.”).
28
   See United States v. Jones, 

, 254 (3d Cir.
2003) (reaching ineffectiveness claim when District Court
“conducted a hearing with [the defendant] and his new
counsel where it specifically considered . . . allegations
concerning the representation he received from his prior
counsel”). The appendix as initially compiled lacked most of
the ineffectiveness-stage papers and transcripts, outside of the
District Court’s decision itself and a single page of
Washington’s new-trial motion. We asked the government to
supplement our record with the relevant filings (which are all
sealed on the District Court docket and, as a result, are not
readily available to us), so as to allow for the determination of
the sufficiency of the trial record and a more-searching
review of Washington’s ineffectiveness claim. We thank the
government for filing the supplement.
29
   We note that Washington initially asked for substitution of
counsel, but not a full hearing on trial counsel’s constitutional
effectiveness. A district court is ordinarily required to warn
pro se litigants when a filing recharacterization might
implicate the second-or-successiveness bar of the
Antiterrorism and Effective Death Penalty Act of 1996. See
Castro v. United States, 

, 383 (2003); 28 U.S.C.
§ 2244(b). However, Washington’s recharacterized filing




                               18
                2. Strickland v. Washington and
                   Standard of Review

       “Regardless of whether an ineffective assistance of
counsel claim is raised in a motion for a new trial, on
collateral review, or on direct appeal, the standard of review
is the same.”30      Under the familiar two-part standard
established in Strickland v. Washington,31 Washington bears
the burden of showing 1) that trial counsel’s actions “were
not supported by a reasonable strategy” and 2) that trial


could not be counted as an initial 28 U.S.C. § 2255 motion, as
he was not yet “in custody under sentence of a [federal]
court.” 28 U.S.C. § 2255(a); see also United States v.
Stockstill, 

, 497 n.10 (4th Cir. 1994) (“Because
[the defendant] advanced his claims prior to sentencing, a
§ 2255 motion would not have been appropriate at the
time.”). Because the § 2244(b) bar was not implicated, and
because the mere possibility of preclusion does not otherwise
“significantly alter[]” Washington’s rights, no warning was
necessary here. Mala v. Crown Bay Marina, Inc., 

, 245 (3d Cir. 2013). But see Mui v. United States, 

, 51 (2d Cir. 2010) (“We hold that a defendant who
raises on direct appeal ineffective assistance claims based on
the strategies, actions, or inactions of counsel that can be, and
are, adjudicated on the merits on the trial record, is precluded
from raising new or repetitive claims based on the same
strategies, actions, or inactions in a Section 2255
proceeding.”).
30
     United States v. Bishop, 

, 469 (5th Cir. 2010).
31
     

(1984).




                                19
counsel’s errors were prejudicial.32 “[B]oth deficiency and
prejudice must be proven to have a valid claim for relief.”33
On appeal of the District Court’s decision, we exercise
plenary review over the legal components of ineffectiveness,
assess any underlying findings of fact for clear error, and
“exercise independent judgment on whether those facts, as
found by the District Court, show that counsel rendered
ineffective assistance.”34

        We agree with the District Court that the general
allegations of alcohol use do not require a departure from
Strickland’s two-prong standard—a point conceded by
Washington in his new-trial memorandum.35 Alcohol or drug
use by trial counsel can certainly be relevant to both parts of
an ineffectiveness inquiry, especially if amplified or systemic,
or on close questions of strategy and jury perception. But on
these facts, alleged substance abuse is not, without more, one
of the rare forms of dereliction amounting to the per se denial




32
     

.
33
  United States v. Travillion, 

, 289–90 (3d Cir.
2014).
34
  United States v. Davenport, 

, 608 (3d Cir.
2015).
35
  See Washington New 

; Sealed
Supplemental Appendix 78; see also United States v. Cronic,

, 659–60 & nn.25–26 (1984).




                                 20
of a defendant’s Sixth Amendment right to the effective
assistance of counsel.36

              3. Trial Counsel’s Cross-Examination
                 of Agent Bowman

        Washington now limits his ineffectiveness allegation
to the cross-examination of ATF Agent John Bowman, which
allowed the prosecutor to bring out Washington’s previous
drug conviction on redirect. He argues that trial counsel’s
line of questioning lacked a strategic basis and caused him
prejudice, as it undermined the “not committed to the crime”
theory of defense.

       By way of background: Agent Bowman, who managed
the ATF’s investigation of the conspiracy, was called to
testify as the government’s final witness. His testimony
established, among other things, the authenticity of the
recorded calls and meetings among the conspirators (or
“conspirator,” in the case of the undercover Agent Edwards)
and their incriminating nature. For instance, Bowman

36
   See Williams v. Trammell, 

, 1200–01 (10th
Cir. 2015) (analyzing substance abuse ineffectiveness under
Strickland), cert. denied, 

(2016); Frye v. Lee,

, 907 (4th Cir. 2000) (“[I]n order for an
attorney’s alcohol addiction to make his assistance
constitutionally ineffective, there must be specific instances
of deficient performance attributable to alcohol.”); see also
Berry v. King, 

, 454 (5th Cir. 1985) (“[U]nder
Strickland the fact that an attorney used drugs is not, in and of
itself, relevant to an ineffective assistance claim.” (emphasis
in original)).




                               21
testified that at the March 5 meeting, Berry assured Agent
Edwards that Washington was committed to the robbery plan.

       Trial counsel’s extensive cross-examination of Agent
Bowman dealt in part with inconsistencies in the investigation
and in ATF’s targeting of Washington. Counsel also probed
the racial dimensions of ATF sting operations; Bowman
admitted that he had participated in three Philadelphia sting
operations, all of which targeted only African American
defendants. (A similar response had earlier been elicited from
Agent Edwards, who admitted that perhaps two defendants in
over 13 scenarios were not African American—and both of
those were Latino.)

       Trouble arose when trial counsel began asking
Bowman about Washington’s uneasy fit with the ATF
targeting guidelines’ requirement of prior criminal histories.

      Q: All right. Now we know that you didn’t use – they
      didn’t have my client identified before he was arrested.
      You knew him as Ski, or some other name, right?

      A: Correct.

      Q: So you didn’t know if he had a prior criminal
      history, right?

      A: No, not during the investigation.

      Q: All right. And you found out after the arrest and
      some checking, you found out that my client doesn't
      have a history for robbery, right?

      A: (No verbal response)




                             22
         Q: And he doesn’t have a history for drugs, does he?

         A: I don’t recall.

         Q: If he did, you would recall, sir, wouldn’t you? Isn’t
         that fair?

         A: I don’t want to misstate, but I’m pretty sure he had
         a --

         Q: If you’re not sure, you probably shouldn’t say --

         A: -- drug arrest.[37]

         Q: -- you probably shouldn’t say, you’re not sure. I’ve
         had his record, and I can say, I didn’t see a robbery
         conviction.

         A: I don’t think there’s a robbery conviction, no.

         Q: And I have his record, I didn’t see a drug
         conviction.

         A: I don’t recall.38



37
   Washington argues that the jury twice heard evidence of
Washington’s criminal history, once on direct and once on
rebuttal. As the excerpt shows, however, the initial mention
of Washington’s drug conviction was equivocal—“I’m pretty
sure”—and broached in the context of an arrest, not a
conviction.
38
     S.A. 176–77.




                                  23
         But Washington did have a drug conviction. In fact,
just a few days before Bowman took the stand, the
government had filed its 21 U.S.C. § 851 information
identifying a “prior felony controlled substance violation”
that it intended to use “as the basis for increased punishment”
in the event that Washington was convicted.39

        While Bowman had not directly confirmed
Washington’s criminal history on cross, the prosecutor saw
the door swing open and, on redirect, invited Agent Bowman
to stroll through it:

         Q: [Trial counsel] asked you some questions about Mr.
         Washington’s criminal history.

         A: Yes.

         Q: You said you weren’t sure when he asked you
         specific questions about whether he had a drug
         conviction, whether he had a robbery conviction,
         whether he had a violent crime conviction. You said, I
         don’t recall --

         ...

         Q: You said you weren’t sure, correct?

         A: Correct.

         Q: I want to take a moment and show you Government
         Exhibit 403, 404 and 405. That’s Government Exhibit
         403. Let’s move on to 404. And lastly, we move on to

39
     Section 851 Information, ECF No. 202.




                               24
         Government Exhibit 405. Did you review those three
         exhibits?

         A: Yes.

         Q: And after reviewing them, are you sure whether or
         not Mr. Washington has a prior drug conviction?

         A: He does have a prior drug conviction.40

After this exchange, the issue of Washington’s criminal
history does not appear to have come up again during trial.
Further, trial counsel did not request, and the District Court
did not give, any limiting instruction.

                4. Strickland’s Prejudice Prong

        We may consider the two Strickland prongs in either
order; and, as we have observed, it is “often practical to
consider the prejudice prong first,”41 not the least because we
“prefer[] to avoid passing judgment on counsel’s performance
when possible.”42 Accordingly, we turn first to prejudice,
which requires showing a “reasonable probability”—a
“probability sufficient to undermine confidence in the
outcome”—that, “but for counsel’s unprofessional errors, ‘the
result of the proceeding would have been different.’”43

40
     June 8, 2015 Tr. at 99–100, ECF No. 245.
41
     United States v. Fazio, 

, 426 (3d Cir. 2015).
42
     United States v. Cross, 

, 315 (3d Cir. 2002).
43
   

(quoting 

).




                                25
        At the outset, we agree with the District Court that the
evidence admitted at trial against Washington was daunting
and, generally, damning. His recorded statements alone,
bluster or not, showed a willing and inquisitive member of the
conspiracy. On the day of the robbery itself, Washington
appeared committed to its success.44 Washington attempts to
push back on this reading of the record, but the big picture of
the trial works against him.

       For instance, in support of his argument that the
evidence was not actually “overwhelming,” he points out that
the jury acquitted him of the count-five § 924(c) gun
charge—which, unlike counts 1 through 4, was not a
conspiracy or attempt charge. This is true, but we struggle to
assign it more than limited relevance. The trial evidence
showed that Berry, not Washington, hid the guns in the Eggo
Waffles box, which he then handed to co-conspirator
Johnson.     The guns were found in the minivan, not
Washington’s Chrysler, when the caravan was taken down.
Culpability arguably shifted away from Washington, and he
has not satisfactorily shown how the jury’s apparent doubt
with the firearm count is linked with the quantum of proof on
the remaining counts of the indictment.

       Similarly, Washington points to two jury requests—
one to see the video of the takedown, and another regarding

44
   See Washington New 

(recounting Washington’s concern, during the final pre-
robbery briefing, that co-conspirators Johnson and Ellis had
purchased supplies from a grocery store, where the men could
have been—and were—recorded on the store’s surveillance
system).




                              26
the definition of entrapment or enticement—as indicative of
its hesitance to convict. The video was played back, and both
the prosecution and defense agreed that entrapment was not at
issue. Beyond that, we do not think that the jury’s questions
evince the kind of doubt that might meet Washington’s
burden for showing prejudice. If anything, all we can draw
from the acquittal on this count is that the jury took seriously
its duty to view the trial evidence on a count-by-count basis.45

       Washington also argues that the District Court erred by
failing to separate the Hobbs Act robbery and drug counts in
determining prejudice, contending the testimony about his
drug conviction, and thus his propensity, affected the latter far
more than the former.46 He emphasizes that the defense’s
theory of the case rested in part on caution and lack of
culpable intent, and points to selections of the recordings,
admitted at trial, that show (or so he claims) that he was wary
of cocaine and was not interested in dealing with it or



45
     In fact, the count-five acquittal strikes against
Washington’s claim that the jury used his drug conviction
against him on the grounds of predisposition. Ample
attention was drawn at trial to Washington’s alleged trigger-
happy statements, yet the jury was not convinced of
Washington’s guilt on count five.
46
    While Washington’s PSR grouped the offenses for
sentencing purposes, the District Court did not treat them as a
single unit, imposing separate sentences on the robbery and
drug counts of the indictment. Accordingly, we assume
without deciding that the counts are appropriately
disaggregated for the purposes of Strickland prejudice.




