Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

The Lone Miscreant, the Self-Training Prosecutor, and Other Fictions:
A Comment on Connick v. Thompson
By Susan A. Bandes*

In Connick v. Thompson, the Supreme Court blocked one of the last remaining paths to
prosecutorial accountability for violating constitutionally mandated discovery obligations under
Brady v. Maryland.1 Two terms ago in Van de Kamp v Goldstein the Court expanded the scope
of absolute immunity for individual prosecutors to encompass supervisory failures to train,
supervise or institute programs to comply with Brady. 2 The upshot of Connick v. Thompson is
that now, unless non-compliance is frequent and notorious enough to reach the level of custom,
prosecutors‟ offices are insulated from Section 1983 liability—entity as well as individual--for
failing to comply with Brady. The decision also bodes ill for prosecutorial accountability more
generally, and for failure to train liability across the board.
The Connick decision has attracted notice mainly for its compelling facts. Plaintiff John
Thompson spent eighteen years in prison, fourteen of them on death row, for a crime he didn‟t
commit.3 Beginning prior to Thompson‟s trial in 1985, the team engaged in prosecuting

*Professor of Law and Dean’s Distinguished Scholar, The University of Miami School of Law. I owe thanks to
Don Doernberg and Robert Mosteller for extremely helpful comments on an earlier draft, to Michael Kozik,
University of Miami Law School Class of 2013, for excellent research assistance, and to Tom Lee and the
Fordham Law Review for organizing a superb symposium.
1 Brady v. Maryland, 373 U.S. 83 (1963) imposes on prosecutors a constitutional obligation to share
exculpatory evidence with the defense.
2 Van de Kamp v. Goldstein, 555 U.S. 335 (2009).
3 The verdict in Van de Kamp similarly left Tommy Lee Goldstein with no remedy for prosecutorial
misconduct that led to his spending 24 years in prison for a crime he didn’t commit.

Electronic copy available at: http://ssrn.com/abstract=1842963

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

Thompson for the New Orleans Parish District Attorney‟s Office--Eric Dubelier,4 Jim Williams,
Gerry Deegan, and Bruce Whittaker--concealed from both Thompson and from the courts
exculpatory evidence that it was constitutionally required to produce.5 In 1994 Deegan, who was
terminally ill, confessed to assistant D.A. Michael Riehlmann that he had suppressed blood
evidence in Thompson‟s armed robbery case, and Riehlmann too joined in the conspiracy of
silence. The evidence that would exonerate Thompson was discovered, fortuitously, by his own
private investigator just before he was about to be executed in 2003. A jury awarded Thompson
fourteen million dollars for his wrongful incarceration and time on death row. Connick v.
Thompson vacates that award and leaves Thompson without a remedy for the violation of his
constitutional rights.
As Thompson himself emphasized after the verdict, it isn‟t only about the money. 6
Section 1983 serves a declaratory and deterrent function as well as a compensatory one, and
Connick v. Thompson limits access to the full panoply of Section 1983 remedies. The decision
sends a deeply unfortunate message about both the government‟s duty to prevent prosecutorial
misconduct and the Court‟s duty to acknowledge, remedy and prevent egregious harms.
Dahlia Lithwick called Connick v. Thompson “one of the meanest Supreme Court
decisions ever.”7 The opinion is “mean” not only in the sense in which she uses the word—

4

D.A.Harry Connick Senior’s third in command.
131 S. Ct. at 137-72 (Ginsburg, J., dissenting).
6 His actual words were: “I don’t care about the money.” He went on to say that what he did care about was
why there have been no repercussions of any kind for the prosecutors involved in his case. In addition, he
pointed out that “of the six men one of my prosecutors got sentenced to death, five eventually had their
convictions reversed because of prosecutorial misconduct.” And the misconduct came to light because these
were capital cases and lawyers were appointed on appeal. “There are more than 4000 people serving life
without parole in Louisiana, almost none of whom have lawyers after their convictions are final.” John
Thompson, The Prosecution Rests, but I Can’t, The New York Times, April 10, 2011.
7 Dahlia Lithwick, Cruel but Not Unusual, Slate, Apr. 1, 2011, http://www.slate.com/id/2290036/.
5

Electronic copy available at: http://ssrn.com/abstract=1842963

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

coldhearted and without acknowledgement of the human costs of the government‟s
wrongdoing8-- but in its grudging interpretation of the constitutional violation at issue, its
reductionist notions of what a training regime can accomplish, and its stark indifference to the
deterrent, compensatory and declaratory aims of Section 1983. Connick reveals the relentlessly
atomistic lens through which the current Court views governmental obligations—both those of
the prosecutor and those of the Court itself.
This article will focus on the atomization of official conduct in Connick: how it is
accomplished, and at what costs to the aims of Section 1983 and governmental accountability.
First, it will challenge the central assumption on which Justice Thomas relied in vacating the
opinion below, that Thompson‟s harm can be traced to only a single incident of governmental
misconduct by a lone miscreant. To the contrary, the violation at issue was a group effort, as well
as a reflection of a longstanding culture of disregard for Brady in Connick‟s office. Second, the
article will critique the Court‟s conception of training. It will argue that the majority
misconceives the nature and purpose of training prosecutors and that there are characteristics that
inhere in prosecutorial culture in particular and organizational culture more generally that make
training essential. And finally, it will argue that the Court‟s atomistic vision of Section 1983 and
of its own role in remedying constitutional wrongs is at odds with the aims of the statute it sets
out to construe.

