In the
 United States Court of Appeals
              For the Seventh Circuit
                         ____________

No. 05-1573
JARRETT M. ADAMS,
                                              Petitioner-Appellant,
                                v.

DANIEL BERTRAND,
                                             Respondent-Appellee.
                         ____________
            Appeal from the United States District Court
               for the Eastern District of Wisconsin.
              No. 04 C 389—J. P. Stadtmueller, Judge.
                         ____________
     ARGUED FEBRUARY 21, 2006—DECIDED JUNE 30, 2006
                         ____________


  Before MANION, WOOD, and EVANS, Circuit Judges.
  MANION, Circuit Judge. A Wisconsin jury convicted Jarrett
Adams, along with Dimitri Henley, of five counts of second-
degree sexual assault, and Adams was sentenced to twenty-
eight years in prison. Adams appealed his conviction to the
Wisconsin Court of Appeals and the Wisconsin Supreme
Court, neither of which granted him relief. Having ex-
hausted his state court remedies, Adams filed a petition for
a writ of habeas corpus in the Eastern District of Wisconsin,
challenging both the sufficiency of the evidence and the
effectiveness of his trial counsel. The district court denied
2                                                    No. 05-1573

his petition. We reverse the district court’s decision based
on the constitutional deficiencies of his trial counsel.


                                I.
  In September 1998, Jarrett Adams, Dimitri Henley, and
Rovaughn Hill, three men from Chicago, were selling
cologne at the University of Wisconsin at Whitewater
shortly before fall semester classes were to begin.1 On the
evening of September 5, the men were playing video games
in the dorm room of Shawn Demain, a freshman whom they
had met during their sales rounds. Meanwhile, S.E.S.2, a
female freshman who resided in the same dorm as Demain,
and S.E.S.’s roommate, Heidi Sheets, were attending a party
a few blocks away. While at the party, S.E.S. had approxi-
mately three or four beers. Upon returning to the dorm later
that evening, the women went to Demain’s dorm room,
where they first met the three men from Chicago.
  At trial,3 Sheets and S.E.S. offered different stories of what
happened next. Sheets claimed that both women invited the



1
  The record does not make clear whether this was a legitimate
business nor whether the men were selling door-to-door in the
dorm. However, the propriety of their presence on the campus is
not an issue in this case.
2
  Both parties refer to the victim of the sexual assault as S.E.S.,
which is consistent with the state and district court pleadings. We
adopt the same convention.
3
  The “trial” refers to the trial of Jarrett Adams and Dimitri
Henley on February 8-9, 2000. All three men had been tried
together the previous year, which resulted in a mistrial because
of an improper attempt by the state to amend its charges at
the close of the evidence.
No. 05-1573                                                 3

men up to their room, which was located three floors above
Demain’s. While the men and S.E.S. proceeded to the room,
Sheets went down the hall for approximately twenty-five
minutes to speak with another friend. According to S.E.S.,
however, neither she nor Sheets invited the three men to
their room. Rather, she left Demain’s room by herself and
went to her room. When she opened the door to her room,
the three men suddenly appeared at the door and barged in,
telling S.E.S. to put on music so that they could dance. After
selecting a compact disc, the men took turns dancing
provocatively with S.E.S., groping her breasts and grabbing
her crotch. She claimed she did not resist because she was
scared. The situation deteriorated further. One of the men
pulled S.E.S. onto Sheets’s bed, and Adams, standing before
her, pulled down his pants, leaving him with only his
boxers on.
  At this point, Sheets entered the room. Both Sheets and
S.E.S. testified that Sheets became upset and stormed out of
the room. S.E.S. followed Sheets out of the room, and none
of the men restrained her from doing so. Sheets entered
another room on the same floor, but refused to let S.E.S.
inside. Through the closed door, S.E.S. asked Sheets whether
she was mad at her and overheard Sheets calling her a slut.
During her time in the hall, S.E.S. never indicated that
anything was amiss or attempted to enter another room.
After waiting a few minutes, S.E.S. left the door and decided
to go downstairs instead of returning to her room. Upon
reaching the staircase, however, she came across Hill
waiting for her. He turned S.E.S. around, directing her back
to her room where the other men were waiting.
  Upon returning to her room, the men ordered her to sit on
the floor, and S.E.S. complied, sitting on the floor with her
legs crossed “Indian style.” Shortly thereafter Adams told
4                                                  No. 05-1573