                               27
otherwise becoming involved. In one of these, Washington is
recorded as saying that he “don’t fuck with coke.”47

       Even in light of the defense’s theory of the case,
however, we do not agree that the charges can be so neatly
separated. Washington wants us to view the likelihood of
prejudice from admitting the conviction as higher for the drug
counts than the robbery counts.48 The fundamental flaw of
Washington’s argument is that he never quite explains, in a
way that satisfies his Strickland burden, why he would have
participated in the robbery, or even in its planning stages, if
not for the cocaine. According to the testimony of ATF
Agent Edwards, the “drug courier” told the other members of
the conspiracy that no money would be found in the house.
Even if Washington did not intend to personally handle the
cocaine or move it for sale, he could not help but know that
cocaine was the object of the robbery. Viewed against this

47
     See, e.g., Washington Reply Br. at 7.
48
   While we assume without deciding that Washington could
have prevailed on this theory, we note that a “caution” or
“lack of total commitment” defense is difficult to successfully
mount given the broad liability for drug-conspiracy charges.
See United States v. Caraballo-Rodriguez, 

, 425
(3d Cir. 2013) (en banc) (“To prove a conspiracy, the
government must show: (1) a shared unity of purpose; (2) an
intent to achieve a common illegal goal; and (3) an agreement
to work toward that goal.”); see also Smith v. United States,

, 719 (2013) (explaining withdrawal from a
conspiracy); United States v. Shabani, 

, 17 (1994)
(holding that proof of an overt act is not required in a § 846
conspiracy).




                                28
backdrop, the “I don’t fuck with coke” statement does not
carry the expansive and exculpatory meaning that he would
like to attribute to it. Moreover, we agree with the
government that the broader defense strategy of the case,
which focused on showing that Washington lacked the violent
criminal history required for ATF targeting, was not
necessarily undermined by a fleeting mention of
Washington’s prior drug conviction, especially in light of his
apparent willingness to participate in the broader drug
conspiracy.49

       We do not mean to trivialize the introduction into the
case of Washington’s drug conviction; although we do not
formally reach the Strickland performance prong, we struggle
to perceive a strategic basis for opening the door.
Nevertheless, we agree with the District Court that
Washington has not met his burden, under Strickland, of
showing that the mistake undermined confidence in the jury’s
verdict.50 Accordingly, the ineffective assistance claim fails.


49
  The government notes that no additional details were given
about the drug offense, so the jury did not know its nature or
severity. However, the jury could infer from the line of
questioning that it was not a violent drug offense.
50
   Cf., e.g., Wilson v. Mazzuca, 

, 502, 507 (2d
Cir. 2009) (finding prejudice from admission of criminal
history, in tandem with other errors, where the government
presented a weak case in chief); Gilliam v. Sec’y for the Dep’t
of Corr., 

, 1033–34 (11th Cir. 2007) (per
curiam) (finding no prejudice on § 2254(d) review when
theory of defense was “sufficiently compromised by other
evidence”); Lyons v. McCotter, 

, 532 n.5 (5th




                              29
          B. Mandatory Minimum Due Process
             Challenge

       In challenging his 264-month sentence, Washington
argues that the District Court erred in following the 20-year
mandatory minimum term set forth in 21 U.S.C. § 841(b)(1),
which (as applicable here) kicks in when the quantity of
cocaine is 5 kilograms or above and the defendant has a prior
felony drug conviction. He does not appear to disagree with
the government that, in the ordinary course of things, the
“mandatory” minimum is precisely what it says on the tin.51
Nor does he argue that the facts supporting the mandatory
minimum sentence—an indictment charging 5 kilograms or
more of cocaine, a corresponding jury verdict, and a properly
filed § 851 notice of a prior conviction—were absent or
infirm. Rather, he contends that its application in this kind of
case, where the comprising elements were entirely fictitious
and in the hands of the government, violates his right to due
process.




Cir. 1985) (“[W]e conclude that the prosecutor’s case was far
from overwhelming and that the introduction into evidence of
Lyons’ prior aggravated robbery conviction undermined the
reliability of his present conviction.”).
51
   See, e.g., United States v. Winebarger, 

, 392
(3d Cir. 2011) (“[D]istrict courts are required to sentence
defendants guilty of that crime to a term of imprisonment no
less than the Congressionally prescribed minimum, unless an
explicit exception to the minimum sentence applies.”).




                              30
             1. Standard of Review

       We begin by noting that although Washington did
object to the mandatory minimum at sentencing, he argued
there on the basis of congressional intent, not due process.
The due process argument also does not appear in his three
sentencing memoranda. While Washington’s failure to
develop the constitutional basis for his objection might
ordinarily limit the scope of our review, we retain discretion
to    reach    unpreserved     arguments     in   appropriate
               52
circumstances. Here, the government asks us to conduct de
novo review and responds to Washington’s argument on the
merits. While a party’s concession does not control the
exercise of our discretion, it is certainly a factor we may
consider. Hence, because Washington did raise an objection
to the application of the mandatory minimum sentence, and
the argument that he relied on came within a stone’s throw of
the one he raises now, we will “waive the waiver” and
consider Washington’s claim on the merits.53            As a

52
   See United States v. Turner, 

, 235 (3d Cir.
2013); cf. United States v. Archuleta, 

, 1007
(8th Cir. 2005) (reviewing newly raised constitutional
argument for plain error)
53
  See United States v. Castro-Taveras, 

, 54 (1st
Cir. 2016) (declining to enforce forfeiture when the
government addressed the merits of unpreserved Fifth
Amendment argument); United States v. Pendleton, 

, 948 n.4 (8th Cir. 2016) (“[T]he Government does not
assert forfeiture and instead argues for de novo review on the
merits. Thus, we choose to apply the usual standard for
evaluating the sufficiency-of-the-evidence claim.”); see also
United States v. Jones, 

, 343 (3d Cir. 2016)




                             31
constitutional challenge to the mandatory minimum, it draws
plenary review.54




(“Because we would reach the same result under either
standard of review, we will apply de novo review, which is
more favorable to [the defendant].”).
Our decision in United States v. Joseph, 

(3d Cir.
2013), is not to the contrary.            Joseph “rectif[ied]
imprecisions” in our preservation and waiver jurisprudence,
and clarified too the oft-overlooked distinction between
“issues” and “arguments,” at least as we use those terms in
this Circuit. 

341–42. To the extent the specific
waiver or forfeiture framework in Joseph applies outside of
Fed. R. Crim. P. 12, see 

nn.2–3, it does not limit
our discretion to excuse waiver or forfeiture concerns as we
do here, especially when the government or appellee
overlooks or disregards waiver or forfeiture and instead asks
for review of the merits. See also Government’s Br. in
United States v. Joseph, No. 12-3808, 

, at
*16–20 (invoking waiver).
54
     United States v. Walker, 

, 75 (3d Cir. 2007).




                                32
              2.   Outrageous Government
                   Conduct and Sentencing
                   Factor Manipulation55

      Washington’s due process challenge falls within the
broader category of “outrageous government conduct”—that

55
   The government suggests in its brief that Washington’s
sentencing challenge is foreclosed by a sentence above the
mandatory minimum. See Gov’t Br. at 57–58. We disagree.
The District Court was clearly guided by the mandatory
minimum term on the drug counts in crafting the overall
sentence. As a result, Washington’s challenge remains viable
despite a sentence above the bare minimum authorized by
law. Compare United States v. Cardena, 

,
1001–02 (7th Cir. 2016) (finding that, when the district court
appeared to treat the mandatory minimum as the lower
bracket for determining a below-Guidelines sentence, court
could not say that the mandatory minimum had “absolutely
no effect”), and United States v. Barnes, 

, 98–99
(1st Cir. 2014) (reaching the legality of a mandatory
minimum sentence although the defendant’s net term was 10
years above the minimum because of references throughout to
the mandatory minimum), with United States v. Ramírez-
Negrón, 

, 49 (1st Cir. 2014) (finding no due
process error when a defendant’s sentence was based
“entirely on Guidelines considerations”), and United States v.
Ramos, 

, 1049 (10th Cir. 2012) (concluding
that a defendant lacked standing to challenge constitutionality
of mandatory minimum because the “actual sentence of
eighty-seven months was not affected by the statutorily
prescribed mandatory minimum” but was instead based on
“the § 3553(a) factors and the Guidelines”).




                              33
is, an allegation that the government’s conduct was so
outrageous that due process and fundamental fairness cannot
abide the defendant’s conviction.56 In our hallmark case on
the doctrine, United States v. Twigg, we decided that a meth
scheme that was substantially engineered by the
government—agents supplied precursor chemicals (at a
significant discount), glassware, and a rented farmhouse for a
lab—displayed the requisite level of outrageousness.57
Twigg led to the ultimate sanction: reversal of the defendant’s
conviction.58

       But Twigg, decided in 1978, is apparently one of only
“two reported court of appeals decisions . . . that have deemed
the government’s conduct so outrageous as to violate due
process.”59 We have found no occasion since Twigg in a
published decision to reverse a conviction or invalidate an
indictment on the theory that the government has strayed
outside of the boundaries contemplated by due process.60 In
United States v. Dennis, for instance, we refused to dismiss an
indictment in a reverse sting case not dissimilar to the one


56
  See United States v. Twigg, 

, 378 (3d Cir.
1978).
57
     See 

380–81.
58
     

59
     United States v. Combs, 

, 795 (8th Cir. 2016).
60
   See United States v. Fattah, 

, 813 (3d Cir.
2017) (citing Twigg for the proposition that “[t]his Court has
granted relief on a claim of outrageous government
misconduct only once”).




                                  34
now at bar, while emphasizing the “exceedingly great”
evidentiary burden placed on the challenging defendant.61

       While our Twigg decision recognized an outrageous
government conduct claim in the context of an attack on an
indictment—and, by extension, the fact of the judgment of
conviction itself—other courts have applied similar reasoning
to a narrower universe of sentencing-related claims, often
under the label “sentencing factor manipulation”—although
they have not done so consistently.62 The Eleventh Circuit
described one model of sentencing factor manipulation in
United States v. Ciszkowski:

      [S]entencing factor manipulation occurs when
      the government’s manipulation of a sting
      operation, even if insufficient to support a due
      process claim, requires that the manipulation be
      filtered out of the sentencing calculus.
      Outrageous government conduct would
      necessitate the reversal of a defendant’s
      conviction,      while      sentencing     factor
      manipulation would simply reduce the sentence
      applied to his conduct. . . . When a court filters
      the manipulation out of the sentencing calculus
      before applying a sentencing provision, no


61
   

, 694–95 (3d Cir. 2016); see also United
States v. Mohamud, 

, 435 (9th Cir. 2016)
(recognizing that dismissal is warranted only in “extreme
cases” (citation omitted)).
62
  See United States v. Sed, 

, 229–31 (3d Cir.
2010) (describing the variation across courts of appeals).




                             35
         mandatory minimum would arise in the first
         place.63

       Our previous precedential opinions have declined to
take a definitive stance on the viability of this doctrine in our
Circuit.64 But even assuming without deciding that the
generous Ciszkowski framing of sentencing factor
manipulation should apply—requiring a lesser showing than
an “outrageous conduct” claim, and allowing for a District
Court to depart below the mandatory minimum range—we
find that Washington has failed to demonstrate, on the facts of
this case, that the mandatory minimum should be excised
from the indictment.

       At bottom, Washington argues that the government
was uniquely positioned to determine the salient facts of his
offense, which he was powerless to refute. Working through
its undercover operative and informant, the ATF did indeed
set the amount of the fictitious cocaine (10 kilograms) and

63
   United States v. Ciszkowski, 

, 1270 (11th Cir.
2007); see also United States v. Rivera-Ruperto, 

,
14 (1st Cir. 2017) (“Sentencing factor manipulation occurs
where government agents have improperly enlarged the scope
or scale of a crime. . . . Where the government engages in
such manipulation, we recognize the court’s power to impose
a sentence below the statutory mandatory minimum as an
equitable remedy.” (internal alterations, quotation marks, and
citations omitted)). But see United States v. Lange, 

, 1296 (11th Cir. 2017) (observing that the Eleventh
Circuit “has never reduced a sentence on the basis of
sentencing factor manipulation”).
64
     See 

–31.




                               36
played up the likelihood of resistance (thereby encouraging
the conspirators to arm themselves).

       But even assuming some impropriety here on the part
of the government, most of the factors it created for the crime,
and which were within its unique control, were not the drivers
of Washington’s actual sentence. Agent Edwards told the
conspirators that they would encounter resistance, so they
brought guns—and, had Washington been convicted of the
gun charge, he would have faced an additional mandatory
consecutive term.65 But he was not. Further, Agent Edwards
told the conspirators that they could expect to recover 10
kilograms of cocaine in the robbery, corresponding to 2014
Guidelines base offense level of 30.66 However, because he
was a career offender, Washington’s Guidelines range was
not governed directly by the 10 kilogram drug-quantity
amount—and the District Court sentenced him far below the
recommended Guidelines range anyway.67

       Instead, the 20-year mandatory minimum was the
product of two factors: the 5 kilograms of cocaine charged in
the indictment and found by a jury, and the § 851 statement



65
     See 18 U.S.C. § 924(c)(1).
66
     See U.S.S.G. § 2D1.1(c)(5) (2014).
67
  To the extent the government manipulated factors that have
not been shown to prejudice Washington, the weight of those
factors is diminished. Cf. Werts v. Vaughn, 

,
198 (3d Cir. 2000) (evaluating prosecutorial misconduct due
process claim for presence of prejudice).