8

See also Failure of Empathy and Justice, The New York Times at A26, editorial, April 1, 2011.

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

How Does a D.A.’s Office Cause a Deprivation of Rights?
Connick v Thompson grapples with the question that has vexed the Court in municipal
liability cases since it first permitted Section 1983 suits against governmental entities:9 what does
it mean for wrongdoing to be attributable to—or caused by-- a municipality?10 Section 1983
provides a remedy when the defendant “subjected” the plaintiff, or “caused [the plaintiff] to be
subjected to a deprivation of constitutional rights.” Since agencies can act only through their
agents, the Monell line of cases requires a determination of whether what happened to Thompson
was attributable to the official capacity acts of the DA‟s office, or to the independent acts of one
or two or five bad apples who deviated from official policy in a way that could not have been
reasonably foreseen or prevented.
Since the decision in Monell, 11 the Court has struggled to draw the line between the
respondeat superior liability that it has held the statute prohibits and the supervisory liability it
has held the statute permits. It has been especially wary of imposing liability on the entity based
on a claim that a wrongdoer„s acts are attributable to something the entity failed to do—train,
supervise or discipline subordinates. Yet failure to train liability is essential—without it,
municipal liability is in danger of becoming a mere form of words. Unsurprisingly, explicitly
illegal policies are rarely put in place. An insistence that liability flows only from an explicit
policy essentially immunizes policymakers who simply adopt a facially constitutional policy, or
institute no policy at all, and then fail to prevent or implicitly condone unconstitutional conduct.

9

Monell v. Dep’t of Soc. Servs. of N.Y., 436 U.S. 658 (1978).
Pembaur v. City of Cincinnati, 475 U.S. 469, 477-79 (1986); City of Oklahoma City v. Tuttle, 471 U.S. 808, 823-24
(1985) (plurality opinion); Monell v. Dep’t of Soc. Servs. of N.Y., 436 U.S. 658, 694 (1978).
11
Monell v. Dep’t of Soc. Servs. of N.Y., 436 U.S. 658, 691-94 (1978).
10

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

To guard against the slippery slope scenario it fears, in which every wrongful act of a
subordinate can be linked to some failure of the policymaker, the Court has established high
hurdles to failure to train liability. It requires the plaintiff to show that the need for the training
was obvious, so that failing to train under the circumstances demonstrated deliberate indifference
to the rights of those with whom the untrained employees come into contact, 12 and that the
failure to adequately train was the moving force behind the resulting violation of rights.13 Yet
prior to Connick, failure to train liability could under certain circumstances be established
through proof of the actions or inactions of the policymaker himself, even absent a pattern of low
level violations of rights.14 Alternatively failure to train liability could be proved through
custom—if enough violations by subordinates occur on the policymaker‟s watch, he is on
constructive notice that whatever he is doing is not working properly. The former avenue—proof
of actions or inactions by the policymaker himself—is now narrowed, perhaps to the vanishing
point.
Substantial confusion has been caused by the question of the probative value of a single
incident of wrongful conduct. Much of this confusion arises from a failure to distinguish single
decisions by policymaking officials (which can lead to liability) from single wrongful actions by
subordinates (which, standing alone, cannot lead to liability). In Pembaur v. City of Cincinnati, 15
the Court held that a single decision by a policymaking official could give rise to municipal

12

City of Canton v Harris, 489 U.S.378 at 388 (1989).
Bd. of County Comm’rs v. Brown, 520 U.S. 397, 404 & 407 (1997); City of Canton v. Harris, 489 U.S. 378, 388-90
(1989).
14
Bd. of County Comm’rs v. Brown, 520 U.S. 397, 404 & 407 (1997); City of Canton v. Harris, 489 U.S. 378, 388-90
(1989).
15
Pembaur v. City of Cincinnati, 475 U.S. 469, 480 (1986).
13

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

liability, since it establishes the requisite causal link between the decision and the violation. 16 In
Oklahoma v Tuttle, the Court declined to infer inadequate policies from a single incident by a
low level employee—an unjustified fatal shooting-- without independent evidence of failures at
the policy level that may have led to that incident. 17 The Tuttle Court held that a policy of failure
to train must be shown to have resulted from the deliberate choice of an inadequate training
program by policymakers. But it recognized that if proof of such a policy of inadequate training
did exist, and an affirmative link between that policy and an unconstitutional deprivation of
rights could be shown, the policy did not need to lead to more than one unconstitutional
deprivation to be actionable.18
In Canton v Harris, the Court again emphasized that the key to municipal liability is the
causal link between some action of the policymaker and the resulting injury. It held that a
municipality may cause a deprivation by providing inadequate training in a situation in which the
need for better training is so obvious, and the inadequacy so likely to result in the violation of
constitutional rights, that the policymaker‟s failure to provide that training amounts to deliberate
indifference. 19 In a footnote, it provided an example of such a situation: a police department that
arms its officers and then fails to train them on the constitutional limits on their use of deadly
force when apprehending fleeing felons.20 Canton, like Tuttle, made it clear that had there been
independent proof of a failure to train of the sort that was likely to lead to the deprivation at
issue, the fact that only one such deprivation resulted would not prevent a finding of municipal
liability.
16

Pembaur v. City of Cincinnati, 475 U.S. 469, 480 (1986).
City of Oklahoma City v. Tuttle, 471 U.S. 808, 823-24 (1985) (plurality opinion).
18
Id.
19
City of Canton v. Harris, 489 U.S. 378, 390 (1989).
20
City of Canton v. Harris, 489 U.S. 378, 390 n.10 (1989).
17