her to lie down, which S.E.S. did. Adams tried to remove
her pants, while S.E.S. kept her knees close together in an
attempt to stop him. Eventually, however, in the face of
Adams’s continued tugging, she relaxed her legs. He
removed S.E.S.’s pants and underwear, and then proceeded
to have sex with her. After Adams stopped, the other men
took turns having sex with S.E.S. During this time, one of
the men also rubbed his penis against her face, telling her to
“suck this,” as S.E.S. turned her face to avoid it.
  The three men left, and S.E.S. called her boyfriend, Joshua
Lodwick. Although S.E.S. did not want to go to the hospital
or talk to the police, Lodwick and others convinced her to
do both. Adams, Hill, and Henley were eventually charged
with second-degree sexual assault. They were first tried
together in 1999, but that trial resulted in a mistrial after the
state attempted to amend its charges after the close of
evidence. Subsequently, Adams and Henley were tried
together. Later, Hill was tried separately.
   At the second trial, both John Fiske, Adams’s lawyer, and
Steve Luchsinger, Henley’s lawyer, cross-examined S.E.S.
about earlier statements to her treating doctor and at the
first trial, pointing out inconsistencies among various
accounts she had given. Adams and Henley offered no
defense witnesses, choosing instead to rely on their percep-
tion that the state failed to prove all the elements of second-
degree sexual assault. The jury, however, found Adams and
Henley guilty on all five counts. Hill, on the other hand,
called defense witnesses at his subsequent trial, including
Demain, who testified that he saw S.E.S. smoking cigarettes
with the three men after the alleged assault occurred. This
trial ended in a hung jury. While the state renewed its
prosecution of Hill later, the court eventually dismissed the
case against Hill because the state had failed to turn over
certain police investigation notes to the defense.
No. 05-1573                                                    5

   On appeal, Adams mounted a challenge to the sufficiency
of the evidence, as well as various decisions made by his
trial counsel. The Wisconsin Court of Appeals affirmed the
conviction, and the Wisconsin Supreme Court denied
review. Adams then proceeded to the Eastern District of
Wisconsin, where he filed a petition for a writ of habeas
corpus, again challenging the sufficiency of the evidence
and his attorney’s performance. The district court denied the
petition, finding the evidence sufficient to support Adams’s
convictions and that his attorney made reasonable strategic
trial choices.


                               II.
  Before this court, Adams again challenges the sufficiency
of the evidence and his counsel’s effectiveness. Adams first
claims that the state failed to establish that his sexual contact
with S.E.S. involved either the use or threat of force or lack
of consent, each of which is a required element to prove
second-degree sexual assault in Wisconsin. Adams also
argues that his trial attorney was constitutionally ineffective
based on several events at trial. Specifically, Adams chal-
lenges his counsel’s failure to request a jury instruction for
a lesser included offense, his decision not to vigorously
cross-examine S.E.S., and his failure to properly investigate
Demain.
  On appeal from a denial of habeas corpus relief, we
review issues of law de novo and issues of fact for clear
error. See Perry v. McCaughtry, 

, 688 (7th Cir.
2002). To qualify for habeas relief, Adams must show that
the state court proceedings adjudicating his claim “resulted
in a decision that was contrary to, or involved an unreason-
able application of, clearly established Federal law, as
6                                                  No. 05-1573

determined by the Supreme Court of the United States.” 28
U.S.C. § 2254(d)(1). See also Van Patten v. Deppisch, 

, 1042 (7th Cir. 2006); Bintz v. Bertrand, 

, 865
(7th Cir. 2005). A state court decision constitutes an unrea-
sonable application of clearly established federal law if the
state court correctly identifies the governing legal principle
from Supreme Court jurisprudence but unreasonably
applies that principle to the facts of the present case. See
Dunlap v. Hepp, 

, 741 (7th Cir. 2006). We have
previously explained that an unreasonable decision, in this
context, is one “well outside the boundaries of permissible
differences of opinion.” McFowler v. Jaimet, 

, 455
(7th Cir. 2003).


