In the
United States Court of Appeals
For the Seventh Circuit
____________________
Nos. 16-1209 & 16-1325
UNITED STATES OF AMERICA,
Plaintiff-Appellee,
v.
ANTONIO WALTER and KENNETH BELL,
Defendants-Appellants.
____________________
Appeals from the United States District Court for the
Northern District of Illinois, Eastern Division.
No. 12 CR 516 — Elaine E. Bucklo, Judge.
____________________
ARGUED MAY 25, 2017 — DECIDED AUGUST 29, 2017
____________________
Before WOOD, Chief Judge, and BAUER and HAMILTON, Cir-
cuit Judges.
WOOD, Chief Judge. The Supreme Court’s decision in Brady
v. Maryland, 373 U.S. 83 (1963), might seem a bit strange to
someone who thought that the adversary system in criminal
cases allows each side to adopt a “no holds barred” litigation
stance. But that is not the way the Constitution structures
criminal procedure. From the Fifth Amendment’s privilege
against compulsory self-incrimination, to its double-jeopardy
2 Nos. 16-1209 & 16-1325
clause, to the rights conferred by the Sixth Amendment both
to be informed of the nature and cause of the accusation and
to be confronted with witnesses, to the due process right to be
convicted only upon proof beyond a reasonable doubt, the
system is replete with safeguards for an accused.
The obligation of the prosecution to turn over any favora-
ble evidence to the defendant, first announced in Brady, is one
aspect of the due process right. Appellants Antonio Walter
and Kenneth Bell assert in this appeal that the prosecution
failed to live up to its Brady obligations. They also argue that
the district court abused its discretion under Federal Rule of
Evidence 404(b) by admitting, over objection, evidence of
Bell’s prior drug sales. Even granting that there may have
been a Rule 404(b) error, it is a close question whether it was
harmless, especially for Walter. In the end, however, this does
not matter, because we conclude that the Brady error requires
a new trial for both defendants.
I
In November 2010, federal and state authorities arrested
over 100 people as part of a two-year investigation known as
“Operation Blue Knight.” Bell was one of those arrested, and
on November 15, 2010, he was charged in federal court with
one count of heroin distribution. See 21 U.S.C. § 841(a)(1). As
that case progressed, a different investigation led in July 2012
to indictments of Bell and Walter for conspiring from 2007
through November 2010 to sell over 1,000 grams of heroin in
violation of 21 U.S.C. § 846. Bell wound up pleading guilty to
the 2010 charge after his motion to consolidate the two cases
was denied. After a six-day trial on the 2012 charges, a jury
Nos. 16-1209 & 16-1325 3
convicted both Walter and Bell on October 24, 2013. The dis-
trict court sentenced Walter to 335 months and Bell to 276
months in January 2016.
The government’s theory was that Walter and Bell were
members of the same drug trafficking organization, which ran
“drug spots” (that is, street corners where heroin was sold) on
Chicago’s West Side. Bell supplied heroin for three drug spots
located at Kedzie Avenue & Ohio Street, Chicago Avenue &
Christiana Street, and St. Louis Avenue & Ohio Street. Walter
was a supervisor who oversaw drug spots. The business
model was straightforward: organization members mixed
Bell’s heroin with Dormin (a sleeping pill) and divided it into
user quantities (“blows”), a process called “going to the ta-
ble.” Runners then took packs of blows to the drug spots,
where other members sold them to users.
The government’s case was not ironclad. It rested on evi-
dence that Bell was inexplicably wealthy (e.g., flush with cash
and cars, able to take extravagant trips), physical samples of
heroin seized from organization members, and expert testi-
mony about drug trafficking. Conspicuously absent was any
direct evidence tying either Walter or Bell to the alleged con-
spiracy. Because there were no controlled buys or recorded
incriminating statements, the government’s case hinged on
witness testimony. Seven witnesses identified Walter as a par-
ticipant in the drug organization; five of them fingered Bell as
the drug supplier. The problem was that these witnesses, un-
surprisingly for this type of case, were hardly model citizens.