                                  37
filed by the government.68 The latter, as the Supreme Court
has indicated, is a matter of discretion “similar to the
discretion a prosecutor exercises when he decides what, if
any, charges to bring against a criminal suspect . . . and is
appropriate, so long as it is not based upon improper
factors.”69 Washington does not argue that the process
envisioned by § 851 was not properly followed or was based
on impermissible considerations.70


68
   The career offender Guideline itself is based on the offense
statutory maximum—here, life in prison, with or without the
§ 851 enhancement—so in that sense the Guidelines
sentencing range was determined by a drug quantity. See
U.S.S.G. § 4B1.1(b) (2014); 21 U.S.C. 841(b)(1)(A). Again,
though, the District Court did not sentence in accordance with
that range, and—as we discuss infra—the 5 kilogram amount
is far below what courts have approved in other cases.
69
   United States v. Labonte, 

, 762 (1997); see
also United States v. Sanchez, 

, 671–72 (2d Cir.
2008) (rejecting due process challenge when the government
filed § 851 notice against one defendant, but not his
codefendants).
70
   That is not to say that we affirmatively endorse the
prosecution’s decision here, which has the unavoidable
appearance of punishing Washington for exercising his right
to go to trial. But on these facts, this is not enough to declare
the government’s actions beyond the pale or invidiously
motivated, especially with the longstanding recognition—
both by us and by the Supreme Court—of the deference
afforded to prosecutorial decisions. For better or worse,
prosecutors have a great deal of power to use specific




                               38
       So it comes down, in the end, to the drug quantity. We
acknowledge Washington’s concerns, which are well stated
and logical, that the drugs did not exist, and that his ironclad
mandatory minimum has no real-world foundation. Other
courts of appeals, however, have roundly rejected claims that
amounts greater than 5 kilograms, or even 10 kilograms,
amount to sentencing factor manipulation.71 Further, Agent

charging decisions to guide mandatory sentencing exposure.
By way of example, a defendant in one recent New Jersey
stash house case was charged in part with conspiring to
possess with intent to distribute more than 5 kilograms of
cocaine, exposing him to the mandatory minimum term.
When the defendant agreed to plead guilty, the government
filed a superseding information that simply deleted the drug
quantity from the conspiracy charge, thereby eliminating the
mandatory minimum. See U.S. Dep’t of Justice Press
Release, Burlington County, New Jersey, Man Sentenced To
Eight Years In Prison For Scheme To Rob Drug Dealers At
Gunpoint,       https://www.justice.gov/usao-nj/pr/burlington-
county-new-jersey-man-sentenced-eight-years-prison-
scheme-rob-drug-dealers (Feb. 8, 2017; archived at
https://perma.cc/Y5XD-UULW); United States v. Forman,
D.N.J. Crim. No. 1:14-cr-00152, ECF Nos. 27, 81.
71
   See United States v. Hare, 

, 102–03 (4th Cir.
2016) (collecting cases for the proposition that “15 to 20
kilograms of cocaine” amounts to “considerably less than the
quantity of cocaine at issue in other stash house sting cases”);
United States v. Sanchez, 

, 1414 (11th Cir.
1998) (“The fact that the government’s fictitious reverse sting
operation involved a large quantity of drugs does not amount
to the type of manipulative governmental conduct warranting
a downward departure in sentencing.”).




                              39
Edwards testified at trial that the amount chosen for the sting
was a “conservative” number based upon the drug weights
found in “a typical [Philadelphia] stash house.”72 He
explained that the proposed scenario “always has to be
realistic” or it might be questioned by the robbery crews.73
Washington has not offered anything to the contrary. Put
simply, there is not enough here for us to conclude that the

72
     June 3 Tr. at 84.
73
   June 3 Tr. at 85–86. These statements were made in the
context of Agent Edwards’s trial testimony, not at sentencing.
It does not appear that the justifications for the amount
chosen were re-raised at sentencing. We acknowledge that
Agent Edwards’s testimony indicates that all Philadelphia
stash-house stings crafted in accordance with ATF
methodology will involve, in some sense, an amount above
the mandatory minimum threshold. Insufficient evidence was
presented to allow the determination of whether a lesser
quantity, below the mandatory minimum amount, would have
sufficed to entice a four-man crew. See, e.g., June 3 Tr. at
129 (testimony by Agent Edwards that his “courier” wanted
only one to one and a half kilogram as a nonparticipant). But
see 

(“Although it is certainly
feasible that . . . the agents could have used some lesser
quantity of drugs and still made the deals look realistic, the
mere fact that they did not, without more, does not establish
that the agents engaged in the kind of extraordinary
misconduct . . . that is required of a successful sentencing
manipulation claim.” (internal quotation marks and citations
omitted)). A district court is, of course, free to probe this
reasoning, especially if culpability or entrapment are raised as
specific defenses.




                              40
government chose the 10 kilogram amount primarily, or even
secondarily, “to inflate [Washington’s] sentence upon a
conviction.”74

        Washington encourages us to follow the reasoning of
United States v. McLean, in which a different judge in the
Eastern District of Pennsylvania sentenced below the
mandatory minimum, on due process grounds, in a reverse-
sting stash house case.75 McLean, which is nonbinding,76 is
also distinguishable. The defendant there received a “split”
jury verdict on the amount of cocaine involved: 5 kilograms
with regard to conspiracy but 500 grams with regard to
attempt.77 We detect no equivalent ambiguity in the jury’s
verdict on Washington’s ultimate culpability, and therefore
reject this argument.78


74
     

.
75
     

, 942–45 (E.D. Pa. 2016).
76
  The government sought to appeal the McLean sentence, but
(as the government explained at oral argument) was unable to
obtain the Solicitor General’s permission to pursue the
appeal. See C.A. No. 16-3227 (order dismissing appeal
entered Sept. 15, 2016). The defendant appealed the
judgment of conviction, which we recently affirmed. See
generally United States v. McLean, No. 16-2993, 

(3d Cir. Aug. 3, 2017) (nonprecedential).
77
     See 

–40 & n.13.
78
  We note that McLean contains an extensive recitation of the
facts and factors that caused its district court to depart below
the mandatory minimum. While constitutional challenges to




                               41
        In sum, we conclude that the 5 kilograms of cocaine
charged in the indictment and found by the jury did not
amount to an impermissible manipulation of sentencing
factors by the government. To the extent that the fictitious 10
kilogram quantity is relevant, we find too that Washington
has shown neither improper manipulation nor prejudice.
Nevertheless, we remind the government that we have
expressed misgivings in the past about the wisdom and
viability of reverse stash house stings. That this case fell on
the safe side of the due process divide should not be taken to
indicate that all such prosecutions will share the same fate.
As one of our colleagues said in a prior case, “I do not find it
impossible for the Government to exercise its discretion
rationally to set up stash house reverse stings. But I share the
concern that this practice, if not properly checked, eventually
will find itself on the wrong side of the line.”79

          C. Selective Enforcement Discovery Claim

        Finally, Washington appeals in part the denial of his
pretrial motion for discovery, which he filed in order to
“prepare a motion to dismiss the indictment on the basis of
racial profiling and/or selective prosecution of racial
minorities by the ATF Office in Philadelphia, in conjunction


mandatory minimum sentences draw de novo review, it might
be the case that a district court’s factfinding and underlying
reasoning, as opposed to its application of a legal standard,
may be due some level of deference. We need not resolve the
question in this appeal.
79
  See 

(Ambro, J., concurring in part
and dissenting in part).




                              42
with the local U.S. Attorney’s Office.”80 He contends that the
District Court erred in applying a strict discovery standard—
Armstrong/Bass—to the portions of his motion that pertained
to law enforcement and ATF material on stash-house reverse
stings, as opposed to those portions (the denial of which he
does not appeal) that sought information related to the
prosecution of those offenses.         Instead of employing
Armstrong/Bass, Washington contends, we should follow the
Seventh Circuit’s opinion in United States v. Davis, which
appeared to depart from the Armstrong/Bass model for claims
of selective enforcement in stash house cases.

       While discovery rulings are ordinarily reviewed for
abuse of discretion, “we exercise de novo review over the
standards the district court used in exercising its discretion.”81
And although we decline to adopt Davis wholesale, we

80
     Washington Discovery, 

, at *2.
81
   Redland Soccer Club v. Dep’t of the Army, 

,
845 (3d Cir. 1995); see also Koon v. United States, 

, 100 (1996) (“A district court by definition abuses its
discretion when it makes an error of law.”), superseded by
statute on other grounds as stated in United States v.
Thurston, 

, 70 (1st Cir. 2004). Although the
government disputes whether Washington’s appellate claim
matches what he raised below, its response brief generally
answers on the merits; the procedural objection is to the scope
of his request, not the consistency of his legal theory. We are
satisfied, from our review of the record, that Washington
adequately developed the claim across his District Court
submissions.




                               43
nevertheless agree with the Davis court that district judges
have more flexibility, outside of the Armstrong/Bass
framework, to permit and manage discovery on claims like
Washington’s. Accordingly, as explained further below, we
will vacate the District Court’s discovery orders and issue a
limited remand for further post-judgment proceedings.

             1. Substantive Equal Protection
                Claims: “Clear Evidence” of
                Discriminatory Effect and
                Intent

       Washington’s argument rests on the distinction
between “selective prosecution” and “selective enforcement,”
labels that we (and others) sometimes deploy
interchangeably. Here, we use them as Washington does.
“Prosecution” refers to the actions of prosecutors (in their
capacity as prosecutors) and “enforcement” to the actions of
law enforcement and those affiliated with law-enforcement
personnel.

        We start with a point of commonality. Substantive
claims of selective prosecution and selective enforcement are
generally evaluated under the same two-part test, which is
derived from a line of seminal Supreme Court cases about the
collision between equal protection principles and the criminal
justice system.82     A defendant challenging a criminal
prosecution at either the law enforcement or prosecution

82
   See Whren v. United States, 

, 813 (1996)
(“[T]he Constitution prohibits selective enforcement of the
law based on considerations such as race.”); Wayte v. United
States, 

, 608 (1985).




                             44
inflection points must provide “clear evidence” of
discriminatory effect and discriminatory intent (the latter is
sometimes referred to as “discriminatory purpose”).83
Meeting this standard generally requires evidence that
similarly situated individuals of a difference race or
classification were not prosecuted, arrested, or otherwise
investigated.84



83
   See, e.g., United States v. Taylor, 

, 197 (3d
Cir. 2012); Harajli v. Huron Twp., 

, 508 (6th
Cir. 2004); United States v. Alameh, 

, 173 (2d
Cir. 2003); see also United States v. Whitfield, 649 F. App’x
192, 196 n.11 (3d Cir. 2016) (nonprecedential) (“[T]he prima
facie elements for both selective prosecution and selective
enforcement are the same: discriminatory effect and
discriminatory intent.”), cert. denied, 

(2017);
Lacey v. Maricopa Cty., 

, 920 (9th Cir. 2012)
(discussing a civil selective enforcement claim); Hill v. City
of Scranton, 

, 125 (3d Cir. 2005) (same);
Marshall v. Columbia Lea Reg’l Hosp., 

, 1168
(10th Cir. 2003) (“These standards have been applied to
traffic stops challenged on equal protection grounds.”). We
cite Whitfield for its description of the law in our Circuit and
do not assign it the weight of precedent.
84
   See United States v. Hare, 

, 98–99 (4th Cir.
2016); United States v. Brantley, 

, 1271 (11th
Cir. 2015); Johnson v. Crooks, 

, 1000 (8th Cir.
2003) (“When the claim is selective enforcement of the traffic
laws or a racially-motivated arrest, the plaintiff must normally
prove that similarly situated individuals were not stopped or
arrested in order to show the requisite discriminatory effect




                              45
                2. Armstrong/Bass: “Some Evidence”

       A criminal defendant, however, will not often have
access to the information, statistical or otherwise, that might
satisfy a “clear evidence” burden. Thus, the two component
cases that make up the Armstrong/Bass test—United States v.
Armstrong85 and United States v. Bass86, both of which arose
from selective prosecution challenges—propounded a facially
less rigorous standard for criminal defendants seeking
discovery on an anticipated selective prosecution claim.
Instead of “clear evidence,” a successful discovery motion
can rest on “some evidence.”87 “Some evidence” must still
include a showing that similarly situated persons were not
prosecuted.88     Furthermore, under Armstrong/Bass, the


and purpose.”); see also Gov’t of V.I. v. Harrigan, 

, 36 (3d Cir. 1986).
85
     

(1996).
86
     

(2002) (per curiam).
87
   See 

(“[A] defendant who seeks
discovery on a claim of selective prosecution must show
some evidence of both discriminatory effect and
discriminatory intent.”); see also United States v. Arenas-
Ortiz, 

, 1068 (9th Cir. 2003) (“The showing
necessary to obtain discovery is somewhat less” than
prevailing on the merits).
88
   

; 

(“Under
Armstrong, therefore, because respondent failed to submit
relevant evidence that similarly situated persons were treated
differently, he was not entitled to discovery.”).




