William Johnson v. GERARDO ACEVEDO, Warden, Hill Correctional Center

In the
United States Court of Appeals
For the Seventh Circuit
No. 08-1731
WILLIAM JOHNSON,
Petitioner-Appellee,
v.
GERARDO ACEVEDO, Warden,
Hill Correctional Center,
Respondent-Appellant.
Appeal from the United States District Court
for the Northern District of Illinois, Eastern Division.
No. 06 C 5351—Harry D. Leinenweber, Judge.
ARGUED SEPTEMBER 3, 2008—DECIDED JULY 14, 2009
Before EASTERBROOK, Chief Judge, and CUDAHY and
SYKES, Circuit Judges.
EASTERBROOK, Chief Judge. William Johnson is serving
sentences aggregating 50 years’ imprisonment for armed
robbery, aggravated battery, and use of a weapon by a
convicted felon. A jury in Illinois convicted him of these
offenses following a trial at which one of his cousins,
plus two employees of a McDonald’s restaurant, testified
that he entered the restaurant with a sawed-off shotgun

-- 1 of 18 --

2 No. 08-1731
and robbed the cash registers, shooting and injuring one
of the employees in the process. Johnson testified at trial
that he had been buying auto supplies while his cousin
robbed the restaurant. A federal district court issued a
writ of habeas corpus under 28 U.S.C. §2254 after conclud-
ing that the prosecutor had violated the rule of Doyle v.
Ohio, 426 U.S. 610 (1976), by asking Johnson why he
failed to relate this story when interviewed the day after
the robbery. 544 F. Supp. 2d 683 (N.D. Ill. 2008). The
state contends that Anderson v. Charles, 447 U.S. 404 (1980),
permits the prosecutor to pursue this line of questioning,
and that any error is harmless.
A jurisdictional problem delayed consideration of this
appeal. The district court entered a judgment that reads,
in full: “IT IS HEREBY ORDERED AND ADJUDGED
that the petitioner, William Johnson’s petition for a Writ
of Habeas Corpus is conditionally granted.” This sounds
like a writ will issue in the future if some condition is
satisfied—but what condition? The district judge’s
opinion says that the state must release Johnson unless
he is retried within 90 days. That is not a “conditional
grant” of anything; it is a decision that Johnson is
entitled to a writ that allows the state to choose between
retrial and release. But no such writ was issued. A judg-
ment needs to do more than just say that some petition or
motion has been granted; it must provide the relief to
which the victor is entitled. Until that happens there is
no final decision and nothing to appeal.
Every judgment must be self-contained and specify
the relief being awarded. See, e.g., Reytblatt v. Denton, 812

-- 2 of 18 --

No. 08-1731 3
F.2d 1042 (7th Cir. 1987). A writ of habeas corpus is
enforceable by contempt, so it is especially important
that the court set out precisely what is required. District
courts would cause fewer problems of this kind if they
obeyed Fed. R. Civ. P. 58(b)(2), which says that the
judge must personally review and approve any judg-
ment other than one implementing a jury’s verdict or
denying all relief. This judgment was not approved by
the judge; it is signed only by a deputy clerk (and not
really “signed” even by the clerk; there is just a line with
“/s/” followed by a typed name). Rule 58(b) requires
review and approval by a judge because deputy clerks
cannot read judges’ minds and may not use legally ap-
propriate language even if they can discern a judge’s
objective. (Few deputy clerks are lawyers.) This court
put the appeal in stasis while the parties returned to the
district court and obtained a proper final judgment.
That has been done, so the appeal can proceed.
Johnson’s cousin, Jameel White, testified that he drove
with Johnson to a McDonald’s restaurant, which the two
entered to order a meal. (We have drawn this informa-
tion, and much of what follows, from the opinion of the
Appellate Court of Illinois. No one contests that court’s
narration of what happened at trial.) According to
White, Johnson unexpectedly pulled out a sawed-off
shotgun and began a robbery; White fled across the
street to a Trak Auto store, where Johnson eventually
joined him. The two paid a third party to give them a
ride. White went home, and Johnson went to a motel.
Two employees of the restaurant picked Johnson out of
a lineup as the robber and identified him at trial. An