                              A.
  Adams first mounts challenges to the sufficiency of the
evidence that established two of the elements of second-
degree sexual assault. On direct appeal, the Wisconsin Court
of Appeals reviewed Adams’s sufficiency claims under the
standard annunciated in State v. Poellinger, 

,
755 (Wis. 1990): “an appellate court may not reverse a
conviction unless the evidence, viewed most favorably to
the state and the conviction, is so insufficient in probative
value and force that no trier of fact, acting reasonably, could
have found guilt beyond a reasonable doubt.” Wisconsin
effectively duplicates the Supreme Court standard for
sufficiency challenges, see Jackson v. Virginia, 

,
324 (1979), and Adams does not claim that the Wisconsin
Court of Appeals applied the wrong standard.4 Rather,


4
    As this court has repeatedly stated on habeas review, a
                                               (continued...)
No. 05-1573                                                    7

Adams suggests that the Court of Appeals applied the
proper standard unreasonably.
  Wisconsin defines the crime of second-degree sexual
assault as “sexual contact or sexual intercourse with another
person without consent of that person by use or threat of
force or violence.” Wis. Stat. § 940.225(2)(a). Turning to
Adams’s initial challenge, he contends that the state failed
to prove the “use or threat of force or violence” element. A
Wisconsin court has previously explained:
    The “use or threat of force or violence element of
    violence” element of second-degree sexual assault
    under Wis. Stat. § 940.225(2)(a) is satisfied if the use or
    threat of force or violence is directed to compelling
    the victim’s submission. The element is satisfied
    whether the force is used or threatened as part of the
    sexual contact or whether it is used or threatened as part
    of the sexual contact to compel the victim’s submission.
State v. Hayes, 

, 215 (Wis. 2004). The Wiscon-
sin Supreme Court has clearly stated that “use of force”
would include both forcible contact or the force used as the
means of making contact. See State v. Bonds, 

,
267 (Wis. 1991). To put it another way, force used during the
sexual act or contact to compel submission can satisfy the
statute, not just force prior to the actual sexual act. In Bonds,


4
   (...continued)
state court need not cite to the applicable Supreme Court case
annunciating the standard, as long as the state court precedent
cited agrees with it. See, e.g., Charlton v. Davis, 

,
374 (7th Cir. 2006) (“A state court need not even be aware of
Supreme Court precedent so long as neither the reasoning nor the
result of the state-court decision contradicts them.”) (internal
citations omitted).
8                                                 No. 05-1573

for example, the Wisconsin Supreme Court found that
“forcibly grabbing her nipple and then squeezing and
pulling it” sufficed for the force element of second- degree
sexual assault, though it also was the extent of the sexual
contact. 

In addition to the use of actual force, a
threat of force that compels submission also satisfies this
requirement. “The message conveyed by a threat is deter-
mined by the context in which it occurs.” State v. Jaworski,

, 30-31 (Wis. Ct. App. 1986). This determina-
tion requires an examination of such factors as the place
where the event occurred, the victim’s understanding of the
threat, and physical disparities between victim and assail-
ant. See 