                               46
defendant’s showing must be “credible” and cannot generally
be satisfied with nationwide statistics.89

        Armstrong/Bass has proven to be a demanding
gatekeeper. In developing it, the Supreme Court sought to
“balance[] the Government’s interest in vigorous prosecution
and the defendant’s interest in avoiding selective prosecution”
by creating a standard that, while difficult to meet, derived
from “ordinary equal protection standards” and was not
“insuperable.”90     The lived experience, however, has
resembled less a challenge and more a rout, as practical and
logistical hurdles abound—especially to proving a negative.91

89
   See 

; United States v. Thorpe,

, 657 (6th Cir. 2006); see also United States v.
Al Hedaithy, 

, 607–08 & n.24. (3d Cir. 2004)
(rejecting under Armstrong/Bass a selective prosecution
discovery request premised on “numerous newspaper
articles” showing rampant cheating on the Test of English as
a Foreign Language exam; “[t]he defect in Al Hedaithy’s
proffer is that none of this evidence indicates that similarly
situated persons were treated differently. Demonstrating that
thousands of other people have also cheated on the [] exam
does nothing to identify persons who are similarly situated”).
90
     

, 470.
91
   See, e.g., Donna Coker, Foreword: Addressing the Real
World of Racial Injustice in the Criminal Justice System, 93 J.
Crim. L. & Criminology 827, 828–29, 846–47 (2003)
(discussing, among other things, the problems with the
“similarly situated” discovery standard, including the
possibility that the “data . . . may simply not exist” or is in
“the exclusive control of the government”); Richard H.




                               47
The government itself concedes that “neither the Supreme
Court nor this Court has ever found sufficient evidence to
permit discovery of a prosecutor’s decision-making policies
and practices.”92

       So, too, in Washington’s case, as the District Court
here found that his discovery motion had fallen short of
Armstrong/Bass. His list of three prior stash house cases, the
District Court determined, revealed nothing about similarly
situated individuals who were not ensnared in stash-house
stings, and Washington had otherwise not shown
discriminatory intent/purpose.93

                3. Armstrong/Bass in “Selective
                   Enforcement” Cases

      On appeal, Washington does not argue that the District
Court’s Armstrong/Bass analysis was wrong, but rather that

McAdams, Race and Selective Prosecution: Discovering the
Pitfalls of Armstrong, 73 Chi.-Kent L. Rev. 605, 640 (1998)
(“The Armstrong holding and the implications of its
reasoning create a barrier to discovery that, for the great
majority of criminal cases, is insuperable.”); 

; see also Whitfield, 649 F. App’x at 196 n.11;
Erik Luna, Transparent Policing, 

. Rev. 1107, 1139
(2000) (“The bar for selective enforcement and prosecution
claims has been set at a nearly unreachable height for the vast
majority of criminal defendants, an example of an abstract
right with no practical remedy.”).
92
     Gov’t Br. at 31.
93
     Washington Discovery, 

, at *7.




                              48
Armstrong/Bass—which arose from discovery aimed at
claims of selective prosecution, not selective enforcement—
should not have applied at all to the subset of his claims
seeking law-enforcement evidence.             “The sort of
considerations that led to the outcome in Armstrong,” he
contends, “do not apply to . . . ATF agents engaged in racial
discrimination when selecting targets for sting operations, or
when deciding which suspects to refer for prosecution.”94
Washington also points to the difficulty of obtaining pre-
discovery statistics in selective prosecution cases, arguing
that requiring the same in law-enforcement cases—when
there are likely to be no records of similarly situated
individuals who were not arrested or investigated—would
transform the functional impossibility of Armstrong/Bass into
a complete impossibility.95 While substantive selective
prosecution and enforcement cases must ultimately reach the
same destination—“clear evidence” of discriminatory
purpose/intent and effect—Washington suggests that
enforcement cases, which do not implicate the heightened
protections afforded to prosecution decisions, should be
permitted to travel on a less rocky path.

       We have not previously addressed this particular
prosecution/enforcement distinction in a precedential
decision.96 And it is true that Armstrong and Bass, both of

94
     Washington Br. at 19.
95
     See 

820 F.3d at 100).
96
  See Whitfield, 649 F. App’x at 196 n.11. The government
says that we have been “reluctant to permit discovery into the
government’s investigatory and prosecutorial practices
without a substantial showing by the defendant.” Gov’t Br. at




                                 49
which arose from selective prosecution claims, were
grounded in part on the special solicitude courts have shown
to prosecutors’ discretion. For instance, the Armstrong Court
said that a “selective-prosecution claim is not a defense on the
merits to the criminal charge itself, but an independent
assertion that the prosecutor has brought the charge for
reasons forbidden by the Constitution,” and because of the
great deference owed to prosecutorial decision-making, the
Court was reluctant to abrogate the “background
presumption” that “the showing necessary to obtain discovery
should itself be a significant barrier to the litigation of
insubstantial claims.”97

       Other courts of appeals, however, have extended the
reasoning of Armstrong/Bass to claims of selective
enforcement and have applied the same burden (“some
evidence”) to the related discovery requests. The Fourth and
Tenth Circuits are two,98 and until recently the Seventh

31. While that is true, the two cases the government cites—
Al Hedaithy and United States v. Abuhouran, 

(3d Cir. 1998)—discussed prosecutorial decision-making,
such as the (nonconstitutional) challenge to substantial
assistance motions in Abuhouran, 

. They
do not provide a definitive answer to the question here:
whether we may look behind the law-enforcement curtain.
97
  

–64 (internal quotation marks
and citation omitted).
98
  See United States v. Alcaraz-Arellano, 

, 1264
(10th Cir. 2006) (“[Armstrong’s] elements are essentially the
same for a selective-enforcement claim.”); United States v.
Mason, 

, 829–30 (4th Cir. 2014).




                              50
Circuit appeared to be another. In United States v. Barlow,
the Seventh Circuit addressed a discovery claim based on
racial profiling, a “selective law enforcement tactic.”99 In
deciding that the District Court had not abused its discretion
in denying discovery, the Barlow court followed Armstrong
(Bass had not yet been issued), finding that defendant Barlow
had not presented relevant and reliable data on the “similarly
situated” prong of the test.100

                4. The Seventh Circuit’s Davis
                   Decision

        But in United States v. Davis,101 the en banc Seventh
Circuit appeared to narrow the scope of Armstrong/Bass.
Davis was an appeal from a pretrial order granting discovery
in a stash-house reverse-sting case.102 The defendants had
alleged that the prosecutor, FBI, and ATF had engaged in
racial discrimination, pointing to some discomfiting statistics:
out of 94 defendants across 20 Northern District of Illinois
stash-house sting prosecutions, only six were non-Hispanic


99
     

, 1010 (7th Cir. 2002).
100
      See 

101
      

(7th Cir. 2015).
102
   We need not address Davis’s procedural intrigue, although
we note that it marked the dividing line between the en banc
majority and dissent. See 

(Rovner, J.,
joined by Hamilton, J., dissenting) (“For all of the prudential
reasons that we do not permit civil litigants to manufacture
appellate jurisdiction, we should not allow an appeal based on
the sort of non-final dismissal that was fabricated here.”).




                                51
whites.103 The statistics, however, revealed nothing about
similarly situated persons who were not prosecuted.104
Nevertheless, the district court granted a broad discovery
order, reasoning in part that the “overwhelming majority” of
those prosecuted being persons of color met, by inference, the
defendants’ burden under Armstrong/Bass.105

       The Seventh Circuit agreed with the government that
the district court’s reasoning was inconsistent with
Armstrong. If Armstrong’s record, which showed the
exclusive prosecution of African Americans for crack
offenses, was not sufficient, then a showing that “three-
quarters of the defendants in stash-house cases have been
black does not suffice.”106

       The Seventh Circuit then addressed whether
Armstrong/Bass was the relevant test at all. In this Circuit’s
view, the key distinction lay between prosecutors, who are
“protected by a powerful privilege or covered by a
presumption of constitutional behavior” recognized by
Armstrong, and FBI/ATF agents, who “regularly testify in
criminal cases” and whose “credibility may be relentlessly
attacked by defense counsel.”107 For these and other reasons,
the Seventh Circuit decided that “the sort of considerations

103
      

–15.
104
      See 

105
   See Order at 2, United States v. Davis, N.D. Il. Crim. No.
13-cr-63-2 (order entered October 30, 2013).
106
      

–20.
107
      

                               52
that led to the outcome in Armstrong do not apply to a
contention that agents of the FBI or ATF engaged in racial
discrimination when selecting targets for sting operations, or
when deciding which suspects to refer for prosecution.”108

       Having ruled that Armstrong/Bass did not quite govern
the law-enforcement aspects of the defendants’ discovery
request, the Seventh Circuit decided that the District Court’s
comprehensive discovery order was nonetheless an abuse of
discretion. Sweeping and overbroad, the order engulfed too
much that did implicate prosecutorial discretion and was not
tailored to the boundaries of the case nor the scope of the
defendants’ proffer.109

        On remand, instead of issuing a “blunderbuss order,”
the district court was ordered to take “measured steps” to
determine the scope and boundaries of discovery.110 First, the
district court was to determine whether there was reason to
believe that race played a role in the investigation—that
“forbidden selectivity occurred or plausibly could have
occurred”111—by evaluating the evidence already of record,
new evidence acquired by the defendants, and (if necessary)
the affidavits and limited testimony of case agents. If the
inquiry gave the district court reason to believe that similarly
situated persons would not have been pursued by law
enforcement, in camera disclosure of targeting criteria might


108
      

109
      See 

110
      

at 723.




                              53
be called for. If the trail of breadcrumbs continued,
additional targeted inquiries might be justified; and if the
obtained information crossed the Armstrong/Bass threshold,
the discovery could be “extended to the prosecutor’s
office.”112




112
    

Although it is of limited relevance to the
actual legal issue on appeal, the “switch” in Davis arose after
years of unease in Seventh Circuit district courts—and in the
Northern District of Illinois in particular—about reverse sting
cases. See, e.g., United States v. Paxton, No. 13 CR 103,

, at *5 (N.D. Ill. Apr. 17, 2014) (granting
discovery under Armstrong/Bass, in part because “no white
defendants have been indicted for phony stash house cases
since 2009, despite the diverse makeup of the Northern
District of Illinois”). Post-Davis, the controversy continues.
See Jason Meisner & Annie Sweeney, Lawyers: ATF Stings
Racially Biased; U. of C.-led Team says Stash House Cases
Show Feds Unfairly Targeted Minorities, Chi. Trib., Mar. 5,
2017,             at          C1,            available         at
http://www.chicagotribune.com/news/local/breaking/ct-atf-
stash-house-sting-racial-discrimination-met-20170303-
story.html (last visited Aug. 21, 2017; archived at
https://perma.cc/XY4G-MKYG). A report in one pending
case, prepared by Columbia Law professor Jeffrey Fagan,
concludes among other things that “race remains a
statistically significant predictor of selection as a Stash House
defendant.” See Report of Jeffrey Fagan, United States v.
Alfred Washington, N.D. Il. Crim. No. 12-CR-632, ECF No.
510-2.




                               54
              5. Davis’s Application to
                 Washington’s Claims

      In sum, despite not being a straightforward affirmance
of a pro-defendant discovery decision, Davis does more or
less what Washington would like this Court to do: find
Armstrong/Bass inapplicable in part and send the case back to
the District Court to make additional inquiries—bolstered,
perhaps, by whatever evidence has become available since.