-- 3 of 18 --

4 No. 08-1731
employee of Trak Auto testified that Johnson entered
the store and waved a fistful of crumpled bills. Police
found in Johnson’s car, which had been left in the restau-
rant’s parking lot, some crumpled currency and a sawed-
off shotgun; a spent shell was in the gun’s chamber.
(The robber had fired one round inside the restaurant.)
A torn portion of a $20 bill in the car matched the remain-
der of that bill eventually recovered from Johnson.
Johnson offered a completely different version of
events. He testified that White drove him to the Trak Auto
store so that he could buy auto supplies. (The car was
Johnson’s but White was the driver, Johnson testified,
because Johnson’s driver’s license had been revoked.)
White dropped him off and continued to the McDonald’s
restaurant. White met Johnson later, looking jittery,
and said that Johnson’s car had been involved in some
misconduct and should be abandoned. White then ar-
ranged for a third party to drive them away, taking
White home and Johnson to a motel.
After finding the shotgun and some of the loot in John-
son’s car, police naturally wanted to talk with him.
They tracked him down at the motel and gave him
Miranda warnings; an Assistant State’s Attorney asked
him about the events. The prosecutor tried to offer the
resulting statement at Johnson’s trial, but the judge ex-
cluded it on state-law grounds. (The prosecutor had not
turned over a copy of the statement during discovery, as
Illinois law requires.) A prosecutor made this offer of
proof:
A.S.A. Keating would state that [Johnson] told him
that he woke up at approximately 6:00 a.m. on

-- 4 of 18 --

No. 08-1731 5
11/11/98, went for some car parts, had to take
Theresa [his sister] to work. He was planning on
going to Robbins[, Illinois,] that night for some
peace talk for some shooting earlier. He thought
he would die that night, so he drank some wine,
smoked some crack and weed. He went to a
friend’s house where he had sex in a bathroom,
and he went to his sister’s after that, and he re-
members being in the hotel when the police
came to get him.
This account of his activities on the day of the robbery
did not include any information about White driving
Johnson to Trak Auto and the other events to which
Johnson had testified. The prosecutor asked Johnson a
total of 25 questions that were variations on the theme:
“If what you have just testified is true, why didn’t you
tell the Assistant State’s Attorney when you made
your statement to him?” The state judge sustained objec-
tions to 10 of the questions, so there was no error in that
respect, see Greer v. Miller, 483 U.S. 756 (1987), but
he allowed the other 15, and the federal district judge
thought this an egregious violation of Johnson’s rights.
Doyle holds that a defendant who receives Miranda
warnings, and invokes his right to keep silent, cannot
be cross-examined about that silence at trial. The Court’s
rationale is that Miranda warnings should not become
a trap. Questions of the “why didn’t you say this earlier?”
variety ask the jury to infer that an innocent person
would have spoken, but Miranda warnings supply an
explanation other than guilt for a suspect’s silence. See
Brecht v. Abrahamson, 507 U.S. 619, 628–29 (1993). A corol-

-- 5 of 18 --

6 No. 08-1731
lary is that, if the suspect chooses to speak after re-
ceiving Miranda warnings, then any difference between
what the person says before trial, and testimony at trial,
may be the subject of questions and comment. That’s
the holding of Anderson v. Charles:
Doyle does not apply to cross-examination that
merely inquires into prior inconsistent state-
ments. Such questioning makes no unfair use of
silence, because a defendant who voluntarily
speaks after receiving Miranda warnings has not
been induced to remain silent. As to the subject
matter of his statements, the defendant has not
remained silent at all.
447 U.S. at 408. And an out-of-court statement may be
called “inconsistent” with the in-court statement because
of a curious omission as well as a flat contradiction.
Sherlock Holmes recognized in Silver Blaze that the dog’s
failure to bark, when barking would have been expected,
conveyed a powerful message. So the Court remarked:
Each of two inconsistent descriptions of events
may be said to involve “silence” insofar as it omits
facts included in the other version. But Doyle does
not require any such formalistic understanding
of “silence,” and we find no reason to adopt such
a view in this case.
Id. at 409. This is the idea behind the questions that the
prosecutor asked on cross-examination. Johnson told the
Assistant State’s Attorney what he had done on the day
of the robbery. His story left out all of the vital details
that he supplied on the stand; the prosecutor then