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

In sum, municipal liability for failure to train could, prior to Connick, be established in
two ways: either through evidence that the policymaker had a policy of failure to train, or
through evidence of a pattern of violations by subordinates, from which it can be inferred that the
policymaker should have known of the need for more effective training. Prior to Connick, it had
never been the rule that a policymaking official was allowed a few “free” violations, or even one
free violation, before he could be held liable. The Court recognized that in some cases the need
for training is so obvious, and the lack of training so likely to create a constitutional violation, 21
that a failure to provide such training could establish the policymaker‟s deliberate indifference
even if it lead to only a single claim of constitutional harm, as in the fleeing felon situation
described in Canton.
Connick argued that a Brady violation is not akin to the fleeing felon situation described
in Canton—that it is not sufficiently obvious that the failure to train prosecutors about their
Brady obligations will lead to Brady violations. He argued that therefore municipal liability for
failure to train prosecutors about their discovery responsibilities should never lie for what he
called a “single incident” (more on that incorrect characterization shortly); that liability cannot
lie unless there is a pattern of Brady violations. The Thomas majority enthusiastically embraced
this argument. It held that no amount of independent proof of deliberate indifference to the need
to train prosecutors about their Brady violations can suffice in the absence of proof of a pattern
of constitutional violations obvious enough22 to come to the policymaker‟s notice. 23
21

Thompson v. Connick, 578 F.3d 293 (5th Cir. 2009) (Pardo, J. opinion in en banc case) Thompson v. Connick, 578
F.3d 293, 314 (5th Cir. 2009) (per curiam) (Pardo, J., writing separately), rev’d, 131 S. Ct. 1350 (2011).
22
It does not specify how these violations would come to the policymaker‟s notice. This is a particularly troubling
question in a situation where the harm is a failure to turn over evidence; something that by its nature is difficult to
discover, particularly when no monitoring system is required to be put in place. Imbler v. Pachtman, 424 U.S. 409,
443-44 (1976) (White, J., concurring).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

In this case there was substantial independent evidence that Connick not only should
have been but was in fact on notice that he was providing inadequate training on a core
prosecutorial function. Essentially he provided no formal training, no formal supervision, and a
clearly communicated policy of presumptive non-disclosure unless required by law, coupled with
a lack of guidance, or in some cases misinformation, on what was required by law.
Unsurprisingly, there was substantial evidence that this approach led to persistent “misperception
and disregard of Brady‟s disclosure requirements.”24
In dissent, Justice Ginsburg observed:
Unquestionably, a municipality that leaves police officers untrained in constitutional
limits on the use of deadly weapons places lives in jeopardy. But as this case so vividly
shows, a municipality that empowers prosecutors to press for a death sentence without
ensuring that those prosecutors know and honor Brady rights may be no less “deliberately
indifferent” to the risk to innocent lives.25

23

The Court is on the brink of holding that except in the very narrow situation described in Canton v Harris—
arming police without training them on deadly force—municipal policymakers are insulated from liability for
failure to train until that failure can be causally linked to a pattern of unconstitutional conduct obvious
enough to come to their actual or constructive notice. The reasoning of Connick v Thompson—that
prosecutors can train and regulate themselves-- certainly seems to insulate prosecutors’ offices from entity
liability for failure to train on any constitutional violation, since the argument for the need to train assistant
prosecutors on their Brady obligations is particularly strong. Whether the reasoning insulates entities other
than prosecutors’ offices from failure to train liability absent a pattern of misconduct is a more difficult issue.
Arguably the opinion’s reasoning relies heavily on the ability of prosecutors, as legal professionals, to train
and regulate themselves. However, there is language in both the majority and concurring opinions to suggest
that the Canton “failure to train on deadly force” example is no longer to be considered an illustration of the
type of situation in which failure to train liability can lie based on deliberate indifference absent a pattern of
unconstitutional subordinate conduct, and is now to be regarded as the sole situation in which such liability is
available. See e.g. Connick, 131 S. Ct. at 1367 (“Were Thompson’s theory the law, there would have been no
need for Canton’s footnote to confine its hypothetical to the extreme circumstance of arming police officers
without [training them about the limitations upon shooting fleeing felons]”) (emphasis supplied).
24
131 U.S. at 1385 (Ginsburg, J., dissenting).
25
131 S. Ct. at 1385 (Ginsburg, J., dissenting).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

The Lone Miscreant Rides Again: Crafting a “Single Incident” Narrative
There weren’t four instances. There was one Brady violation that possibly could have involved
one to four prosecutors.26

The Connick opinion holds that the district attorney‟s office may not be held liable under
Sec. 1983 for failure to train its prosecutors based on a single Brady violation. 27 It is important
to take a step back at this juncture and ask: In what sense is this series of acts by at least five
prosecutors28 over a period of more than eighteen years a single incident? Not in the sense that it
was a rogue, unforeseeable act by a subordinate that could not fairly be ascribed to the actions of
the policymaker. Only in the sense that it culminated in the wrongful conviction and near
execution of only a single man.
When judges are faced with allegations of governmental misconduct, they must make
choices about what factors are relevant or important, what causal links exist between various
acts, and whether to categorize incidents as isolated or part of a pattern. 29 Judges make these
choices in light of assumptions about the underlying constitutional and statutory values at issue,
and more generally, about how various types of actors ought to behave and how the world
works.30 As I‟ve argued elsewhere, courts display a persistent tendency to portray governmental
misconduct as isolated rather than part of a larger pattern.