        However, there are good reasons to be cautious about
Davis, and its practical application in this case is not quite as
straightforward as Washington suggests. While the Seventh
Circuit did not follow Armstrong/Bass, Davis does not clearly
state whether the test adopted in its stead was a variation of
Armstrong/Bass or, alternatively, was intended to be a
complete departure. For instance, Davis does not explicitly
discuss the discriminatory purpose/intent prong of the
traditional Armstrong/Bass analysis.113 Davis might therefore
be fairly described as an opinion entirely about discriminatory
effect as a gateway to discovery. Moreover, Davis does not
mention the Seventh Circuit’s earlier decision in Barlow at
all—not to harmonize it, distinguish it, or explicitly overrule
it.114 Davis also arose on a different posture, where the

113
    It is perhaps true that, in a given investigation, a finding
that a defendant would not have been prosecuted if he had
been non-Hispanic white is enough to suggest an inference of
discriminatory purpose/intent.
114
    At least one court has observed this ambiguity in the
Seventh Circuit’s case law in declining to adopt the
prosecution/enforcement discovery dichotomy. See United
States v. Lamar, No. 14 CR 726, 

, at *5 n.3




                               55
defendant had prevailed below and, thus, benefitted from
partial appellate deference to the trial court’s exercise of
discretion. Here, by contrast, the District Court’s decision
was not favorable to Washington; this Court’s deference thus
tips the other way. The Davis framework was further
influenced by the Seventh Circuit’s review of a pretrial
decision, as indicated by the court’s repeated references to
expediency—“limited inquiries that can be conducted in a
few weeks” so as to not “sidetrack[]” the case.115 While any
framework must be mindful of the pretrial context in which
discovery motions will be filed and decided, we are reviewing
a final judgment, one which (as discussed further below) is
not unwound if we decide to remand.

                6. Strict Application of Armstrong/Bass
                   is Inappropriate

       Despite our caution, we find ourselves in agreement
with the core rationale of Davis: the special solicitude shown
to prosecutorial discretion, which animated the Supreme
Court’s reasoning in Armstrong and Bass—and our own
reasoning in our pre-Armstrong/Bass case law on the same
subject116—does not inevitably flow to the actions of law

(S.D.N.Y. Aug. 7, 2015). The government, for its part,
argues that Davis is wrongly decided, and points in particular
to the Barlow that did not bark. See Gov’t Br. at 39 n.13.
115
      

.
116
   See, e.g., United States v. Torquato, 

, 569–70
(3d Cir. 1979) (discussing the need to “minimize the intrusion
on the prosecutorial function” in the context of the burden
required to obtain an evidentiary hearing); see also In re
Grand Jury, 

, 1030 (3d Cir. 1980).




                                56
enforcement, or even to prosecutors acting in an investigative
capacity. Prosecutors are ordinarily shielded by absolute
immunity for their prosecutorial acts,117 but police officers
and federal agents enjoy no such categorical protection.118
And, as the Davis court observed, officers and agents are
expected to testify in criminal cases, with their honesty and
candor “open to challenge.”119         That aspects of law
enforcement and prosecutorial discretion are often
intertwined does not make the distinction between the two
realms any less legitimate; courts are often called upon to
determine whether specific acts fall more into one category
than the other.120


117
   See, e.g., Buckley v. Fitzsimmons, 

, 268–71
(1993); Odd v. Malone, 

, 208–09 (3d Cir. 2008).
118
    See Orsatti v. N.J. State Police, 

, 483 (3d Cir.
1995); Forsyth v. Kleindienst, 

, 1216 (3d Cir.
1979) (“[F]ederal law enforcement officers are entitled only
to qualified, or good faith, immunity.”).
119
      

.
120
    The government suggests that when a district court is
presented with mixed claims, or some selective enforcement
and some selective prosecution claims, applying the
Armstrong/Bass standard across the board is appropriate.
Gov’t Br. at 28. This contention was rejected by 

(observing that, if the “measured steps” discovery
rises to the level required by Armstrong/Bass, the
investigation can “extend[] to the prosecutor’s office”), and
we agree that it unduly penalizes a defendant who casts a
wide net. That said, it remains within the discretion of a
district court—and, indeed, remains within the discretion of




                                57
       A challenge to a law-enforcement policy also
implicates another area where immunity is limited. The ATF
reverse sting model is familiar to us and other courts precisely
because it is a defined operation, one with policies, manuals,
targeting criteria, and standards. Its appearance from coast to
coast is not some kind of convergent law-enforcement
evolution, but instead is due to the promulgation of official
policies by a federal agency. Claims of unconstitutional
policies or practices, lodged against entities rather than
individuals, often cannot be met with qualified or good-faith
immunity defenses at all.121

        In sum, while we do not lightly depart from the well-
established Armstrong/Bass framework, the enforce-
ment/prosecution distinction is a legitimate one, and we
therefore join the Davis court in finding Armstrong/Bass to be
distinguishable on these facts. Accordingly, motions for
discovery seeking information on putative claims of
unconstitutional selective enforcement are not governed by
strict application of the Armstrong/Bass framework.

      Nevertheless, and as tacitly acknowledged by the
Seventh Circuit, courts contemplating motions for discovery

this District Court—to determine that a “selective
enforcement” claim was either not appropriately raised or was
simply a prosecution claim tailored to avoid the requirements
of Armstrong/Bass. As always, a court must look beyond the
labels affixed by the party and focus on the substance of what
is sought.
121
   See Carver v. Foerster, 

, 102–04 (3d Cir.
1996) (citing, among other things, Owen v. City of
Independence, 

(1980)).




                              58
on selective enforcement claims must still be guided by the
spirit of Armstrong/Bass, which incorporates the demands
placed on the underlying substantive claims: not just “some
evidence,” but the heightened “clear evidence” standard.
Further, while we agree with a general approach of taking
“measured steps” over the course of discovery, we decline to
mandate a precise system or order that a district court must
follow. As we have often said, matters of docket control and
discovery are committed to broad discretion of the district
court.122 We are confident in the ability of district courts to
react to the particular circumstances of a case—the likelihood
of a near-term trial date, the complexity of the underlying
matter, the strength of a defendant’s discovery proffer, the
similarity to previous cases raising similar concerns, the need
to avoid overly prejudicial or irrelevant disclosure, and so
on—in crafting a measured approach to discovery. Finally,
we note that although we are now in a post-trial posture, the
fact of the matter is that most, if not all, appeals from criminal
discovery orders will be properly brought only after judgment
is entered.123




122
    See, e.g., Sempier v. Johnson & Higgins, 

, 734
(3d Cir. 1995); In re Fine Paper Antitrust Litig., 

, 817 (3d Cir. 1982); United States v. Newman, 

, 739 (3d Cir. 1973) (referring to criminal discovery
rulings under Fed. R. Crim. P. 16).
123
    See United States v. Sciarra, 

, 627–28 (3d
Cir. 1988).




                               59
              7. Selective Enforcement Discovery
                 Standard

        We therefore hold as follows. In ruling on a pretrial
discovery request that alleges selective prosecution and/or
selective enforcement, a district court applies Armstrong/Bass
to claims that implicate protected prosecutorial functions,
such as those that arose in the namesake cases. If claims of
selective law enforcement are raised, or there are “mixed”
claims that involve prosecutors acting in investigative or
other capacities (in short, performing functions that ordinarily
would not draw absolute immunity), the standard guiding the
district court’s discretion is different. While Armstrong/Bass
remains the lodestar, a district court retains the discretion to
conduct a limited pretrial inquiry into the challenged law-
enforcement practice on a proffer that shows “some
evidence” of discriminatory effect. The proffer must contain
reliable statistical evidence, or its equivalent, and may be
based in part on patterns of prosecutorial decisions (as was
the case in Davis) even if the underlying challenge is to law
enforcement decisions.124 Distinct from what is required
under Armstrong/Bass, a defendant need not, at the initial

124
     We do not reach the question of the geographical
boundaries of the initial evidence the defendant must
provide—whether, in other words, the application of a law
enforcement policy or practice in the defendant’s specific
district might be contextualized by its application elsewhere,
so long as the defendant adequately connects the practice
elsewhere to his or her situation. We leave this issue to the
district court’s discretion and common sense, in light of the
need to show that the policy ultimately acted upon, or did not
act upon, persons similarly situated to the defendant.




                              60
stage, provide “some evidence” of discriminatory intent, or
show that (on the effect prong) similarly situated persons of a
different race or equal protection classification were not
arrested or investigated by law enforcement. However, the
proffer must be strong enough to support a reasonable
inference of discriminatory intent and non-enforcement.

        If a district court finds that the above has been met, it
may conduct limited inquiries of the sort recommended in
Davis, and cabined to the same considerations of judicial
economy and the need to avoid protracted pretrial litigation of
matters collateral to the upcoming trial—as well as the need
to avoid impinging on other areas of executive privilege.
Areas of consideration could include the testimony, in person
or otherwise, of case agents or supervisors, and the in camera
analysis of policy statements, manuals, or other agency
documents. Relevant information, having passed the filter,
can also be disclosed to the defendant, although the district
court retains discretion to forgo disclosure of or otherwise
restrict the use of information that, while relevant to a
selective enforcement claim, might not ordinarily be the sort
of discovery material available to a criminal defendant under
Fed. R. Crim. P. 16 or Brady and its progeny.

       Throughout, the district court must be mindful that the
end “goal” of such a discovery motion is a valid claim of
selective enforcement under the heightened substantive
standards, which we are not asked to diminish or distinguish.
If the district court’s initial or secondary inquiry sees that
destination recede or stand still, not advance, the court
operates within its discretion to deny additional discovery and
to proceed to trial.




                               61
        That limited discovery of this sort may be granted in
one case does not guarantee—and should not guarantee—that
it will be granted in another, similar case, even within the
same district.125 But courts may, of course, consider the
product of earlier investigations in deciding whether to
conduct pretrial discovery on the individual claims they
happen to confront.

              8. Remand is Necessary for the
                 District Court to Exercise its
                 Discretion under the Correct
                 Framework

        Having set forth the governing standard for selective
enforcement cases, we address its application to
Washington’s case. It is clear that the District Court thought
itself bound by the more-demanding Armstrong/Bass standard
across the entirety of Washington’s discovery request, and
then again on reconsideration. Because it exercised its
discretion under the incorrect standard, we would normally
remand for the District Court to reconsider its ruling in light
of its now-enhanced discretion. The government, however,
advances two primary reasons why, in its view, remand is
unnecessary.

       First, the government emphasizes that it “did not
actually select or target any of the defendants,” suggesting
that a selective enforcement claim is categorically



125
   See Threadgill v. Armstrong World Indus., Inc., 

, 1371 (3d Cir. 1991) (“[T]here is no such thing as ‘the
law of the district.’”).




                              62
forestalled.126 This argument was raised in and rejected by
Davis. We agree with the Seventh Circuit that, although
Berry “himself initiated matters by [asking] the informant for
robbery opportunities and then chose his own comrades . . .[,]
it remains possible that the [government] would not have
pursued the investigation had [the crew] been white.”127

        Second, the government argues in essence that the
matter need not be remanded because any error was harmless;
Washington received everything to which he was entitled
when the District Court gave him a redacted portion of an
ATF manual on the eve of trial. The Fourth Circuit took such
an approach in United States v. Hare, decided shortly after
Davis. Despite quoting Davis with approval and exhibiting
some discomfort with the Armstrong/Bass test as applied to
stash-house cases, the Fourth Circuit decided that the
defendants “ha[d] not shown that they are entitled to
discovery beyond what the government has already
produced.”128 While Washington also has not shown that he
is “entitled” to anything beyond what he has already received,
we think that the District Court, not our Court, is better
positioned to make that determination.129



126
      Gov’t Br. at 35 n.10.
127
      

–23.
128
      

, 101 (4th Cir. 2016).
129
    Further, we note that 1) the District Court copied the
approach taken in the Alexander Northern District of Illinois
case, and thus may not have been independently exercising its
discretion; and 2) material relevant to a trial defense does not




                                63
        Accordingly, we will vacate the District Court’s
discovery orders and remand for a renewed decision under the
framework we articulate today. We emphasize that we are
not directing the District Court to grant discovery; our
collective thumbs are not on the scale. Rather, we commit the
inquiry to the District Court’s considerable discretion. We
note that the District Court may, if it so chooses, consider
additional information offered by Washington on remand as
part of his proffer, as well as any relevant information (such
as testimony about the racial cast of prior prosecutions) that
was disclosed at trial.

        Two administrative considerations require additional
attention. First, as indicated by the Supreme Court in
Armstrong itself, discovery requests like Washington’s exist
outside of the framework of Fed. R. Crim. P. 16, and are
neither a challenge to nor a defense against the government’s
actual case.130 It is well established, moreover, that both
discovery orders and substantive equal protection challenges
are appealable only after entry of final judgment.131

necessarily coincide with what is relevant to a challenge to an
indictment on equal protection grounds.
130
      See 

–64.
131
    See Jarkesy v. SEC, 

, 26 (D.C. Cir. 2015);
Adapt of Phila. v. Phila. Hous. Auth., 

, 360 (3d
Cir. 2006); United States v. Howard, 

, 552 (9th
Cir. 1989); cf. United States v. Zone, 

, 1106–07
(9th Cir. 2005) (per curiam) (merging discovery and
substantive inquiry when underlying Double Jeopardy claim
would have been appealable under the collateral order
doctrine).