Their knowledge of the defendants’ involvement in the organ-
ization stemmed from their own participation. All seven had
been charged with or convicted of drug crimes, and six of
them were testifying pursuant to agreements that held out the
4 Nos. 16-1209 & 16-1325
possibility of reduced sentences. As the defense was at pains
to point out, personal involvement, lengthy criminal histories,
and a desire to secure lenient treatment all raised major cred-
ibility concerns.
Two aspects of the trial concern us. First, Bell’s lawyer
called to the stand two officers involved with Operation Blue
Knight and elicited detailed testimony about its thorough-
ness. Information came out that three of the government’s wit-
nesses had been arrested during the course of that investiga-
tion. Counsel took care, however, to avoid revealing that Op-
eration Blue Knight had uncovered two instances in which
Bell was selling heroin in controlled buys, and that he eventu-
ally had been arrested for doing so. These omissions might
have given the jury the inaccurate impression that two inves-
tigations in a row had failed to yield any concrete evidence of
Bell’s wrongdoing. Such an inference would have reinforced
the defense’s central argument—that the government’s wit-
nesses were lying to save themselves. The prosecution sought
to rebut that theory on cross-examination by eliciting testi-
mony from FBI Task Force Officer Michael Lipsey that Oper-
ation Blue Knight had produced a recording of Bell selling
heroin to a confidential informant. The defense’s objection to
that line of questioning was overruled, though the district
court did issue a limiting instruction cautioning the jury to
consider the evidence “only as a rebuttal to evidence pre-
sented by defendant Bell about the investigation conducted
by law enforcement in this case.”
The second issue relates to the government’s failure to
disclose a damaging remark by one of its witnesses,
Dushae Nesbitt, about a key government witness,
Edmund Forrest. Forrest was a career criminal and relatively
Nos. 16-1209 & 16-1325 5
senior organization member who had known Walter “all [his]
life,” and Bell for “a lot of years.” He testified on the first day
of trial pursuant to a plea agreement under which the
government agreed to recommend 15 years off his 30-years-
to-life guideline range, and which allowed for a free fall to a
floor of 10 years. Even so, Forrest’s initial testimony offered
only lukewarm support for the government’s case. He stated
that he had seen Walter in the room while heroin was being
prepared for sale, but he did not remember Walter’s actively
participating. He maintained that Bell was rarely present, and
he did not recall how often Bell supplied heroin. Forrest’s
memory improved after he spoke with the prosecution during
a lunch break. In the afternoon he testified that Bell dropped
off heroin to Walter around three times a month, and that cash
from the day’s sales was usually handed over to Walter.
Nesbitt testified on the second-to-last day of trial. He did
so involuntarily—he was the only witness who did not have
an agreement with the government—and invoked his Fifth
Amendment rights until the government obtained an immun-
ity order. Nesbitt was less taciturn outside of the courtroom.
The critical event for the defendants’ Brady claims came when,
on October 22, 2013, while the trial was still ongoing, Nesbitt
was out in the hallway talking to FBI Agent Helen Dunn. He
told Dunn that Forrest, while out on bond and cooperating
with the government, was still “at the table” and was still sell-
ing narcotics for a supplier known as “KMART.” This directly
contradicted Forrest’s earlier testimony that he had given up
drug sales while on bond. Dunn informed the prosecution of
the exchange either that evening or the following day (the last
day of trial). The government said nothing to the defense
about Nesbitt’s revelations.
6 Nos. 16-1209 & 16-1325
Indeed, the defense heard nothing about Nesbitt’s conver-
sation with Dunn until almost two months later, long after the
jury had returned its verdict. On December 18, 2013, the gov-
ernment sent defense counsel a letter in which it had this to
say about the hallway encounter:
On or about October 22, 2013, FBI Agent Helen
Dunn had a conversation with DuShae Nesbitt
outside of Judge Bucklo’s courtroom in the
Dirksen Building. Nesbitt told Agent Dunn that
Edmund Forrest is still selling drugs and is “at
the table” on a regular basis. According to
Nesbitt, Forrest is selling drugs for FNU LNU
a/k/a “KMART,” who is a cousin of Steven
Collins a/k/a “Cupcake.” Nesbitt said that
Forrest is using Collins’ apartment near Chicago
Avenue and Spaulding Street in Chicago.