26
27

Counsel for Connick at oral argument in the Supreme Court.

131 U.S. at 1362.
131 S. Ct. at 1384 (Ginsburg, J., dissenting). (Four prosecutors worked together on prosecuting Thompson, and a
fifth was the recipient of Gerry Deegan’s deathbed confession).
29
Susan Bandes, Patterns of Injustice: Police Brutality in the Courts, 47 Buff. L. Rev. 1275 at 1275 (1999).
30
I have elsewhere discussed in detail many of the assumptions that underlie the courts’ tendency to disaggregate
governmental misconduct. These include selective empathy, in this case a strong identification with the forces of
law and order, the assumption that the status quo is just and fair, fear of the chaos that will ensue if liability is
28

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

The story of a few bad apples in an otherwise pristine barrel is both comforting and
seductive. It conforms to standard narrative conventions of good and evil actors, a focus on
action rather than inaction, easy causal links, a clear moral, and a simple and satisfying
dénouement. When governmental wrongdoing is at issue, the story also serves to reassure that
the world is just and that the forces of crime and chaos will be reliably reined in by a benign
government. The Connick majority and concurring opinions dramatically illustrate how judges
portray wrongful conduct as the fault of one or two malevolent actors rather than as the product
of systemic dysfunction.31 In an office described as having “one of the worst Brady records in
the country,”32 one bad apple was held entirely to blame. The lone “miscreant,”33 Gerry Deegan,
died in 1994, and in any case would have been absolutely immune from suit under Van de
Kamp. Thus no one is held accountable for withholding evidence that kept an innocent man in
prison for 18 years and nearly led to his wrongful execution. Moreover, for the Court, no
systemic changes appear to be in order.
How did this egregious wrongdoing, in which several assistant prosecutors participated,
in an office notorious for its discovery abuses both by the policymaker and by his subordinates,
end up on the shoulders of one man, long dead? As I will describe, for the Court to place the
blame solely on Deegan‟s shoulders and to exonerate Connick‟s office required it to disaggregate
a complex pattern of official misconduct at every conceivable juncture. Illustrating the
available, and a preference for individual stories of motive, fault and blame over complex stories of a series of
interlocking decisions, often made in good faith. Susan Bandes, Patterns of Injustice: Police Brutality in the Courts,
47 Buff. L. Rev. 1275, 1317-40 (1999). See also Susan Bandes, Tracing the Pattern of No Pattern: Stories of Police
Brutality, 34 Loyola Los Angeles L. Rev. 665 (2001).
31
Susan Bandes, Patterns of Injustice: Police Brutality in the Courts, 47 Buff. L. Rev. 1275, 1330 (1999).
32
Connick v. Thompson, 131 S. Ct. 1350, 1384 (2011) (Ginsburg, J., dissenting).
33
131 S. Ct. at 1368 (Scalia, J., concurring)” “The withholding of evidence in *Thompson’s] case was almost
certainly caused…by miscreant prosecutor Gerry Deegan’s willful suppression of evidence he believed to be
exculpatory…)

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

mechanics of disaggregation in this case requires a quick overview of the facts. Thompson was
convicted of both an armed robbery and a murder he did not commit. He served eighteen years in
prison, fourteen of them on death row, and narrowly escaped execution. As discussed earlier,
four assistant prosecutors worked together on the Thompson prosecutions and deliberately
withheld and concealed blood evidence that would have exonerated Thompson in his armed
robbery case--an undisputed violation of Brady v. Maryland. 34 The robbery conviction led, by
design, to Thompson‟s inability to take the stand in his own defense in the murder case, allowing
the actual perpetrator‟s testimony to stand uncontradicted. The office had deliberately tried
Thompson first for the robbery, though it occurred after the murder, in order to disable him from
testifying. The armed robbery also provided the statutory aggravating factor that made the
murder death-eligible and ultimately put Thompson on death row. The prosecution argued in
closing at Thompson‟s murder trial that “because he was already serving a near-life sentence for
the armed robbery, the only way to punish him for murder was to execute him.” 35 Thompson
was sentenced to death.
Over the eighteen years that Thompson was imprisoned, the D.A.‟s office never turned
over the Brady material. As discussed above, it was ultimately discovered by one of Thompson‟s
own investigators less than a month before Thompson was to be executed. After this fortuitous
discovery, it came to light that in 1994, nine years after Thompson‟s trial and five years before
his execution date, assistant D.A. Gerry Deegan had confessed on his deathbed to another
assistant D.A., Michael Riehlmann, that he had (according to Riehlmann‟s later report)
34