Court of Appeals did not dwell very long
on Adams’s sufficiency challenge, simply concluding that
“[t]he victim’s testimony, if believed, was sufficient to
support the charges the jury was instructed on.” Wisconsin
v. Adams, No. 02-0039-CR, 2002 Wisc. App. LEXIS 1214, at
*9-10 (Nov. 7, 2002). Viewing the evidence in the light most
favorable to the conviction, we agree that sufficient evidence
established both use of force and threats of force. Specifi-
cally, S.E.S. testified that the men groped her breasts and
crotch while dancing with her in her room. After she
followed Sheets out of the room, she stated that she at-
tempted to go downstairs, only to have one of the men
physically turn her around and lead her back into the room.
Upon her return to the room, she attempted to keep her legs
together, but Adams persisted in taking off her pants and
underwear. While S.E.S. did not want him to do this, she
eventually could no longer keep her legs together in light of
his repeated attempts. All of these actions were uses of force
to compel her submission to sexual contact or intercourse.
While this is not a case of a beating or extreme violence
against S.E.S., the statute does not require that type of force.
No. 05-1573                                                9

It only requires force to compel submission—precisely what
happened here according to S.E.S.’s testimony.
   Moreover, even if there were insufficient evidence of
actual force, sufficient evidence supported a finding of a
threat of force. As stated above, threats of force are evalu-
ated based on the context in which they accrue. See 

. Viewing the evidence in the light most
favorable to the conviction, the context in this case estab-
lishes a threat directed to obtaining S.E.S.’s submission.
S.E.S. testified that as she entered her room, she was
surprised by three men that she barely knew, who followed
her into her room uninvited. The men then ordered her to
put on music and groped her while dancing. One of the men
pulled her onto a bed, while Adams dropped his pants. She
was able to leave the situation temporarily when she
unsuccessfully tried to enter another room where her
roommate had relocated. When she attempted to go down-
stairs, she was met at the stairwell by one of the men and
returned to her room. At that time, the room was dark, and
she was surrounded by three men who ordered to sit down,
then lie down, with Adams proceeding to take off her pants
and underwear despite her attempts to prevent him. While
this case does not feature an explicit threat, given the
number of men, their orders to S.E.S., and their aggressive,
sexually-charged actions despite her unwillingness, the
context described by S.E.S. was one of intimidation geared
to making her submit to sexual intercourse. The Wisconsin
Court of Appeals, therefore, applied Jackson reasonably to
the force element.
  Adams also argues that the evidence of lack of consent
was insufficient. Wisconsin defines consent as “words or
overt actions by a person who is competent to give informed
consent indicating a freely given agreement to have sexual
intercourse or sexual contact.” Wis. Stat. § 940.225(4).
10                                                No. 05-1573

Adams relies on the fact that S.E.S. eventually took part in
the sexual intercourse as an overt act indicating consent.
However, such a reading would hollow out the consent
requirement, basically removing from the realm of sexual
assault any instance when the victim did not overtly resist.
“A victim of sexual assault is not required to resist the
assault to establish that the act was nonconsensual.” State v.
O’Brien, 

, 18 (Wis. 1999); see also State v. Clark,

, 721 (Wis. 1979). At trial, S.E.S. expressly
testified that the men engaged in intercourse with her
without her giving any verbal consent. Moreover, S.E.S.
stated that she committed no overt acts that would suggest
consent to the men, instead testifying that she complied
with their orders out of fear. A reasonable juror could have
concluded that she merely followed directions rather than
gave consent. See 

. The state,
therefore, introduced sufficient evidence that would allow
a jury to conclude neither words nor overt actions supplied
the necessary consent element. Again, the conclusion of the
Wisconsin Court of Appeals was not unreasonable.


                              B.
   Turning to Adams’s second group of claims, he brands his
trial counsel constitutionally deficient because of a number
of questionable trial decisions. The Wisconsin Court of
Appeals properly referenced Strickland v. Washington, 

(1984), in its evaluation of Adams’s ineffective
assistance challenges. The well-known Strickland analysis
requires Adams to demonstrate that: (1) his counsel’s
performance fell below an objective standard of reasonable-
ness, and (2) the deficient performance caused him preju-
dice. See, e.g., Harris v. Cotton, 

, 555 (7th Cir.
No. 05-1573                                                   11

2004); 