-- 6 of 18 --

No. 08-1731 7
implied that the story told in court must be a recent
invention.
The Appellate Court of Illinois and the federal district
court found a constitutional problem for two reasons.
First, they observed that Johnson’s statement to the
Assistant State’s Attorney (at least as summarized in
the offer of proof) did not mention either White or the
robbery, so that the questioning at trial did not cover
the same “subject matter” as the statement. Second, they
thought it hard to see how Johnson could be questioned
about inconsistencies between his trial testimony and
his statement to the Assistant State’s Attorney, when
that statement was not in evidence.
Neither of these points is enough, independently, to
support the conclusion that the questions transgressed
the holding of Doyle. That the statement to the Assistant
State’s Attorney did not mention White is the very thing
that made it so curious. Johnson had been arrested at
the motel. He was later questioned about what he did
on the day of the robbery. He waived his right to remain
silent (and thus, one might think, abandoned the founda-
tion for Doyle’s rule). Yet instead of relating facts that,
if true, would have led to his release from custody,
Johnson said only inconsequential things. A jury might
conclude that he did this because he had yet to invent a
story that was consistent with innocence—and Charles
holds that once a suspect agrees to speak he may be
questioned about telling omissions, which are a form
of inconsistency. See Phelps v. Duckworth, 772 F.2d 1410
(7th Cir. 1985) (en banc).

-- 7 of 18 --

8 No. 08-1731
As for the fact that the prosecution did not perform
its obligations in discovery: A violation of state law is not
the basis for federal collateral relief. See, e.g., Estelle v.
McGuire, 502 U.S. 62 (1991). Illinois is free to hold as a
matter of its own law that, if a defendant’s out-of-court
statement has been excluded from evidence, no cross-
examination to draw out inconsistency between that
statement and the in-court testimony is allowable. But
that is not what the state’s appellate court held. It
relied on Doyle rather than any state rule restricting cross-
examination. Yet Doyle does not establish a federal rule
that the out-of-court statement must be introduced
before cross-examination about revealing omissions may
be conducted; Doyle applies only when the suspect
invokes his federal right to remain silent, which Johnson
did not.
The state discovery violation has presented a compli-
cation of a different sort for this proceeding, however.
All we have is the offer of proof. The full statement to
the Assistant State’s Attorney is not in the record—by
which we mean, not in the federal record. And the
absence of the statement makes it difficult to apply
Doyle. When, precisely, did Johnson receive Miranda
warnings? (Doyle applies only to post-warning statements
or omissions. See Fletcher v. Weir, 455 U.S. 603 (1982).)
Was Johnson really silent about White’s role? We don’t
know. What, if anything, did he say about how he came
to have part of the $20 bill? We don’t know. Did he say
anything about how White could have carried a sawed-off
shotgun without his knowledge? We don’t know.

-- 8 of 18 --

No. 08-1731 9
Unless Charles means that a defendant who says
anything may be cross-examined about his failure to tell
the whole story, the absence of the full statement
impedes evaluation. And it is hard to think that the
Supreme Court meant to say that, if a suspect tells a knock-
knock joke after receiving Miranda warnings, and
then clams up, Doyle drops out of the picture. The
suspect must say enough to make the omission of the
story told at trial a sound basis for an adverse inference.
We just don’t know whether that is true of the difference
between Johnson’s statement to the Assistant State’s
Attorney and his testimony at trial. For all the record
shows, Johnson chatted amiably until the Assistant
State’s Attorney asked about the robbery, after which
he invoked his right to silence.
Normally the prisoner bears the burden of production
and the risk of non-persuasion in a federal collateral
attack; after all, the petitioner is contesting a final decision
of a state court. But the fact that the state’s appellate
court ruled in Johnson’s favor on the Doyle issue may
have led him to think it unnecessary to mount a full
presentation in the federal proceeding. What’s more, we
do not know whether Johnson (or his lawyers) ever
received a copy of the full statement. It is the Attorney
General of Illinois who wants to contest the state judi-
ciary’s resolution of the Doyle question—yet the Attorney
General did not furnish the federal court with a complete
copy of Johnson’s statement. Under the circumstances, it
is best to proceed as the state court did: To assume that
Doyle barred at least some of the 15 questions to which
objections were overruled, and to ask whether the
error was harmless.