The December 18 letter noted that a follow-up interview had
not yielded any more information.
The government sent a second letter to the defense on
February 10, 2014. That letter relayed the results of an
interview of Nesbitt after he pleaded guilty to some state drug
charges:
Nesbitt said that Forrest was “at the table”
mixing heroin for street distribution from at
least June 2013 until September 2013. Nesbitt
knows this because Nesbitt saw Forrest “at the
table” two or three times during that time
period. They were mixing heroin for the
Chicago/Christiana drug spot, which was
selling about $3,000–$4,000 of heroin per day.
Nos. 16-1209 & 16-1325 7
The apartment they used to mix the heroin was
on Cicero near Madison. Nesbitt was running
bundles and “working packs” at
Chicago/Christiana, which led to the two state
cases that Nesbitt recently pled guilty to.
Nesbitt said he also gave Forrest money from
heroin sales on occasion during that period and
saw others give Forrest money from heroin
sales. Nesbitt said that FNU LNU aka
“KMART” was running the Chicago/Christiana
drug spot.
Based on conversations with Forrest, Nesbitt
believes that Forrest may have been “at the
table” for Chicago/Christiana prior to June 2013
as well.
***
As you know, Forrest was on bond for his
federal case during 2013. He is currently in cus-
tody.
A final letter dated July 8, 2014, confirmed that Dunn’s hall-
way conversation with Nesbitt had occurred before the con-
clusion of the trial.
The jury returned its verdict of conviction on
October 24, 2013. Sentencing, however, was delayed
repeatedly: ultimately Bell was sentenced on January 20, 2016,
and final judgment in his case was entered on February 4,
2016; and Walter was sentenced on January 21, 2016, with
final judgment entered on February 2, 2016. After the jury’s
verdict but long before final judgment, on July 19, 2014, Bell
and Walter moved for a new trial. That motion argued that
8 Nos. 16-1209 & 16-1325
their Brady rights were violated by the government’s failure
promptly to pass along what Dunn learned from Nesbitt
during the October 22, 2013, hallway conversation. Two
months later, well before the district court ruled on the
motion, this court decided United States v. Gomez, 763 F.3d 845
(7th Cir. 2014) (en banc). The new framework adopted in
Gomez convinced Bell and Walter that the district court’s
decision to admit the evidence of Bell’s prior drug sales also
required reversal. They amended their motion for new trial to
add that argument.
Rule 33 of the Federal Rules of Criminal Procedure nor-
mally requires motions for new trial to be filed within 14 days
of the jury’s verdict, unless the motion is based on newly dis-
covered evidence, in which case it may be filed within three
years of verdict. Walter and Bell did not know of the Nesbitt
statements until well after the 14-day mark, and the govern-
ment does not appear to have stood on any objection to the
timeliness of their motion. These deadlines are not, in any
event, jurisdictional. See Eberhart v. United States, 546 U.S. 12,
19 (2005). With respect to the Rule 404(b) point, the govern-
ment tried to have it both ways: it filed a response with the
following comment: “This motion is not based on newly dis-
covered evidence. As a result, it is untimely and should be
denied. That said, the government respectfully requests that
the Court address the merits of the motion as well and deny
it on that basis as well.” In light of the government’s final
statement, the district court chose to decide the Rule 404(b)
argument on its merits, as it was entitled to do. We thus place
no weight on the timing of either aspect of the new-trial mo-
tion.
In separate orders issued the same day, December 22, 2014,
the district court denied both aspects of the amended motion
Nos. 16-1209 & 16-1325 9
for new trial. With respect to the Rule 404(b) argument, the
court reasoned that Bell had opened the door to the testimony
about his earlier drug sales by mischaracterizing Operation
Blue Knight. The court rejected the Brady argument on the
ground that neither defendant could demonstrate any preju-
dice from the government’s failure to disclose Nesbitt’s state-
ments to Dunn, even granting that they were favorable to the
defense as impeachment evidence and they were not dis-
closed in time for use at the trial. Forrest’s statement, the court
thought, was corroborated by other sources, which lessened
its importance. Similarly, Nesbitt’s remark would have done
little further damage to Forrest’s credibility, because Forrest’s
credibility was already in tatters—defense counsel had exco-
riated him on cross-examination, highlighting his extensive
criminal history, his supposed motive to lie to preserve his
deal, and the inconsistencies within his trial testimony. Both
defendants have now appealed.