                               64
Accordingly, by remanding for partial reconsideration of
Washington’s discovery request, we do not unwind his
conviction or otherwise undermine the jury’s verdict. If
discovery is granted, and if it leads to a successful selective
enforcement claim, then his constitutional rights can be
vindicated at that time by striking the indictment in whole or
in part.132 Second, Washington did not file a motion to
dismiss the indictment pursuant to Fed. R. Crim. P. 12(b), as
his gateway discovery request was denied. Despite the
requirement in Rule 12(b)(3) that certain motions be made
“before trial,” we will not require defendants to file quixotic
substantive motions even before their predicate discovery
motions are granted or denied. In any event, we note that as
of the 2014 revision to Fed. R. Crim. P. 12(b)(3), the
language of the rule makes clear that the substantive motion
must be made pretrial only if “the basis for the motion is then
reasonably available.”

III.         CONCLUSION

        For the foregoing reasons, we will affirm the judgment
of the District Court, vacate the discovery orders, and remand
for further proceedings.


132
   See United States v. Jones, 

, 978 & n.8 (6th
Cir. 1998) (remanding for discovery on a selective
prosecution claim, while noting that the remand “does not
warrant a new trial, but only gives [the defendant] the
opportunity to move to dismiss the indictment following
discovery”); cf. United States v. Brizendine, 

,
222 (D.C. Cir. 1981) (explaining how the court can “provide
effective relief” on appeal from final judgment).




                              65
United States v. Washington, 16-2795

McKEE, Circuit Judge, concurring in part and dissenting in
part.

        I agree with the Majority’s thoughtful and persuasive
discussion of the discovery and ineffective counsel issues this
case presents. I therefore join Part II.A and Part II.C of the
Majority Opinion. However, I disagree with my colleagues’
rejection of Washington’s sentencing manipulation claim and
his assertion that the mandatory minimum sentence should not
apply in these unique circumstances. Accordingly, I must
respectfully dissent from Part II.B.

           I.      Stash-House Sting Operations

       Arguably, undercover sting operations, including ones
involving fictitious stash houses, can be a valuable
investigative tactic for ferreting out those individuals who
would otherwise commit crimes in their communities. I also
agree that “[c]ourts should go very slowly before staking out
rules that will deter government agents from the proper
performance of their investigative duties.”1

        However, the potential for abuse and mischief that is
endemic to fictitious stash-house stings should not be ignored.
The U.S. Court of Appeals for the Fourth Circuit has cautioned
that stash-house stings “appear[] highly susceptible to abuse.”2
The Court of Appeals for the Ninth Circuit is “wary of [stash-
house] operations” due to the “the ease with which the
government can manipulate . . . factors [like drug
quantities][.]”3 The Ninth Circuit specifically warned that one
of the problems with such operations is that they ignore
questions about whether a planned stash-house robbery is
within a defendant’s actual “ambition and means.”4 Indeed,

1
  United States v. Montoya, 

, 3 (1st Cir. 1995)
(citation omitted).
2
  United States v. Hare, 

, 103–04 (4th Cir. 2016),
cert. denied, 

(2016), reh’g denied, 

(2016).
3
  United States v. Briggs, 

, 730 (9th Cir. 2010).
4
  

colleagues also express reservations about these operations
here, even though they ultimately conclude that Washington is
not entitled to relief. Moreover, federal courts have noted that
such sting operations risk opening the door to the very kind of
racial profiling Washington is alleging here.5 All of these
problems with stash-house operations have led noted jurist
Richard Posner of the Court of Appeals for the Seventh Circuit
to conclude, on the whole, that such operations are “a
disreputable tactic.”6

       The facts of this case illustrate that these cautions and
misgivings are well-founded. This investigation began with a
confidential informant (“Roc”) advising a supervising ATF
agent that he knew of an individual (Dwight Berry) who
wanted to rob a drug stash house. After Berry was identified,
the ATF embarked upon inventing a scenario that would
include weapons, a “crew,” and a mythical quantity of cocaine
that would be the bait for those who would become ensnared
in ATF’s trap.

5
  In 2013, for example, Chief Judge Ruben Castillo of the
U.S. District Court for the Northern District of Illinois
ordered the disclosure of prosecutorial records after defense
attorneys filed for discovery. The defense attorneys argued
that since 2011, all of the stash-house targets charged in
Chicago’s federal courts had been minorities—19 African-
American and seven Latino defendants. Chief Judge Castillo
ordered discovery “on the sensitive issue of potential racial
profiling” after concluding that “the defendants ha[d] made a
strong showing of potential bias in the history of the
prosecution of . . . ‘phony drug stash house rip off cases.’”
Order, United States v. Brown, No. 12-cr-632, ECF No. 153,
at 1 (N.D. Ill. July 31, 2013). Other district courts also have
ordered discovery into the basis for ATF and federal
prosecutors identifying suspects for investigation. See Maj.
Op. at 54 n.112.
6
  See, e.g., United States v. Kindle, 

, 414 (7th
Cir. 2012) (Posner, J., dissenting) (criticizing that “[l]aw
enforcement uses [such stings] to increase the amount of
drugs that can be attributed to the persons stung, so as to jack
up their sentences”), reh’g en banc granted, opinion vacated
(Jan. 16, 2013), on reh’g en banc sub nom. United States v.
Mayfield, 

(7th Cir. 2014).
                               2
       It was the Government, not Berry, that selected
cocaine—instead of, for example, marijuana—as the drug of
choice for the stash house. Although no cocaine actually
existed, the Government decided to entice targeted individuals
with a predetermined quantity of cocaine—10 kilograms—
which was double the amount needed to statutorily trigger the
mandatory minimum provisions. We are told that this quantity
was necessary in order to portray a “credible” stash house in
the Philadelphia region.

       After initiating a plan to rob a stash house with Roc and
an undercover Agent, Berry presumably enlisted Washington.
Washington was a resident of the community whom the
confidential informant had not initially targeted, and he was
not of any initial interest to the Government based on past
criminal activity.

       I realize, of course, that even though Washington was
just a secondary target, his statements during the planning
meetings and subsequent phone conversations show that he
was neither a shrinking violet nor reluctant recruit. Rather,
Washington was clearly interested in participating and even
offered a number of disturbingly violent ideas that he thought
would facilitate the planned robbery.

       Nevertheless, it comes as no surprise that, “having
yielded to an extraordinary inducement [Washington] would
do everything possible to earn the promised reward.”7
According to the Special Agent’s testimony at trial, a single
kilogram of cocaine was worth upwards of $40,000, as Berry
(the person who enlisted Washington) no doubt knew.

       Despite the Government’s claim that the 10-kilogram
quantity was only selected to make the scheme credible,
nothing suggests that Washington was motivated by any
knowledge of a specific drug quantity, nor is there any
evidence of him having any involvement with stash-house
robberies.8 To the contrary, Washington initially told the

7
 See, e.g., 

(Posner, J., dissenting).
8
 The Agent had only told Berry that he saw over 10
kilograms of cocaine inside a cooler when the two men met.
                               3
group that he did not want to be involved with cocaine. He
explained that he “don’t fuck with coke” and that he didn’t
“really do this shit.” The Agent understood Washington’s
claim that he didn’t “really do this . . .” to mean that
Washington did not deal in home invasion robberies. Yet, the
Agent and Roc forged ahead, greasing the skids to involve
Washington in the criminal conspiracy. Washington was
ultimately arrested and charged with conspiracy to possess, and
attempt to possess, 5 kilograms or more of cocaine with intent
to distribute after he carried out the Government-contrived
crime.

        Despite Washington’s initial statements of disinterest in
cocaine and stash-house robberies, I agree that Washington’s
ultimate actions do establish his intent to carry out an armed
theft of cocaine from a stash house. However, that should not
obscure a more fundamental point. As another appellate court
has explained, “[t]he risk [of targeting] . . . generalized
populations [with stash-house investigations] is that the
government . . . create[s] a criminal enterprise that would not
have come into being but for the temptation of a big payday, a
work of fiction spun out by government agents to persons
vulnerable to such a ploy who would not otherwise have
thought of doing such a robbery.”9

       Here, the Government created a criminal scheme that
would not have otherwise existed. Washington had no prior
history of stash-house robberies (or violent crimes generally,
for that matter), and he expressed reluctance to get involved
with cocaine. Thus, here, as in similar cases, there is a strong
possibility that had Washington not been “fooled into
conspiring and attempting to steal fictitious drugs,”10 he may
well not have been sucked back into the criminal justice
system. This is particularly true because he was not even the
intended target of this operation. Despite his criminal past,

The District Court makes no finding that Berry then told
Washington of the exact quantity of drugs to be obtained in a
potential stash-house robbery before Washington joined the
initial planning meeting.
9
  United States v. Black, 

, 303 (9th Cir. 2013).
10
   United States v. Yuman-Hernandez, 

, 474 (9th
Cir. 2013).
                               4
Washington was not necessarily destined to commit future
crimes. “Criminals do sometimes change and get their lives
back on track,” and, as Judge Posner reminds us, “we don’t
want the government pushing [criminals] back into a life of
crime.” 11,12

         II.    Sentencing in Stash-House Sting Cases

        As is all too often the case, not only do stash-house
stings risk ensnaring those who might otherwise not have
committed crimes, but also the resulting convictions regularly
give rise to particularly dubious applications of the Sentencing
Guidelines and mandatory minimum sentences. Here, as is
typical of these stings, the Government intentionally set the
amount of fictitious drugs at a level that substantially increased
Washington’s sentencing exposure.


11
     

–16 (Posner, J., dissenting).
        As I suggested earlier, given Washington’s statements
during this scheme, he is not the best example of someone
being lured into criminality who may otherwise have
continued restoring his life in the community. Nonetheless,
he still had the support of a family, and at the sentencing
hearing, his loved ones told the court that Washington, after
serving time for his first conviction, was “out doing the right
thing . . . doing really good,” having, for example, acquired
his own business and taking children in the community to
baseball games. Sentencing Tr. 36. His mother stated: “He
was doing a lot of good things and how he got caught up in
that situation is beyond me.” 

during the scheme notwithstanding,
concerns that have been expressed about fictitious stash-
house schemes are no less valid. The tactic still is troubling.
12
   See Alfred Blumstein and Kiminori Nakamura, Redemption
in the Presence of Widespread Criminal Background Checks,
47 Criminology 327, 327–59 (2009) (“Recidivism probability
declines with time ‘clean,’ so some point in time is reached
when a person with a criminal record, who remained free of
further contact with the criminal justice system, is of no
greater risk than a counterpart of the same age [who has no
criminal record] . . . .”).
                                 5
       The potential for mischief and abuse is rewarded and
encouraged by applying an extraordinarily heavy mandatory
sanction that I doubt Congress ever intended to apply where no
drugs exist,13 and where the defendant would not have
committed a crime without the government’s assistance. Here,
the Government decided to charge Washington with a
conspiracy involving 5 kilograms or more of cocaine. As the
majority notes, given that quantity, Washington’s prior
convictions subjected him to a 20-year mandatory minimum
sentence. Accordingly, the District Court concluded that it was
required to impose the 20-year mandatory minimum sentence
that Washington received.

        Surely, sentences should bear some rational relationship
to culpability. Otherwise, the entire enterprise of criminal
sanctions is reduced to little more than an abstract matrix of
numbers and grids. Yet, on this record, there is absolutely
nothing to suggest that Washington would not have conspired
to rob a stash house containing, for example, a kilogram less
than the 5-kilogram mandatory trigger. No mandatory
minimum would have “applied” had this trap been baited with
the illusion of a stash house containing four kilograms
(translating roughly to upwards of $160,000 in value based on
the trial testimony)—thereby placing him beyond the reach of
the perceived need to impose a 20-year statutory mandatory
minimum sentence.14

       It is worth repeating that Washington had no prior
history of robbing stash houses containing any quantity of
cocaine (let alone 10 kilograms of it), or any history of

13
   See infra Part III for a discussion of what, ostensibly, were
Congress’s original intentions for tying mandatory minimums
to specific drug quantities.
14
   I recognize that the 5-kilogram cutoff is equally arbitrary
when defendants are sentenced for a quantity of drugs that
actually exists. Some degree of arbitrariness may be
necessary to any sentencing scheme, and this is no less true
when sentencing ranges are largely determined by artificially
constructed Federal Sentencing Guidelines ranges. However,
that practical reality does not minimize or negate the very real
issues of unfairness and the potential for sentencing
manipulation in these kinds of cases.
                               6
committing violent crimes. In addition, as I have noted, he
initially stated that he did not want to get involved with
cocaine. Even if we accept the deterrent value of mandatory
minimum sentences, it is fanciful to believe that Washington
would not have been deterred from future criminal activity had
a much shorter period of incarceration been imposed. As Judge
Posner has argued in similar circumstances, if a shorter
sentence had been imposed, “[could] there be any serious
concern that upon emerging [from prison, Washington] would
embark on a career of robbing stash houses? That if approached
by anyone [subsequently] inviting him to launch such a career
he would listen to the person?”15 I think not.