-- 9 of 18 --

10 No. 08-1731
The harmless-error question has some difficulties of its
own. The first is the standard of federal review. Section
2254(d) provides:
An application for a writ of habeas corpus on
behalf of a person in custody pursuant to the
judgment of a State court shall not be granted
with respect to any claim that was adjudicated on
the merits in State court proceedings unless
the adjudication of the claim—
(1) resulted in a decision that was contrary to,
or involved an unreasonable application of,
clearly established Federal law, as determined
by the Supreme Court of the United States; or
(2) resulted in a decision that was based on an
unreasonable determination of the facts in
light of the evidence presented in the State
court proceeding.
How does this language apply when the state court has
found a violation of the Constitution but concluded that
the error was harmless? One approach, which the
district court used in reliance on Fry v. Pliler, 551 U.S. 112
(2007), is to tackle the issue independently, using the
standard laid down in Brecht: Whether the Doyle viola-
tion “had substantial and injurious effect or influence in
determining the jury’s verdict.” The other possible ap-
proach is to ask whether the state court applied federal
law unreasonably when holding that the error was harm-
less beyond a reasonable doubt. See Mitchell v. Esparza,
540 U.S. 12 (2003). The reasonable-doubt standard is
the right one on direct appeal, see Chapman v. California,

-- 10 of 18 --

No. 08-1731 11
386 U.S. 18 (1967), just as the substantial-effect standard
is appropriate on collateral review.
Esparza holds that, when a state court has found a
constitutional error harmless beyond a reasonable
doubt, the federal court’s initial question is whether that
decision represents an “unreasonable application of
clearly established Federal law”. The Justices wrote: “We
may not grant [a] habeas petition . . . if the state court
simply erred in concluding that the State’s errors were
harmless; rather, habeas relief is appropriate only if
the [State] Court of Appeals applied harmless-error
review in an ‘objectively unreasonable’ manner.” 540 U.S.
at 18.
The district court did not mention Esparza. It relied on
Fry for the proposition that a federal court should make
an independent decision (though under the Brecht
standard rather than the reasonable-doubt standard). The
question at issue in Fry, however, was not whether a
federal court should disregard a state court’s considered
decision on the subject of harmless error. It was instead
what a federal court should do if the state court con-
cludes that no constitutional error occurred and there-
fore does not make a harmless-error decision. If the
federal court concludes that constitutional error took
place, and the last state court to address the case did not
apply any kind of harmless-error review, which is the
appropriate standard: Brecht or Chapman? The Court
held that Brecht supplies the standard for a federal court
to use in all collateral attacks, when making an independ-
ent evaluation of a trial error’s effects.

-- 11 of 18 --

12 No. 08-1731
Fry did not overrule Esparza. To the contrary, the Justices
wrote, “[i]n Mitchell v. Esparza, 540 U.S. 12 (2003) (per
curiam), we held that, when a state court determines
that a constitutional violation is harmless, a federal court
may not award habeas relief under §2254 unless the
harmlessness determination itself was unreasonable.
Petitioner contends that §2254(d)(1), as interpreted in
Esparza, eliminates the requirement that a petitioner
also satisfy Brecht’s standard. We think not.” 551 U.S.
at 119.
If the state court has conducted a harmless-error
analysis, the federal court must decide whether that
analysis was a reasonable application of the Chapman
standard. If the answer is yes, then the federal case is
over and no collateral relief issues. That’s the holding of
Esparza. If the answer is no—either because the state
court never conducted a harmless-error analysis, or
because it applied Chapman unreasonably—then §2254(d)
drops out of the picture and the federal court must
make an independent decision, just as if the state court had
never addressed the subject at all. And we know from
Fry that, when this is so, a federal court must apply the
Brecht standard to determine whether the error was
harmless. See also, e.g., Smiley v. Thurmer, 542 F.3d 574,
583–84 (7th Cir. 2008); Aleman v. Sternes, 320 F.3d 687 (7th
Cir. 2003).
So did the state’s appellate court analyze the harmless-
error question reasonably? Because its opinion is unpub-
lished (and not available on the Internet), we reproduce
the pertinent portion here:

-- 12 of 18 --

No. 08-1731 13
Before a constitutional error can be held harmless,
the reviewing court must be able to conclude
beyond a reasonable doubt that the error did not
contribute to the finding of guilt. People v. Averhart,
311 Ill. App. 3d 492, 505–06, 724 N.E.2d 154 (1999)
citing Chapman v. California, 386 U.S. 18, 24, 87
S. Ct. 824, 828, 17 L. Ed. 2d 705, 710–11 (1967). The
burden of proof is on the State to show beyond a
reasonable doubt that the jury verdict would
have been the same absent the error. People v.
Thurow, 203 Ill. 2d 352, 363, 786 N.E.2d 1019 (2003).
There are five factors courts consider in determin-
ing whether a Doyle violation is harmless beyond a
reasonable doubt: (1) the intensity and frequency
of the references to the defendant’s silence;
(2) which party elected to pursue the line of ques-
tioning; (3) the use that the prosecution made of
the defendant’s silence; (4) the trial court’s oppor-
tunity to grant a motion for mistrial or to give a
curative jury instruction; and (5) the quantum of
other evidence establishing defendant’s guilt.
People v. Dameron, 196 Ill. 2d 156, 164, 751 N.E.2d
1111 (2001).
In light of the overwhelming quantum of evidence
of defendant’s guilt as established from the eye-
witness identification testimony, we conclude
that the Doyle violation was harmless beyond a rea-
sonable doubt. Defendant was identified by two
eyewitnesses whose identifications were reliable.
Factors that are considered in determining the
reliability of an out-of-court identification include

-- 13 of 18 --

14 No. 08-1731
the witness’s opportunity to view the accused at
the time the crime occurred, the witness’s degree of
attention, the accuracy of the witness’s prior
description, the degree of certainty demonstrated
by the witness at the subsequent confrontation,
and the amount of time that passes between the
crime and the confrontation. People v. Taylor, 143
Ill. App. 3d 252, 255, 492 N.E.2d 1011 (1986). In this
case, the armed robbery occurred in a fully illumi-
nated restaurant with the offender in close proxim-
ity to the eyewitnesses, making no attempt to
conceal his identity. Moreover, the armed robbery
lasted approximately ten minutes, during which
time the offender ordered Susanna Ramos to open
her cash register and moments later ordered the
manager, Alicia Ortega, to sequentially open four
cash registers.
The record indicates that Monique Nolan and
Susanna Ramos both had sufficient opportunity to
observe the offender during the armed robbery and
both were paying attention. Nolan testified that
during the armed robbery she stood only five to
six feet away from the offender and that she con-
tinually looked and peeked at the offender from a
corner of the employee area of the restaurant.
Ramos testified that when the offender first ap-
proached her at the front counter, she looked him
in the face and asked for his order. According to
Ramos, the offender stood at the front counter
for about a minute without responding and then
walked away. Minutes later, the offender returned

-- 14 of 18 --

No. 08-1731 15
to the counter where he pointed a sawed-off shot-
gun at Ramos’ chest and demanded that she open
her cash register. Ramos testified that at this time
the shotgun actually touched her chest and she
looked directly at the offender’s face for one to
two minutes before turning away because she
was frightened.
Defendant argues that Nolan’s and Ramos’ failure
to describe his facial hair rendered their identifica-
tions unreliable. We disagree. “Experience tells us
that an identification is not usually made by distin-
guishing separate features but by the total impres-
sion made upon the witness.” People v. Smith, 52 Ill.
App. 3d 583, 587, 367 N.E.2d 756 (1977). Thus,
courts have generally held that an eyewitness’s
failure to include facial hair in an initial descrip-
tion constitutes a minor discrepancy, affecting
the credibility of the description, but not destroy-
ing the validity of the identification testimony.
Taylor, 143 Ill. App. 3d at 255. Under the circum-
stances of this case, Nolan’s and Ramos’ failure
to describe defendant’s facial hair in their initial
description of defendant did not render their
identification testimony unreliable.
Therefore, we find that in light of the overwhelm-
ing quantum of evidence of defendant’s guilt as
established from the eyewitness identification
testimony, the Doyle violation was harmless be-
yond a reasonable doubt. We are convinced that
the jury would have convicted defendant