II
We begin with the claim, raised by Bell and joined by Wal-
ter (even though there was no comparable evidence against
him), that the district court erred by allowing Officer Lipsey
to testify about Bell’s sale of drugs to an informant. “We re-
view the district court’s decision to admit evidence under
[Federal] Rule [of Evidence] 404(b) for abuse of discretion
only.” United States v. Curtis, 781 F.3d 904, 907 (7th Cir. 2015).
Even if the court’s decision to admit the evidence was mis-
taken, however, “evidentiary errors are subject to harmless er-
ror review.” United States v. Dvorkin, 799 F.3d 867, 884 (7th Cir.
2015) (alteration and citation omitted). “To determine
whether an evidentiary error is harmless, we consider
whether, to the average juror, the prosecution’s case would
10 Nos. 16-1209 & 16-1325
have been significantly less persuasive absent the error.” Id.
(citation omitted).
Rule 404(b) bars otherwise relevant evidence “of other
crimes, wrongs, or acts if the purpose is to show a person’s
propensity to behave in a certain way.” Gomez, 763 F.3d at 855.
But it allows the use of other-act evidence for “another pur-
pose,” a category that includes (but is not limited to) “proving
motive, opportunity, intent, preparation, plan, knowledge,
identity, absence of mistake, or lack of accident.” FED. R. EVID.
404(b). Though Rule 404(b)’s framework is superficially
straightforward, “confusion arises because admissibility is
keyed to the purpose for which the evidence is offered, and
other-act evidence is usually capable of being used for multi-
ple purposes, one of which is propensity.” Gomez, 763 F.3d at
855.
Gomez is our most recent word on how to distinguish be-
tween propensity and non-propensity purposes. There we
stressed that Rule 404(b) “allows the use of other-act evidence
only when its admission is supported by some propensity-
free chain of reasoning.” Id. at 856. The question is not
“whether the proposed other-act evidence is relevant to a non-
propensity purpose but how exactly the evidence is relevant
to that purpose—or more specifically, how the evidence is rel-
evant without relying on a propensity inference.” Id. Bell’s
theory is that Officer Lipsey’s testimony does not pass that test
because his past heroin sale is relevant only to show a propen-
sity for drug dealing.
Bell has a point. The government argues that it wanted to
use the evidence of the prior sales to rebut the inference Bell
was trying to raise that he had a clean record, by misleadingly
implying that not only did he elude the 2012 investigation, but
Nos. 16-1209 & 16-1325 11
also that he was never caught red-handed by Operation Blue
Knight. This, it says, is a non-propensity purpose that justifies
admitting the challenged testimony. As it sees things, the tes-
timony of the prior sales would correct Bell’s mischaracteriza-
tion of Operation Blue Knight. We do not dispute the notion
that correcting a misleading suggestion or implication can be
a valid non-propensity reason for admitting other-act evi-
dence—Rule 404(b)’s list is illustrative, not exhaustive. See
e.g., Curtis, 781 F.3d at 909–11. But we are skeptical that this is
the use to which the government intended to put the Opera-
tion Blue Knight evidence. The only reason to correct the rec-
ord was to show that Bell, contrary to his insinuation, had in
fact been caught selling heroin. And the only reason that con-
clusion mattered was to invite the jury to infer that he was
likely doing so again in the charged conspiracy.
The district court tried to address this problem with its
limiting instruction following Officer Lipsey’s testimony. The
instruction was broadly in line with Gomez, see 763 F.3d at
860–61, but it was incomplete. It left out the critical point that
the Operation Blue Knight purchases could not be considered
for propensity purposes. The government argues that, even if
it was flawed, the instruction resolved any problem because
its contents resembled the instruction the defense requested,
and Bell’s lawyer said “if we give [the jury] this instruction,
the matter is cured.” But Bell’s lawyer was talking about his
own proposed instruction, which (unlike the one the court
gave) spoke to the propensity issue by forbidding the jury to
consider the controlled buy “as evidence of either defendant’s
guilt in this case.”