. Regarding the performance
inquiry, we presume that counsel’s actions “fall within the
wide range of reasonable professional assistance,” Strick-

, and defer to strategic decision-making
by a trial attorney, see Brown v. Sternes, 

, 691 (7th
Cir. 2002). See also Hough v. Anderson, 

, 891 (7th
Cir. 2002) (“Our review of the performance of counsel must
be highly deferential.”) (internal citations omitted). Despite
this weighty deference, we nonetheless must carefully
consider whether the attorney brought to bear the skill and
knowledge that allows for a proper adversarial testing
process, considering all the circumstances. See 

. To succeed in meeting the prejudice prong of
Strickland, Adams must further show a reasonable probabil-
ity that, but for counsel’s errors, the result of the proceeding
would be different. See id.; see also 

. A
reasonable probability is defined as one that is sufficient
to undermine confidence in an outcome. See 

. “On a habeas appeal, we look at [the
Strickland] test not directly, but rather ask whether the lower
court made a reasonable application of this law to the facts
of this case.” Floyd v. Hanks, 

, 850 (7th Cir. 2004).
  In this case, Adams points to three main instances of
substandard performance by trial counsel. First, he argues
that Fiske unreasonably failed to request a jury instruc-
tion regarding third-degree sexual assault. Second, he
contends that Fiske erred by not vigorously cross-examining
S.E.S. or clearly presenting various inconsistencies among
her descriptions of events. Finally, and most substantially,
Adams claims that his trial counsel failed in his duty to
conduct a reasonable investigation by not locating and
talking to Demain, who could have conceivably served as a
witness on Adams’s behalf.
12                                              No. 05-1573

                    1. Jury Instructions
  Adams claims that Fiske failed in his professional duties
by not seeking a jury instruction for the lesser included
offense of third-degree sexual assault. In Wisconsin, third-
degree sexual assault is “sexual intercourse with a person
without consent of that person.” Wis. Stat. § 940.225(3).
Basically, this offense has the same requirements as second-
degree sexual assault, except it does not necessitate any
showing of force. Adams believes that, given the paucity of
evidence relating to force at his trial, Fiske should have
requested this additional instruction rather than rely upon
an all-or-nothing gamble.
  Fiske did not overstep the bounds of professional compe-
tence in this respect. According to Fiske’s testimony during
a post-conviction proceeding, he was aware of the lesser
included offense of third-degree sexual assault, and he
understood its constituent elements, including the lack of a
force requirement. Fiske, however, decided that allowing
the main offense to stand on its own would make it more
difficult for the government to prove, given the case’s
weaknesses. He chose to bring the lack of evidence of force
into clear view, which he believed would compel the jury to
acquit on the second-degree charges. Fiske discussed this
strategy with his client, and it comported with Adams’s
claims and overall defense— complete innocence of any-
thing other than consensual sex. Fiske believed that the
inclusion of a lesser count instruction would look like
grasping at straws given this theory of innocence.
  We cannot conclude that this choice was professionally
unreasonable. The strategy Adams now advocates has some
initial appeal in hindsight, as the jury obviously found that
the evidence of force was sufficient. That tactic, however,
No. 05-1573                                                 13

while it might ensure a smaller prison sentence, would also
increase the chance of conviction on some charge. Fiske’s
strategy, on the other hand, while somewhat risky, could
have led to a complete acquittal had the jury agreed that
proof of force was lacking. Counsel’s decision here dwells
in the region of tactics and strategy, and the Wisconsin
Court of Appeals did not act unreasonably when concluding
that Fiske had not committed a Strickland error.