        My concern is exacerbated by the fact that very few
nationally-reported cases of government sting operations or
investigations specify any fictional amount of cocaine that is
less than the 5 kilograms that triggers this mandatory minimum
sentence. Other courts have recognized this problem. For
example, the U.S. Court of Appeals for the Second Circuit
noted in another stash-house case:

     It is unsettling that in this type of reverse sting, the
     government has a greater than usual ability to
     influence a defendant’s ultimate Guidelines level
     and sentence. It appears to be no coincidence that
     the [government] chose to [use] no less than [the
     amount of sham cocaine that would trigger as
     much as 78 more months of imprisonment] . . .16




15
   

(Posner, J., dissenting) (criticizing
the fact that the defendant in that stash-house case was
imprisoned for 27 years—and proposing that a sentence of 5
years was “more than adequate,” in part because, as a result
of the sting, “taxpayers w[ould] be supporting [the defendant]
at considerable expense for the next quarter century”).
16
   United States v. Caban, 

, 93 (2d Cir. 1999)
(recognizing the defendant’s argument as one paralleling
sentence manipulation but concluding that the status of the
doctrine at the time was unclear).
                                 7
In fact, it is usually the government’s initial scripting of the
stash-house operations, including the quantity of drugs, that
automatically subjects defendants to particular sentences.17

        It is very troubling that the government can initiate and
facilitate criminal conduct, and make strategic choices that
result in sentences that have a relationship to culpability that
is, at best, tenuous and theoretical. As other courts have
observed, in fictitious stash-house stings, “the government has
virtually unfettered ability to inflate the amount of drugs”
involved—in addition to selecting the type of drugs—“thereby
obtain[ing] a greater sentence for the defendant.”18 The
government can also “minimize the obstacles that a defendant
must overcome to obtain the drugs.”19 Though the District
Court here felt compelled to rely on the fanciful quantity the
Government selected and to impose the corresponding 20-year



17
   It is also the government’s initial scripting of the type of
drugs that bears on mandatory minimum sentencing. When
asked about choosing that drug for the sting operation in this
case, the Government witness described stash-house stings as
a “technique . . . developed in the 1980s in response to a
trend,” and that “[m]any of the robbery crews . . . specifically
target houses where cocaine is stored.” Trial Tr. 82–83.
Therefore, “[the sting operation] has to be realistic” and
“mirror what’s really going on in the streets for them to
believe it and for our safety.” 

The witness
explained that “when you’re talking about the operation of a
stash house, cocaine lends itself . . . as opposed to say another
drug like marijuana where—if you’re talking about a large
scale, typically you’re talking about a grow house or
something like that.” 

infra, however, my concerns about the
degree to which such street-informed testimony can be tested
leave me doubting whether the government must use cocaine
to achieve its law enforcement objectives. Here, for example,
Berry expressed only a general interest in robbing a drug
stash house without regard for a specific type.
18
   United States v. Briggs, 

, 729 (9th Cir. 2010).
19
   

                                8
mandatory minimum, “the Government assured such a result
in advance by the script that it wrote . . . .”20

         My colleagues correctly note that that there was little,
if any, countervailing evidence for the District Court to
consider in making the factual determination that the agents
could have used an amount less than 10 kilograms in creating
the stash house.21 The only relevant findings stem from the
undercover Agent’s trial testimony that the 10-kilogram
amount was selected because that quantity mirrored drug
weights typically found in stash houses in Philadelphia. He
explained that the proposed scenario had to be realistic, lest
robbery crews question the operation’s legitimacy. He also
testified that that quantity was based on a consultation with the
Drug Enforcement Agency (presumably the Philadelphia
Division), which, he claimed, provides “experts in this
information.”22 Apparently, the DEA is “aware of exactly
what was going on . . . in the Philadelphia Metropolitan region”
and provided the quantity “based on search warrants and
investigations that they had conducted.”23

       Another district court considering a stash-house sting
prosecution using 10 kilograms of cocaine was faced with
similar government evidence. However, unlike here, that court
was able to conclude that “the record [there] [wa]s clear that
[the defendant] was ‘in for a penny, in for a pound,’”24 and that
the evidence before it had established that the defendant was
“‘hungry’ enough to pursue . . .[the] undertaking regardless of
any specific amount of drugs.” 25 That district court explained
that “[o]nce the Government established that [the defendant]

20
   United States v. McLean, 

, 939 (E.D.
Pa. 2016).
21
   While it was ultimately the 5 kilograms of cocaine that the
Superseding Indictment charged that drove Washington’s 20-
year mandatory minimum, the amount the Government
selected allowed it to charge Washington with conspiring to
rob 5 kilograms or more, and thereby trigger the mandatory
minimum.
22
   Trial Tr. 85.
23
   

199 F. Supp. 3d at 935.
25
   

                               9
was willing to engage in an armed robbery of any quantity
large enough to resell, its core law enforcement objective was
met.”26 The court cited to the government’s own testimony
that “the street value” of a single kilogram of cocaine was
$36,000 and that stolen narcotics “represent pure profit,” both
factors that would seem to make the sting “sufficiently alluring
well below 5 kilograms.”27

        My agreement with the Majority on this specific issue
notwithstanding, it is nearly impossible for a defendant to ever
rebut the government’s “expert”-based explanation for why a
given fictitious quantity is necessary or appropriate. Accepting
such testimony at face value invites the mischief I mentioned
at the outset to drive the sentencing. The district court is also
deprived of its well-established sentencing discretion,28 a
concern compounded by the problems the district court in
McLean identified:
      The netherworld of criminal activity is by its very
      nature opaque. For that reason, almost out of
      necessity, law enforcement officers, whose
      experiences give them familiarity with that world,
      are allowed to render certain opinions about use
      of coded language and street slang. When used in
      that way, the opinion testimony is interpretive. In
      stash house sting cases, the Government seeks to
      make [that opinion testimony] dispositive
      because the charges themselves are the product of
      opinion testimony as to 1) the amount of cocaine
      that would be “expected” to be found in a stash
      house, and 2) the necessity of specifying
      substantial amounts to preserve the credibility
      and safety of the operation. There is a third
      unstated premise as well—that the targets of the
      sting would have the same familiarity with the

26
   

27
   

28
   See, e.g., Mistretta v. United States, 

, 390
(1989) (discussing the Sentencing Guidelines and Congress’s
“strong feeling that sentencing has been and should remain
primarily a judicial function” (internal quotation marks
omitted)).
                               10
      quantity of narcotics stored at the average stash
      house.

      By definition, such opinions are supported only
      by personal experience, and the dataset, to the
      extent that one exists, is created by, and only
      accessible to, law enforcement. There are no peer-
      reviewed journals within the narcotics trade.
      There is no way to test the premises on which
      these sting operations are based. None of the
      traditional means by which expert testimony can
      be tested in a systematic way apply here, yet
      courts are expected to accept such opinion as the
      justification for undercover operations that
      inexorably and indiscriminately give rise to large
      mandatory minimum sentences. 29

I agree.

        Thus, regardless of whether a claim of sentencing
manipulation is raised, any proffered evidence about the need
for a given quantity or type of fictitious drugs deserves a great
deal more scrutiny than courts give it. 30 Similarly, requiring
evidence that a defendant only agreed to participate because of
a given quantity or type of drugs seems more than appropriate.
Requiring such scrutiny would not eliminate the myriad of
problems that pervade these fictitious stash-house stings, but it
would at least help minimize the unfairness that can arise from
allowing the government to select the drug and the quantity
that will reap the biggest reward at sentencing with little or no
fear that a sentencing court would ever question the choices.31

29
   

30
   Here, the District Court did not probe the testimony, which,
as the Majority notes, it certainly was free to do. Maj. Op. at
40 n.73. As the Majority further suggests, had there been
more fact-finding by the District Court on this issue, some
deference to the testimony about the drug quantity may have
been appropriate. 

n.78.
31
   To accept, wholesale, the unsubstantiated rationale that a
fictitious quantity of drugs matches what is “realistic” in a
particular geographic region also suggests that defendants
across the United States could theoretically be subjected to
                               11
We should not be “delegat[ing] [sentencing discretion] all the
way down to the individual drug agent operating in the field.”32

        Scrutinizing the basis for the drug quantity would help
restore the alignment between culpability and punishment that
is jettisoned when the government is allowed to control the
defendant’s sentencing exposure. “Deeply ingrained in our
legal tradition is the idea that the more purposeful is the
criminal conduct . . . the more severely it ought to be
punished.” 33 Absent unique circumstances not evident here, a
defendant’s criminal exposure should be linked to actual
culpability regarding his/her dealings in specific drug
quantities.

        Insofar as sentencing manipulation is concerned, “[t]he
question is not whether the underlying criminal conviction is
lawful, but rather whether there is reason to reduce the sentence
due to the inducements used by undercover police or their
agents.” 34 Moreover, “a sentence based on an evaluation of a
defendant’s culpability for particular offense conduct, which
includes a consideration of police inducements,” serves the
retributive goals of “proportional and fair punishment,” is
“compatible with the consequentialist aims of incapacitation
and deterrence,”35 and is “directly supported by the systemic
goal of identifying less blameworthy defendants and mitigating
their sentences accordingly.”36 These fundamental principles
of criminal justice necessitate closer scrutiny for schemes that
originate with, and are driven by, law enforcement because it
is highly unlikely that the Sentencing Guidelines were intended
to apply to such circumstances.37 This scrutiny is appropriate

mandatory minimum sentences if the stash-house drug
quantities allowing for such a sentence happen to be
“realistic” for those geographic areas, as they apparently are
in Philadelphia.
32
   United States v. Staufer, 

, 1107 (9th Cir.
1994).
33
   Tison v. Arizona, 

, 156 (1987).
34
   Eda Katharine Tinto, Undercover Policing, Overstated
Culpability, 34 Cardozo L. Rev. 1401, 1454 (2013).
35
   

(emphasis added).
36
   

37
   See infra Part III.
                               12
even absent specific evidence that the government “intended”
to inflate a defendant’s sentence.38

       My colleagues discuss our precedent in United States v
Twigg39 in rejecting Washington’s claim that the sentence that
resulted from this scheme is a denial of his constitutional right
to due process. I would emphasize, however, that Twigg does
not defeat any claim of sentencing manipulation. Indeed, if
anything, Twigg strongly suggests that we should recognize
some kind of sentencing factor manipulation claim when
appropriate. Although, for reasons the Majority explains, the
conduct here may not have crossed the due process threshold,40

38
   See 

n.34 at 1426 (concluding that “in the
context of a sentencing claim, the requirement of an improper
[police] motive ignores the needed link between the police
conduct and the justification for a reduction in sentence”
because “[r]egardless of whether police officers are explicitly
making strategic choices based on sentencing laws (and the
desire to increase a suspect’s sentence), the motivation for the
law enforcement conduct or the inducements used may or
may not be relevant from the perspective of assessing the
defendant’s culpability”).
39
   

(3d Cir. 1978).
40
   I disagree with the Majority’s suggestion that Washington
has not shown prejudice because Washington’s ultimate
sentence was significantly below the recommended
Sentencing Guidelines range. The Majority, itself, concludes
that the District Court was “clearly guided by the mandatory
minimum term on the drug counts in crafting the overall
sentence.” Maj. Op. at 33 n.55. The District Court never
mentioned whether, or the extent to which, it may have
departed from the recommended Sentencing Guidelines range
had it not been required to impose a sentence of at least 20
years.
      Neither do I find persuasive the distinction the
Majority makes between this case and McLean, to the extent
that Washington could rely on that case for whatever
persuasive value it may have for his due process argument.
The Majority, for example, discusses that the defendant in
McLean received a “split” jury verdict on the amount of
cocaine involved (5 kilograms with regard to conspiracy but
                               13
I believe Washington’s sentencing manipulation claim is more
meritorious than the Majority concludes.

 III.    Sentencing Factor Manipulation and Mandatory
                   Minimum Sentences

       The fact that the sentence was mandatory does not
necessarily deal a fatal blow to Washington’s sentence
manipulation claim. It is difficult to believe that Congress ever
considered requiring the imposition of a mandatory minimum
sentence where 1) the sentence is tied to a fictitious drug
quantity in a criminal endeavor that originates with the
government, and 2) the defendant would not have engaged in
the criminal conduct but for the government’s prompting and
encouragement.

       Congress intended for the 10-year mandatory minimum
sentences to apply to “major traffickers,”41 i.e., “manufacturers
or the heads of organizations.” 42 The 5-year mandatory
minimums were intended to apply to “serious traffickers,” i.e.,
“managers of the retail level traffic . . . in substantial street

500 grams with regard to attempt) and that there was “no
equivalent ambiguity” in the jury’s verdict for Washington
here. But that jury finding, while it highlighted the “inherent
problems” these prosecutions presented for the district court,

, was not one of the “factors”
that led the court to conclude that enforcing the mandatory
minimum would “offend due process.” 