-- 15 of 18 --

16 No. 08-1731
absent the prosecutor’s improper reference to
defendant’s post-Miranda silence.
This passage articulates the Chapman rule accurately, and
it reads like a reasonable application. It is in some
respects favorable to Johnson—it recognizes a weakness
in the eyewitness testimony, and it does not rely on
anything that White said, the torn $20 bill, the fact that
the manager of the Trak Auto store testified that
Johnson displayed a wad of crumpled bills, or the sawed-
off shotgun and loot found in his car.
Johnson contends that Nolan and Ramos gave testi-
mony at odds with that of White, who said that the
robbery started soon after they entered the restaurant,
rather than a few minutes later. Disagreements about
time are common in eyewitness testimony; a few minutes
may seem like much longer to someone staring at the
barrel of a sawed-off shotgun. It is almost always possible
to find inconsistencies in the narrations of multiple eye-
witnesses, and incongruities in the story of any one
eyewitness. The shortcomings of human observation
and memory do not make it unreasonable for a state
court to find an error harmless beyond a reasonable
doubt. Questions that violate Doyle’s rule are unlikely to
change a jury’s mind. These questions do not get damning
evidence before the jury, and “why didn’t you tell this
story before?” is something that jurors are likely to
wonder, whether or not a prosecutor makes the question
explicit.
Perhaps we would have reached a different conclusion,
had we been faced with the harmless-error question in

-- 16 of 18 --

No. 08-1731 17
a case on direct appeal in federal court. But whether the
state court erred (in the sense that federal judges
would have decided otherwise) does not matter under
§2254(d). We ask only whether the state court’s resolu-
tion was reasonable. This considered conclusion was
reasonable, so the state court’s decision must stand.
REVERSED
CUDAHY, Circuit Judge, concurring. The district court,
after what seems to me to be a searching analysis under
Brecht, reached a point where it was in “grave doubt”
about the harmlessness of the Doyle error in Johnson’s
state court trial and, citing Fry v. Pliler, resolved its state
of “equipoise” by granting the habeas petition. See Fry
v. Pliler, 127 S. Ct. 2321, 2328 n.3 (2009) (“We have previ-
ously held that, when a court is ‘in virtual equipoise as
to the harmlessness of the error’ under the Brecht
standard, the court should ‘treat the error . . . as if it
affected the verdict . . . .’ ”) (quoting O’Neal v. McAninch,
513 U.S. 432, 435 (1995)).
The majority resolves the question of the harmlessness
of the Doyle violation by finding that the state court
applied settled law reasonably under § 2254(d), and
therefore does not reach the Fry/Brecht question. This
seems unsettling because it points to tension between
our precedent regarding “reasonableness” under § 2254(d)

-- 17 of 18 --

18 No. 08-1731
and the Supreme Court’s precedent on how to resolve
questions of harmlessness under Brecht. In the end,
I believe the legal gymnastics can, perhaps, be reduced
to a difference of opinion about the effect of the error
in question. The district court judge was convinced
that the Doyle error made a difference in the outcome
of Johnson’s trial (or he had grave doubts, anyway). The
majority on appeal believes that a Doyle violation rarely,
if ever, makes this kind of a difference and that the
state court was therefore reasonable in finding it made
no difference here.
Because habeas petitions are subject to de novo
review, the majority analysis results in a reversal. But the
majority does little to clarify how district courts are to
apply § 2254(d)’s “reasonableness” inquiry in the light
of Fry’s instruction to “assess the prejudicial impact of
constitutional error in a state-court criminal trial under
the . . . standard set forth in Brecht,” and apparently to
grant the petition where “grave doubts” lead to a state
of “equipoise.” Fry, 127 S. Ct. at 2328 & n.3. Although
these complications are troubling, I do not find them
adequate grounds to disagree with the outcome and
therefore I join the majority opinion.
7-14-09

-- 18 of 18 --