A finding that the rule was violated would normally re-
quire us to take the next step and decide whether the error
12 Nos. 16-1209 & 16-1325
was harmless. See FED. R. CRIM. P. 52(a). We regard that as a
close question. It is one that we do not need to resolve, how-
ever, because we conclude that the prosecution’s non-disclo-
sure violated Brady. We trust that in any further proceedings,
the prosecution will take care not to fall into the same
Rule 404(b) trap again.
III
To succeed on a Brady claim, a defendant “bears the bur-
den of proving that the evidence is (1) favorable, (2) sup-
pressed, and (3) material to the defense.” United States v.
Walker, 746 F.3d 300, 306 (7th Cir. 2014) (internal quotation
marks and citation omitted). Both sides agree that Nesbitt’s
hallway disclosures were favorable to the defense, and so we
are left with the question whether the government sup-
pressed anything and if the suppressed evidence was mate-
rial. There is no doubt that the prosecution did not turn over
the evidence in time for use at the trial, but its failure to dis-
close would not have violated Brady if the defense already had
the evidence. See United States v. Morales, 746 F.3d 310, 316
(7th Cir. 2014). Evidence is material under Brady “if there is a
reasonable probability that, had [it] been disclosed to the de-
fense, the result of the proceeding would have been differ-
ent.” Kyles v. Whitley, 514 U.S. 419, 433–34 (1995) (quotation
marks omitted). We review the district court’s denial of a mo-
tion for a new trial based on an alleged Brady violation for an
abuse of discretion. Walker, 746 F.3d at 306.
The government contends that the record reflects that Wal-
ter and Bell did already know the key fact that Nesbitt re-
vealed: that Forrest had lied when he said that he was no
longer dealing drugs while he was cooperating with the gov-
ernment. But the evidence to which it points is fatally vague.
Nos. 16-1209 & 16-1325 13
In arguing for a continuance, Bell’s lawyer said only, “I believe
there may be evidence indicating that there is collusion be-
tween” Nesbitt and Forrest. While cross-examining Nesbitt,
Walter’s lawyer asked if the heroin he had recently been ar-
rested for selling “[w]as … [Forrest]’s heroin?” The lawyer
then added “[b]ut Edmund Forrest is still out on Chicago and
Christiana, isn’t he?” to which Nesbitt replied, “Yes.”
It is pushing too hard to say that a reference to “collusion,”
which is all we have in the first example, is enough to show
that Forrest was out on the street dealing drugs even as he
was cooperating with the government and telling it that he
was doing no such thing. The other two examples to which
the government points—asking if the heroin was Forrest’s and
if Forrest was “out on the street”—would be opaque at best to
the jury. Asking Forrest if he still frequented an area where
heroin was sold is not the same as asking if Forrest was still
involved with selling the drug. Forrest could be “out on Chi-
cago and Christiana” as a customer, or simply because he
lived in the vicinity. Nesbitt’s account to Dunn, set out above,
was in an entirely different league. He unambiguously re-
ported that Forrest was still actively in the drug trade. On top
of that, the statements to which the government points are
equivocal, qualified by such language as “I believe” and “there
may be evidence” (emphasis added). Nesbitt’s statements to
Dunn, in contrast, were unqualified.
The fact that Forrest is an admitted lifelong drug dealer
with at least two controlled-substance convictions (to say
nothing of his firearm convictions) does not undermine the
incremental value of this impeachment evidence. (As we said
earlier, most of the witnesses in this trial had similar flaws.) It
answered the important question whether Forrest sold heroin
14 Nos. 16-1209 & 16-1325
while he was a cooperating witness. Nothing else in the record
directly spoke to that crucial point.
And that was not all that Nesbitt’s hallway remarks added.