Regardless of any “ambiguity,” the jury in McLean still found
the defendant guilty of conspiring to possess 5 kilograms or
more of cocaine which, “absent some constitutional
prohibition,” purportedly “bound” the district court—like
the District Court here—to a mandatory minimum sentence.

41
   U.S. Sentencing Comm’n, Special Report to the Congress:
Cocaine and Federal Sentencing Policy, 119 (1995).
42
   H.R. Rep. No. 99-845, 99th Cong., 2d Sess. 1986, 

; see also 132 Cong. Rec. 27, 193–94 (daily ed.
Sept. 30, 1996); 132 Cong. Rec. 22, 993 (daily ed. Sept. 11,
1986).
42
   H.R. Rep. No. 99-845, 99th Cong., 2d Sess. 1986, 

                               14
quantities.”43 Despite Congress’s intention for mandatory
minimums to reflect culpability based on drug quantities, the
law instead has, over time, targeted low-level offenders (e.g.,
street-level dealers and couriers) more often than high-level
offenders.44 For example, in 2009, offenders sentenced for
relatively minor roles represented the biggest share of federal
drug offenders, while the highest-level traffickers made up a
comparatively small share of federal drug offenders.45 The
disconnect is not explained by the fact that there are more low-
level dealers than high-level traffickers. The U.S. Sentencing
Commission itself concluded in 2011 that “the quantity of
drugs involved in an offense is not as closely related to the
offender’s function in the offense as perhaps Congress
expected.”46

43
   

Sentencing Comm’n, Report to the Congress:
Mandatory Minimum Penalties in the Federal Criminal
Justice System, Appendix D, Figure D-22, available at
https://www.ussc.gov/sites/default/files/pdf/news/congression
al-testimony-and-reports/mandatory-minimum-
penalties/20111031-rtc-pdf/Appendix_D.pdf; see U.S.
Sentencing Comm’n, Special Report to the Congress:
Cocaine and Federal Sentencing Policy, 20–21, 85 (May
2007). See also Deborah Young, Rethinking the
Commission’s Drug Guidelines: Courier Cases where
Quantity Overstates Culpability, 3 Fed. Sent. Rptr. 63 (1990)
(tracking the disproportionate severity of quantity-based
penalties for lower-level drug offenders and further observing
that the quantity-based Sentencing Guidelines often apply to
defendants less culpable than the key drug players, who are
the “primary targets of the laws”).
45
   U.S. Sentencing Comm’n, Report to the Congress:
Mandatory Minimum Penalties in the Federal Criminal
Justice System (October 2011), Appendix D, Figure D-22,
available at
https://www.ussc.gov/sites/default/files/pdf/news/congression
al-testimony-and-reports/mandatory-minimum-
penalties/20111031-rtc-pdf/Appendix_D.pdf.
46
   U.S. Sentencing Comm’n, Report to the Congress:
Mandatory Minimum Penalties in the Federal Criminal
Justice System, 350, (October 2011) available at
https://www.ussc.gov/sites/default/files/pdf/news/congression
                              15
       Thus, there is no reason to believe that Congress
anticipated—much less intended—for quantity-based
mandatory minimums to reflexively apply in stash-house cases
where, as here, the defendant is not only a low-level “drug”
offender, but also became involved with non-existent drugs at
the government’s urging. The circumstances of such phony
stings will rarely lend themselves to a mandatory minimum
sentence, or suggest that Congress intended a mandatory
minimum to apply.         Concluding otherwise risks both
perverting the congressional intent behind the mandatory
minimums and, as I have explained, circumventing federal
judges’ traditional sentencing authority.47

       Moreover, applying mandatory sentences where the
criminal conduct and the type and quantity of drugs exist only
in the law enforcement’s fertile imagination, rather than an
offender’s actual possession, defeats the congressional intent
of requiring judges to impose sentences that are guided by the
factors in 18 U.S.C. § 3553(a). In United States v. Olhovsky,
we stressed that “[18 U.S.C. §] 3553(a) clearly states that a
court must impose a sentence that is ‘sufficient but not greater
than necessary, to comply with the purposes of
[sentencing].’”48 We there quoted the Supreme Court’s
admonition that this requirement, referred to as “the parsimony
provision,” is 18 U.S.C. § 3553(a)’s ‘“overarching
instruction.’”49

al-testimony-and-reports/mandatory-minimum-
penalties/20111031-rtc-pdf/Chapter_12.pdf.
47
   My discussion is limited to sentences imposed as
punishment for involvement in a phony stash-house sting. I
do not intend to suggest that a sentence designed primarily to
incapacitate is necessarily inappropriate. Such sentences may
be necessary for the protection of the community in rare
circumstances. However, phony stash-house stings will
rarely, if ever, present a court with such circumstances, and
when they do, I have every confidence that the district court
will sentence accordingly.
48
   

, 547 (3d Cir. 2009), as amended (May 5,
2009).
49
   

(quoting Kimbrough v. United States, 

, 111 (2008)).
                              16
       Despite our conclusion that 21 U.S.C. § 841(b)’s
mandatory minimum sentence provision does not conflict with
§ 3553(a)’s parsimony provision,50 abandoning the “demand of
parsimony that is the overarching instruction of the
congressionally mandated sentencing factors”51 seems an
unintended result in phony drug stings. There are no drugs that
would otherwise endanger the community, and the criminal
conspiracy probably would never have been hatched but for
law enforcement’s intervention and direction. Congress could
not have intended courts to impose otherwise applicable
mandatory minimum sentences—which we have described as
“draconian”52—where the criminal conduct is the result of the
government’s initiative, rather than a defendant’s. I also find
it hard to believe that Congress would create exceptions to
mandatory minimums that spare actual drug traffickers
exposure to draconian sentences53 while intending those same
harsh sanctions to apply when the government lured a
defendant into being involved with drugs that never even

50
   See, e.g., United States v. Walker, 

, 85 (3d Cir.
2007) (finding that there is no conflict between § 3553 and a
mandatory minimum sentence provision because Ҥ 3553(a)
must be read in conjunction with [] § 3553(e), which prohibits
courts from sentencing a defendant below the statutory
mandatory minimum sentence unless the Government files a
motion permitting such departure”).
51
   

(internal quotation marks
omitted).
52
   See United States v. Williams, 

, 258 (3d Cir.
2002).
53
   In Williams, we addressed one of those exceptions—
Congress’s enactment of the “safety valve” in 18 U.S.C.
§ 3553(f). 

not surprising that Congress did not
include situations such as phony stash-house stings in the
statutory exceptions for applying mandatory minimum
sentences; Congress likely never contemplated that situation.
Williams accurately characterizes the lengths of mandatory
minimums as “draconian,” and exceptions like 18 U.S.C.
§ 3553(f)’s safety valve, and § 3553(e) (granting authority
upon government motion), at minimum, evince Congress’s
intention that the mandatory sentences need not always be
imposed.
                              17
existed.

       In addressing Congress’s intent, I recognize that there is
no ambiguity on the face of the mandatory minimum
sentencing statute. 21 U.S.C. § 841(b)(1) does not distinguish
between roles in a narcotics conspiracy, nor does it require that
drugs actually exist.54 That is not surprising, as it would have
taken something approaching clairvoyance for Congress to
foresee that these severe sentences would extend to situations
where drugs were not actually involved. In any event, it is, of
course, axiomatic that “[w]hen Congress establishes a
minimum sentence for a particular crime, district courts are
required to sentence defendants guilty of that crime to a term
of imprisonment no less than the Congressionally prescribed
minimum, unless an explicit exception to the minimum
sentence applies.”55 But as the U.S. Court of Appeals for the
Eleventh Circuit explained, “[c]onceptually, . . . an adjustment
for sentencing factor manipulation is not a departure” that the
mandatory minimum statute would otherwise forbid.56 This is
because “[w]hen a court filters the manipulation out of the
sentencing calculus before applying a sentencing provision, no
mandatory minimum would arise in the first place.”57

      Ironically, it may well be the lay testimony of Rashida
Clover, Washington’s sister and former caretaker, that best
expresses the arbitrariness of applying the mandatory sentence

54
   As the Government points out, there are only two
circumstances under which a district court can depart
downward from a statutorily authorized mandatory minimum
sentence: the government must file a motion to recognize the
defendant’s “substantial assistance,” or the defendant must
fall within the provisions of the “safety valve” embodied in
18 U.S.C. § 3553(f). See, e.g., United States v. Kellum, 

, 289 (3d Cir. 2004).
55
   United States v. Winebarger, 

, 392 (3d Cir.
2011); see also United States v. Reevey, 

, 113
(3d Cir. 2010) (stating that the “exceptions are the only
authority a district court has to depart below a mandatory
minimum” (quoting 

)).
56
   United States v. Ciszkowski, 

, 1270 (11th Cir.
2007).
57
   

the government initiates the crime and no drugs are
involved. At Washington’s sentencing hearing, she remarked:
         20 years? My brother . . . [has] already spent half
         of his life in jail. . . . That’s not doing anything.
         . . . It’s not rehabilitating him. . . . What he needs
         is education and an opportunity. . . . I understand
         that [the District Court has] guidelines to go by,
         but . . . I can’t imagine that . . . [the] Guideline
         book said . . . to go out and entrap young men
         who are not organized in organized crime and
         sentence people for fake drugs and put their own
         limitations on the amount of the drugs just to
         give them a [minimum] 20 years sentence or
         more. . . . I hardly think whoever created that
         book meant for this to happen. I feel like the
         system is being manipulated by that. And it’s . .
         . embarrassing and it’s hurtful because a lot of
         people are being affected by this. This is not just
         my brother. . . .This is about a lot of people in our
         communities that are affected by this. They
         really are.58

        I agree that applying mandatory minimum sentences in
cases where no drugs exist and the government originates and
perpetuates the criminal activity creates such an unfair and
irrational divergence between culpability and conduct that
Congress could hardly have intended the result.

                          IV.    Conclusion

       This case is the latest illustration of why federal courts
across the country continue to find the government’s reliance
on phony stash-house sting operations disturbing. As I have
explained, these cases raise serious issues of fairness while
destroying the fundamental relationship between culpability
and punishment that is so important to sentencing. The
conduct being sanctioned is the direct result of the
government’s initiative rather than the defendant’s.

      I reiterate that it is exceedingly difficult to conclude that
Congress ever considered that mandatory minimum sentences
58
     Sentencing Tr. 36.
                                  19
would apply here. Nevertheless, it just may be that the ultimate
systematic resolution of this very troublesome approach to
sentencing will have to await clarification by Congress, the
Sentencing Commission,59 or the U.S. Supreme Court.
Meanwhile, it is worth echoing my colleagues’ caution: The
Government’s success today should not be interpreted as a clue
that “all such prosecutions will share the same fate” in the
future.60

        Hopefully, this problem will be resolved by one of the
authorities I have just mentioned. Until that day comes, we are
left with the very poignant observation of Ms. Clover, who has
experienced our sentencing laws “up close and personal.” As
quoted earlier, she was skeptical that “whoever created that
[Sentencing Guidelines] book meant for this to happen,” and


59
   The Sentencing Commission has already “recognized the
potential for government agents to use their knowledge of the
Sentencing Guidelines to manipulate the quantity of drugs
sold in a reverse sting in order to increase a defendant’s
sentence.” United States v. Stavig, 

, 1245–46
(8th Cir. 1996) (discussing how under Application Note 17 of
U.S.S.G. § 2D1.1(b)(17), a district court can depart
downward when law enforcement agents set a price below
market that allows the defendant to purchase a significantly
larger quantity of drugs, and that Application Note 12 of
§ 2D1.1 instructs a district court to remove from the
sentencing calculation the amount that a defendant is unable
to produce if the produced amount is less than negotiated).
The provisions of the Sentencing Guidelines in place “show[ ]
that the Sentencing Commission is aware of the unfairness
and arbitrariness of allowing drug enforcement agents to put
unwarranted pressure on a defendant in order to increase his
or her sentence without regard for his predisposition, his
capacity to commit the crime on his own, and the extent of his
culpability.” United States v. Staufer, 

, 1107
(9th Cir. 1994). But the “Sentencing Commission’s
determination that the defendant may receive a downward
departure when the government artificially lowers the price of
the drugs . . . only addresses one of the ways in which drug
enforcement agents are able to manipulate sentences.” 

Op. at 42.
                              20
that “the system is being manipulated by that.”61 She added
that it is “embarrassing and it’s hurtful because a lot of people
are being affected by this.”62 And so they are.




61
     Sentencing Tr. 36.
62