But see Justice Scalia’s concurrence, arguing that prosecutors had no duty to turn over blood evidence because
they did not know (having determined that they would not inquire into) the suspect’s blood type. 131 S. Ct. at
1368-69. Justice Ginsburg in dissent refers to this as a “don’t ask, don’t tell” view of Brady obligations that garners
no support from precedent. 131 S. Ct. at 1350.
35
131 S. Ct. at (Ginsburg, J., dissenting).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

intentionally withheld the blood evidence. Riehlmann kept this knowledge to himself until the
blood evidence was uncovered by Thompson‟s investigator in 1999.
Thus there is evidence that five or more attorneys had been part of the failure to turn over
evidence or had a hand in covering it up.36 The D.A.‟s office vacated the robbery conviction in
light of the exposure of the exculpatory blood evidence, but chose, after Thompson‟s murder
conviction was reversed, to retry him. In Thompson‟s retrial not only did he testify in his own
defense, he gained access to thirteen pieces of evidence that Connick‟s office failed to turn over
during his first trial, including information that an eyewitness had received a monetary award
and a police report in which the main eyewitness had given a description inconsistent with his
trial testimony, both essential tools for impeaching crucial inculpatory testimony. 37 At retrial
the jury acquitted Thompson in thirty-five minutes. In his subsequent Sec. 1983 suit, a jury found
that Connick‟s failure to train his assistants about their Brady obligations caused the deprivation
of Thompson‟s rights. It awarded Thompson fourteen million dollars, a verdict upheld by the
Fifth Circuit and left in place by an equally divided en banc court. Connick v Thompson
overturned that award, leaving Thompson with no remedy for his constitutional injury.
The operative question is whether Connick should have been on notice that his office had
a problem with Brady training and compliance that was likely to lead to Brady violations if
unaddressed. To find that no such constructive notice existed, the Court portrayed a complex and
persistent pattern of misconduct as not merely a series of isolated, disconnected acts of
misconduct, but a single act of misconduct. First, the Court implied that only prior judicial
36

Connick v. Thompson, 131 S. Ct. 1350, 1384 (2011) (Ginsburg, J., dissenting).
Thompson v. Connick, 553 F.3d 836, 845-46 (5th Cir. 2008), vacated, 578 F.3d 293 (5th Cir. 2009) (en banc), rev’d,
131 S. Ct. 1350 (2011). See also Connick v. Thompson, 131 S. Ct. 1350, 1374-76 & 1376 n.10 (2011) (Ginsburg, J.,
dissenting).
37

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

reversals on Brady grounds can provide notice of prior noncompliance with Brady. It cited no
authority and no rationale for confining itself to judicial reversals while ignoring multiple other
avenues for discovering office non-compliance or the need for training of subordinates.
Demanding judicial reversals sets a bar that is not only unprecedented but onerously high in
Brady cases. Even when prosecutors are caught hiding evidence, courts will reserve a conviction
only if the evidence was so strong that its disclosure would have created a reasonable probability
of a different verdict. “And catching prosecutors who have engaged in such deception can be
extremely difficult.”38 The Chicago Tribune described a number of cases in which evidence
came to light years after a conviction, and by happenstance. In one such case, “evidence
undermining the state‟s case surfaced only after being stolen from a prosecutor‟s office by a man
dating the prosecutor‟s secretary.” 39 The wrongly convicted man had at that point served 21
years in prison. “Evidence has surfaced in other cases only after a judge directed the U.S.
marshal to seize the prosecutors‟ documents, or because newspapers sued under the Freedom of
Information Act, or because of anonymous tips, conversations accidently overheard or papers
spied in a prosecutor‟s hand.”40

38

Ken Armstrong & Maurice Possley, Trial & Error, Chicago Tribune, Jan. 11-14, 1999,
http://www.chicagotribune.com/news/nationworld/chi-dptrialerror-special,0,632955.special See also
Robert P. Mosteller, Failures of the Prosecutor’s Duty to “do Justice” in Extraordinary and Ordinary
Miscarriages of Justice at 13, in THE PROSECUTOR IN TRANSNATIONAL PERSPECTIVE (Erik Luna & Marianne Wade
eds., Oxford University Press forthcoming 2011): “without broad disclosure requirements, the extent of
injustice will remain hidden, unaddressed, and without correction. “ As one amicus brief points out, the
barriers to discovery would be even higher in situations where an inmate who suspected a Brady violation
was unrepresented by counsel. Brief of Amicus Curiae the National Association of Criminal Defense Lawyers
in Support of Respondent in Connick v. Thompson.
39
40

Id.
Id

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

In Connick‟s office, there were in fact multiple other warnings that the office was out of
compliance in ways that were leading to predictable Brady violations—warnings and
condemnations that it received and ignored repeatedly.41 But even by the Court‟s own measure,
there were significant warnings. By the time of Thompson‟s trial in 1994, Connick‟s eleven year
regime had already led to four other judicial reversals on Brady grounds—four other cases in
which the failure to turn over Brady evidence came to light and was serious enough to warrant
reversal. 42
But the Court proceeded to set the bar higher still, rejecting the significance of these four
reversals because they involved the failure to turn over different types of evidence—for example
an arrest report rather than a crime lab report; a report about a weapon rather than blood
evidence. Yet this distinction based on types of evidence rather than the scope of obligations
does not track the usual categories of Brady training43 or address the sorts of misconceptions
about Brady that generally arise 44 or that are at issue in this case.45