                   2. Cross-Examination
  Adams also asserts that his counsel failed to vigorously
confront S.E.S. with various prior inconsistent state-
ments about the assault. At the post-conviction hearing,
Fiske testified that he reviewed S.E.S.’s statements to
doctors, as well as testimony from prior hearings and the
initial trial. As mentioned before, Adams was tried jointly
with Henley. Fiske, along with Henley’s counsel,
Luchsinger, decided on a common trial strategy. Specifi-
cally, the attorneys agreed that, rather than both attacking
perceived inconsistencies, Luchsinger would be the more
aggressive in the cross-examination of S.E.S. This strategy
was born out of concern that if both lawyers aggressively
confronted the assault victim, it might alienate members
of the jury sympathetic to her. Nonetheless, Fiske felt
confident that either Luchsinger or himself would (and did)
address all inconsistencies at trial. Fiske articulated plausi-
ble strategic reasons for his actions, and the Wisconsin
Court of Appeals again reached a reasonable conclusion.


               3. Investigation into Demain
 Finally, Adams alleges that Fiske acted unreasonably
when he failed to locate and call Demain as a witness. As we
14                                                       No. 05-1573

have noted repeatedly, a lawyer owes his client “ ‘a duty to
make reasonable investigation or to make a reason-
able decision that makes particular investigations unnec-
essary,’ and a failure to investigate can certainly consti-
tute ineffective assistance.” Washington v. Smith, 

, 631 (7th Cir. 2000) (quoting 

);
See also 

; 

.
The Wisconsin Court of Appeals concluded that Fiske
reasonably chose to limit investigations into Demain
because he had decided not to call any defense witnesses as
part of his strategy. Adams, 2002 Wisc. App. LEXIS 1214, at
*5-6.
  We disagree. The “failure to investigate a particular lead
may be excused if a lawyer has made a ‘reasonable decision
that makes particular investigations unnecessary.’ ” Wash-

(quoting 

).
“In our assessment of counsel’s actions, we evaluate coun-
sel’s conduct from the perspective at the time the decision
was made to forgo the investigation.” 

. Here, Fiske made an unreasonable decision not to
investigate Demain because Fiske possessed knowledge,
before the trial, that Demain could have swung the case in
his client’s favor.5
  Fiske knew that Demain, like Sheets, appeared throughout
all accounts of the events, both before the assault and after,
and thus was a witness to several key moments in the


5
   As stated previously (and which will be discussed below), the
trial that resulted in Adams’s conviction was actually the sec-
ond trial in this matter. In the first trial, the state presented all of
its evidence, but the district court granted a mistrial based on
an improper amendment of the charges after the close of the
evidence.
No. 05-1573                                                15

chronology. Unlike any other witness, Demain had no prior
relationship with either the men or S.E.S., thus rendering
him impartial and independent. Demain could explain what
occurred in his room prior to the sexual encounter, includ-
ing whether he heard the women invite the men to visit
their room and what happened when S.E.S. left his room.
Further, the police report, which Adams’s attorney reviewed
as part of his preparations, contained a statement from
Demain that he witnessed S.E.S. downstairs with the three
men after the time of the sexual encounter. Demain, there-
fore, appeared to be a witness who could shed considerable,
perhaps conclusive, light on the events of that night.
  Demain gained added importance given Fiske’s unique,
and enviable, position of having seen the state’s entire
case against his client. As mentioned, the initial trial of
Adams, Henley, and Hill ended in a mistrial after the
presentation of the state’s case because of the state’s im-
proper attempt to amend a pleading. Fiske and the other
attorneys witnessed a dress rehearsal of the state’s case and
could prepare accordingly before the second trial. Fiske
knew that Sheets and S.E.S. did not agree on whether the
women invited the men up to their room before the sex-
ual encounter. Demain, who was in the room at the time,
could have resolved this dispute in Sheets’s favor, and
thus completely undercut S.E.S.’s version of uninvited
strangers bursting into her room. Moreover, Fiske reviewed
the police report that indicated that Demain saw S.E.S. with
the men later that night. This would seem to strongly
contradict S.E.S.’s story of a sexual assault, the men quickly
leaving the scene, and S.E.S. calling her boyfriend im-
mediately following the alleged rape. Instead, this would
support a plausible theory of a consensual encounter
between the men and S.E.S., thus damaging the state’s
theory of the case.
16                                                No. 05-1573