The government largely ignores the fact that Nesbitt’s com-
ment (if believed) revealed that Forrest was selling on behalf
of a new supplier named “KMART.” If the jury thought that
KMART had taken over this area, and if it further believed
that there was room for only one supplier, it might have found
that Bell was not that person. Nesbitt and Forrest had been
selling on those corners before Bell’s arrival, and they contin-
ued to sell after Bell’s arrest. That is far from conclusive evi-
dence pointing to Bell’s non-involvement, but it could have
been one piece of a puzzle.
That leaves materiality. The government attacks the mate-
riality of Nesbitt’s statements on two grounds: first, that the
remaining evidence supporting its case was so strong that the
additional information he was providing about Forrest could
not have made any difference; and second, that the use of
Nesbitt’s information to impeach Forrest’s credibility would
have made no difference, because his credibility was already
so damaged. We do not need to find, however, that “but for”
the failure to disclose Nesbitt’s impeachment evidence, the
defendants would not have been convicted. The standard is
only whether there is a reasonable probability of a different
outcome. We conclude that the evidence meets this standard.
The prosecution presented no direct evidence of either de-
fendant’s involvement in the charged conspiracy—the case
came down to the credibility of the government’s witnesses,
and they were questionable at best. Apart from the fact that
the witnesses were involved in the same drug trafficking or-
ganization as Walter and Bell, they were hardly a trustworthy
Nos. 16-1209 & 16-1325 15
bunch. Jeffrey Scott, for example, was testifying to secure a
reduced sentence in connection with his own involvement in
the conspiracy. He had five felony drug convictions; he had
violated the terms of his bond by (among other things) using
marijuana; and he was known as “Scotty Toohigh.” Other wit-
nesses had comparable backgrounds and motivations.
Against that backdrop, Forrest’s testimony was important be-
cause of its detailed, firsthand nature, and because it corrob-
orated what the other witnesses were saying.
Had the defense been able to impeach Forrest with
Nesbitt’s information, Forrest’s reaction could have done
wonders for the defense. Nesbitt’s statements, if believed,
showed that Forrest was actively disregarding his
cooperation agreement, his bond, and the law. Had this been
raised with him on cross-examination, Forrest might have
invoked the Fifth Amendment and refused to testify in order
to avoid either perjuring himself or opening himself up to yet
another drug conviction. At the least, Nesbitt’s statement
would have dented Forrest’s assertion that he had been on the
straight and narrow since he began cooperating with the
government. That matters. Such a claim would enhance the
witness’s credibility with the jury, if one thinks that jurors are
more likely to trust a reformed criminal than an active one.
Yet the implication is flipped if the redemption story turns out
to be a lie. United States v. Boyd, 55 F.3d 239, 246 (7th Cir. 1995).
If the jury in this case had learned of Forrest’s ongoing
criminal conduct, “it might have doubted [his] testimony that
[he] had ‘seen the light.’ Knowing that [he was] lying under
oath about [his] using and dealing in drugs, the jury might
reasonably have supposed that [he was] lying about the
criminal activities of the defendants as well.” Id.
16 Nos. 16-1209 & 16-1325
The defense could even have benefited if Forrest had vig-
orously denied Nesbitt’s accusation. If the jury disbelieved
Forrest, his credibility would have been shot. If they believed
him, Nesbitt’s credibility would have been badly damaged.
And Nesbitt, it bears repeating, was the only organization
witness who was not testifying pursuant to an agreement
with the government. Even if the jury could not decide whom
to believe, the result would have been a more skeptical recep-
tion for both Forrest and Nesbitt.
This is not to say that the government could not have con-
victed the defendants if the defense had been told of Nesbitt’s
statement; that outcome was certainly possible. But the stand-
ard, once again, is only a “reasonable probability” that disclo-
sure would have changed the result of the proceeding. Kyles,
514 U.S. at 433–34; Turner v. United States, 137 S. Ct. 1885, 1893
(2017). That bar has been met. We have no need to consider
whether Walter and Bell’s alternative theory—that they could
have used Nesbitt’s reference to “KMART” to suggest that
Bell was not the supplier during the relevant period—is plau-
sible enough to support relief.
IV
Because Walter and Bell’s rights under Brady were violated
by the government’s failure to disclose material impeachment
evidence to them in time for use at the trial, we VACATE both
convictions and REMAND for a new trial.