41

See e.g. Armstrong and Possley, supra note 39, noting that Connick’s office was condemned repeatedly for
withholding evidence, and detailing specific warnings to the office to change its behavior in this regard. See
Connick, 131 S. Ct. at 1382 (Ginsburg dissenting) noting that after the Supreme Court decision in Kyles v. Whitley
featuring many instances of the state’s failure to discover exculpatory evidence Connick stated that he saw no
need to make any changes. See also 131 S. Ct. at 1376, describing Connick’s decision to abort grand jury
proceedings against the prosecutors who had withheld the lab report in Thompson’s case after one day because
the grand jury “would make his job more difficult.” See also 131 S. Ct. at 1387 noting that Connick himself had
previously been indicted for suppression of evidence. See also Brief of the Innocence Network as Amicus Curiae in
Support of Respondent in Connick v. Thompson, which stated that of the 36 capital convictions during Connick’s
tenure, more than half of those convicted subsequently asserted that Connick’s office withheld Brady material.
Courts found that evidence had been improperly withheld in 9 of those 19 cases.
42
131 S. Ct. at 1360.
43
See e.g. http://www.sdsheriff.net/legalupdates/docs/bradytrainingbulletin.pdf (visited on April 22, 2011) setting
forth Brady obligations by categories of favorable evidence included in Brady, such as “evidence directly opposing
guilt,” “evidence indirectly opposing guilt,” “evidence supporting defense testimony,” “evidence supporting a
defense motion that would weaken the prosecution’s case” and “evidence impeaching a prosecution witness’s
credibility” rather than by type of document (crime lab report) or type of evidence (blood evidence).
44
See e.g. amicus brief of National Association of Criminal Defense Lawyers in Support of Respondent in Connick v.
Thompson, detailing recurrent issues that include whether impeachment evidence is exculpatory, whether

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

The majority and concurrence applied a similar methodology to the series of decisions by
several prosecutors in regard to this particular case—the decisions of the attorneys who tried the
case to withhold Brady evidence and to deliberately avoid mentioning that evidence at trial; the
decision of attorney Reihlman to keep quiet after heading a deathbed confession regarding
Thompson, even as Thompson sat on death row facing imminent execution. That several
prosecutors not only failed to turn over the evidence but covered up the failure to do suggests an
office culture that breeds noncompliance with Brady. Justice Thomas pared all this information
down to a single incident, insufficient to provide notice to the policymaker, through several
methods of disaggregation. He dismissed every action subsequent to the initial failure to turn
over the evidence—in other words the entire cover-up that kept Thompson locked up for
eighteen years-- as “contemporaneous or subsequent conduct [that] cannot establish a pattern”
because it provides no notice to the entity and no opportunity for the entity to conform to
constitutional dictates.46 But of course Thompson did not complain only about his initial
conviction. He complained about the entire course of conduct that kept him imprisoned for
eighteen years, on death row for fourteen, and facing execution dates on five occasions. As
Justice Ginsburg said, the prosecutors hid material they were constitutionally obligated to turn

evidence must be disclosed, whether evidence that is not exculpatory by itself but might exonerate the defendant
if combined with other evidence known by defense or ascertainable by the prosecution must be disclosed, how
materiality is measured, and whether materiality is judged cumulatively or based on each individual piece of
evidence.
45
For example, there was confusion in the office about whether evidence impeaching credibility was governed by
Brady, and in addition Connick himself incorrectly believed that inadvertent conduct is excusable under Brady. 131
S. Ct. at 1350.
46
Connick at 10 n7 Connick v. Thompson, 131 S. Ct. 1350, 1360 n.7 (2011) (citing City of Canton v. Harris, 489 U.S.
378, 395 (1989) (O’Connor, J., concurring in part and dissenting in part)).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

over, not only before or during the trial, but well after that. They did so “despite multiple
opportunities, spanning nearly two decades, to set the record straight.”47
The Court dismissed the four previous Brady violations during Connick‟s tenure (which
included a high profile Supreme Court reversal of the work of Jim Williams, also one of the
prosecutors in this case). It held that any pattern of violations must be a pattern of failure to train
about this particular type of Brady evidence, or about “the specific scenario related to the
violation at issue.”48 It dismissed the failure to turn over, prior to Thompson‟s murder trial,
thirteen additional pieces of evidence evidence which once in his possession helped win
Thompson acquittal at his murder retrial. It classified those thirteen pieces of evidence as
irrelevant because there was no explicit finding that the evidence was Brady material. 49 These
are, in the light most favorable to the majority, interpretive choices that skew relentlessly in one
direction.
But the final and most devastating choice lies buried in a footnote and is hard to defend
under any interpretive criteria. In this footnote the Court refused to consider the argument that
Connick created a culture of indifference in the district attorney‟s office, stating “This argument
is essentially an assertion that Connick‟s office had an unconstitutional policy or custom. The
jury rejected this claim, and Thompson does not challenge that finding.”50 The majority‟s
assertion is a serious misstatement of both the record and the law.51 As the Court has consistently
47

Connick, 131 S. Ct. at 1370.
Connick v. Thompson, 131 S. Ct. 1350, 1363 (2011).
49
Connick v. Thompson, 131 S. Ct. 1350, 1364 n.11 (2011). But see 131 S. Ct. at 1374 n11 (Ginsburg, J., dissenting)
(explaining lack of explicit finding that failure to turn over these additional pieces of evidence violated Brady).
50
Connick v. Thompson, 131 S. Ct. 1350, 1364 n.10 (2011).
51
The instructions permitted the jury to find a policy on two separate grounds, as an “official policy,” or through a
failure, through deliberate indifference, to establish policies and procedures to protect one accused of a crime
from these constitutional violations.” Thompson v. Connick, 553 F.3d 836, 847 (5th Cir. 2008), vacated, 578 F.3d
293 (5th Cir. 2009) (en banc), rev’d, 131 S. Ct. 1350 (2011). The first ground appears to ask the jury to find an
48

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

held from Monell onward, there are only two avenues for finding municipal liability, a policy
emanating from the policymaker, which includes a policy of failure to train, or a custom shown
by a pattern of subordinate misconduct. The jury found that the entity had no explicit policy on
Brady compliance, but it found that Connick‟s office had a policy of failure to adequately train,
within the meaning of the governing precedents: Canton v. Harris and Bryan County v. Brown.52
Thus there is no defensible argument for excluding evidence that Connick‟s office created a
culture of indifference—a culture that created and illustrated the need for training and
supervision.