   Demain was not an irrelevant bystander. Rather, he
was present at two crucial events where the evidence is
sharply conflicting. Given the testimony at the first trial, it
was essential that Demain be found in order to determine
what he saw and heard before informed decisions about
strategy (no defense witnesses) could be made. Although
Fiske made minimal attempts to locate Demain, when the
initial attempts failed, he let the matter drop. Fiske did not
hire an investigator, and his total efforts were comprised of
asking witnesses and the University Demain’s location. As
Luchsinger, the co-defendant’s counsel, admitted during
a postconviction proceeding, he and Fiske had no real
strategic reason for their decision not to call Demain besides
“that we were not going to present a defense and ride that
horse, hoping that no defense would carry, would outweigh
the advantages of a favorable witness.” Despite the fact that
Fiske had the benefit of hearing the prosecution’s case
before a mistrial, knew the relevant inconsistencies in the
stories, and was aware that Demain had information about
these crucial facts and could resolve these key points in his
client’s favor, he chose not to track down Demain because
he had already decided not to call witnesses in the second
trial. This knowing disregard was unreasonable. See Wash-
ington at 631.
  Fiske committed to a predetermined strategy without a
reasonable investigation that could have produced a pivotal
witness. Therefore we find that the conclusion of the
Wisconsin Supreme Court (that Fiske’s decision not to call
defense witnesses without first locating and interviewing
Demain was reasonable trial strategy) was not a reasonable
application of the first prong of Strickland.
No. 05-1573                                                 17

   Turning to the second prong of the Strickland analysis,6 we
ask whether there is a reasonable probability that, but for
Fiske’s error, the trial result would be different. See Strick-

. As mentioned previously, a reasonable
probability is one undermining our confidence in an
outcome. See 

In this case, Adams has established
prejudice from the failure to investigate and call Demain.
This was a very close case, and its outcome was highly
contingent on the credibility of S.E.S. Looking at Demain’s
testimony in Hill’s trial (and in the police report), he was
able to place S.E.S. with the men smoking cigarettes after the
alleged assault. He also testified that the women left with
two of the men and that they were going to the women’s
dorm room. In addition, he testified that the third man, Hill,
stayed behind for about thirty minutes playing a video
game. This could have helped explain why S.E.S. encoun-
tered Hill on the stairs after her closed-door conversation
with Sheets. Such testimony completely contradicts S.E.S.’s
description of the events of that night. Nor can we blind
ourselves to the results of Hill’s trials. Hill, who was
charged with precisely the same offenses as Henley and
Adams, was tried in February 2001 and presented Demain’s
testimony. This trial resulted in a hung jury. Eventually, the
charges against Hill were dismissed at a later trial because
of the prosecution’s failure to turn over police notes regard-
ing their investigation. We do not contend that the result
from Hill’s trial would necessarily dictate the same result
had Fiske investigated Demain. But the outcome of that
trial, together with Demain’s testimony and the relatively
thin evidence presented at Adams’s trial, does undermine


6
  As the Wisconsin Court of Appeals concluded that counsel’s
actions regarding Demain were reasonable, it did not, nor was
required to, inquire into prejudice.
18                                               No. 05-1573

our confidence in the outcome, such that there is a reason-
able probability that the outcome of Adams’s trial would be
different.


                            III.
  Adams failed to show that the Wisconsin Court of Ap-
peals acted unreasonably when it found sufficient evidence
to support his convictions. His trial counsel’s failure to
investigate properly a crucial witness, stemming from an
intransigence regarding his trial strategy, however, consti-
tuted ineffective assistance of counsel. The Wisconsin Court
of Appeals acted unreasonably in its application of Strick-
land to the facts on this point. We, therefore, REVERSE the
decision of the district court and REMAND with instructions
to grant the writ, unless the state gives Adams a new trial
within 120 days.

A true Copy:
       Teste:

                          _____________________________
                           Clerk of the United States Court of
                             Appeals for the Seventh Circuit




                   USCA-02-C-0072—6-30-06