Erasing the Entity: The Myth of the Self-Training, Self-Regulating Prosecutor

The Connick court atomized not only the prosecutorial misconduct at issue but also the
prosecutor‟s office as an entity. In the office depicted in the majority and concurring opinions,
each assistant prosecutor is an island entire unto himself. 53 The Court ascribed each incident of
misconduct solely to the individual, and assumed that each prosecutor is responsible for his own
training and regulation. It repeatedly erased and ignored the dynamics of the governmental entity
and the role of the entity in setting, communicating and enforcing standards of conduct. The
Connick opinion rejected the need for prosecutorial training based on assumptions about the
ability of assistant prosecutors to train themselves in the law and adhere to ethical rules—
assumptions that are most generously described as wishful thinking.
th

explicit policy, and the 5 Circuit assumed that this is how the jury interpreted it. See Thompson v. Connick, 553
F.3d 836, 851 (5th Cir. 2008), vacated, 578 F.3d 293 (5th Cir. 2009) (en banc), rev’d, 131 S. Ct. 1350 (2011).
52
See e.g. Canton v Harris, 489 U.S. 378, 390 (1989) (Only where a failure to train reflects a “deliberate’ or
“conscious” choice by a municipality—a “policy” as defined by our prior cases—can a city be liable for such a
failure under Sec. 1983).
53
John Donne, Meditation XVII, “No man is an island entire unto itself.”

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

The Court‟s faith in the self- regulating prosecutor is contradicted by ample evidence of
prosecutorial misconduct, some of it willful, but much of it unintentional. 54 Moreover the Court‟s
portrayal of the prosecutor as autonomous, self directing agent ignores the substantial and
growing body of knowledge about how bureaucratic structure influences ethical decision
making.55 This body of knowledge, coupled with a rich literature about the particular ethical
challenges of the prosecutor‟s office and how they are best addressed, illuminates the
tremendous importance of training and the misguided nature of the Court‟s approach.
The Court holds that absent a pattern of violations, “a district attorney is entitled to rely
on prosecutors‟ professional training and ethical obligations.” 56 In a vote of confidence for legal
education that should strike fear in the heart of every law professor, the Court held that whereas
police might need some training before they are sent out into the street with guns, prosecutors
learn what they need to know in law school. Even if they don‟t actually learn about Brady in law
school (and the vast majority do not, since criminal procedure is not usually a required course,
and the introductory criminal procedure course does not usually cover Brady in any case), they
obtain the tools they need to learn about Brady, on their own, or with the help of CLE courses,
when the time comes. And to the extent they have an ethical obligation to turn over Brady
evidence, the Court is reassured that all lawyers have also satisfied character and fitness
standards, that those who violate ethics rules are subject to discipline, and that prosecutors have a
special ethical obligation to do justice as well as to seek conviction.
The air of unreality that pervades the majority‟s descriptions of the prosecutorial function
is palpable. As a foundational matter, it is bizarre to assume that a professional degree obviates
54

See supra text accompanying notes 61-65.
See supra text accompanying notes 60-82.
56
131 S. Ct. at 1361.
55

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

the need for specialized training. 57 But more specifically, there is ample evidence that the honor
system the Court advocates has not proved a viable option for the training and discipline of
prosecutors. It is well documented that prosecutorial misconduct, prominently including the
violation of Brady obligations, is not only all too common, but a leading cause of wrongful
convictions. 58 It is also well documented that discipline of prosecutors is rare almost to the
vanishing point, so discipline cannot be relied upon as a deterrent to or remedy for misconduct. 59
When institutional dynamics are made part of the equation, the need for training becomes
clear. Norms of ethical conduct, for prosecutors or other professionals, are not inert precepts
transmitted in a vacuum. They are understood, given shape and refined in institutional and social
contexts, through both implicit and explicit means, through official choices to act and official
choices to refrain from action. 60 Institutions are uniquely placed to shape and guide moral
intuitions—for better or for worse. They provide systematic feedback through the provision of

57

See Brief of Amicus Curiae the National Association of Criminal Defense Lawyers in Support of Respondent in
Connick v Thompson, analogizing this argument to taking a general medical practitioner with no training in surgery
and placing him in a clinic, expecting him to rely on attendance at medical school, on the job training, and
professional responsibility to provide competent care.
58
See Daniel S. Medwed, Emotionally Charged: The Prosecutorial Charging Decision and the Innocence Revolution,
31 Cardozo L. Rev. 2187, 2187 (2010). See generally Ken Armstrong & Maurice Possley, Trial & Error, Chicago
Tribune, Jan. 11-14, 1999, http://www.chicagotribune.com/news/nationworld/chi-dptrialerrorspecial,0,632955.special. See also Erwin Chemerinsky, Head in the sand over prosecutorial misconduct, the
National Law Journal, April 25, 2011 (reporting on a recent study by the Northern California Innocence Project at
Santa Clara University Law School documenting frequency of misconduct) and Margaret Z. Johns, Reconsidering
Absolute Prosecutorial Immunity, 2005 Brigham Young University Law Review 53, 59-64 (discussing several major
studies detailing frequency of prejudicial prosecutorial misconduct).
59
See Armstrong and Possley, supra note 39, discussing rarity with which prosecutors are disciplined. See also
Margaret Johns, Reconsidering Absolute Prosecutorial Immunity, 2005 BYU L. Rev. 53, 71 (discussing the lack of
safeguards against or consequences for prosecutorial misconduct). See also Brief of Amicus Curiae National
Association of Criminal Defense Lawyers (citing New York and California task forces which concluded that
“prosecutorial misconduct is a substantial cause of wrongful convictions, errant prosecutors are virtually never
disciplined, and the widespread lack of discipline causes such misconduct to occur.”) See generally Fred C.
Zacharias, The Professional Discipline of Prosecutors, 79 U.N.C. L. Rev. 721 (2001) (discussing why discipline of
prosecutors is so rare).
60
See generally Susan Bandes, The Negative Constitution: A Critique, 88 Mich. L. Rev. 2271 (1990) (critiquing use of
action/inaction distinction in determining governmental accountability).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

incentives and disincentives that convey the norms of acceptable and unacceptable behavior. 61
Thus it is essential to look to the implicit and explicit norms of the institution and determine
what sorts of ethical rules are being transmitted and reinforced.
Institutions shape ethical behavior by rewarding or punishing awareness of ethical conflict,
and by rewarding or punishing those who take the initiative to confront such conflicts. When the

workplace creates a conflict between ethical behavior and institutional values, or implicitly
condones or overlooks unethical behavior that advances institutional goals, ethical behavior
tends to lose out. The choice rarely operates on a conscious level. “Left to our own devices,
without feedback or correction, we are likely to believe what is self-protective or self-deceptive.
We are not particularly good at identifying and correcting our own assumptions, biases and blind
spots…The problem is not only that we articulate self-protective and self-defensive excuses; the
problem is that we have every incentive to believe them.” 62 People reconfigure their notions of
what counts as ethical so they do not have to confront the tension between doing good and doing
well, or the acute discomfort of regarding themselves as unethical. 63 Institutions that encourage
employees to avert their eyes from questionable behavior, or to place the protection of the entity
above the observance of ethical obligations, exacerbate this tendency toward self-protection and
supports self-deception.64 Psychologist Gerg Gigerenzer calls these “split brain institutions.” 65 For

61

Gerg Gigerenzer, Gut Feelings: The Intelligence of the Unconscious at 157-58 (2007).
Susan A. Bandes, Is it Immoral to Punish the Heedless and Clueless?, 29 Law and Philosophy 433, 445 (2010).
63
Susan Bandes, Loyalty to One’s Convictions: The Prosecutor and Tunnel Vision, 49 How. L.J. 475, 487 (2006).
64
Gerg Gigerenzer, Gut Feelings: The Intelligence of the Unconscious at 198 (2007).
65 Id.
62

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

those who work in such institutions, self protective behavior gradually ceases to be viewed as
unethical and begins to look acceptable and even desirable.66
This dynamic— the convenient reconfiguring of the notion of ethical behavior—is one the
prosecutor‟s office has to guard against with particular vigilance, due to the inherent tension

between the prosecutor‟s dual roles as advocate and minister of justice. 67 When the institutional
incentives emphasize only one aspect of that dual role—the role of adversary zealously focused
on obtaining convictions-- the consequences are entirely predictable. Prosecutors are frequently
faced with hard ethical choices, and the admonition to do justice is vague enough to seem to
justify, or at least excuse, a wide range of actions. After a while a choice may appear less as an
ethical quandary and more as a justifiable action to help victims, to keep bad people off the
street, and to “protect the reputation of the agency itself so that it can continue to do its important
work.”68 Such all too human tendencies to do what is expected and what is rewarded can be
counteracted, or they can be exacerbated, by the norms and expectations of the institution. In
some cases these norms are transmitted, at least in theory, by written policies or legal constraints.
More often, administrative norms are clearly communicated through less traceable
channels69…through the behavior of…colleagues and supervisors, through observing how things

See Max H. Bazerman and Ann E. Tenbrunsel, Stumbling Into Bad Behavior, The New York Times at A21,
April 21, 2011, discussing the “ethical fading” that enables people to behave unethically and also overlook the
unethical behavior of others while maintaining a positive self image in situations in which there is a tendency
toward “motivated blindness” about information that works against self interest or the interest of the
employer.
67 See e.g. Fred C. Zacharias, Structuring the Ethics of Prosecutorial Trial Practice: Can Prosecutors do Justice?
44 Van. L. Rev.45, 46 (1991).
68 Susan Bandes, Loyalty to One’s Convictions: The Prosecutor and Tunnel Vision, 49 How. L.J. 475, 487 (2006).
69 The failure to promulgate specific policies protects policymaking official and keeps responsibility and
blame at low levels. It perpetuates the appearance that street level officers are making autonomous,
disconnected decisions. Bandes, Patterns of Injustice, supra note 30 at 1329.
66

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

are done, what is rewarded, what is punished, and what is ignored. 70 And indeed it is a
commonplace that obtaining convictions tends to be the key to prosecutorial advancement. As
Daniel Medwed put it, “A series of factors cause trial prosecutors to view their jobs primarily
through the lens of gaining “wins” (convictions) and avoiding “losses” (acquittals).” 71
Prosecutors may violate Brady intentionally or unintentionally. Either type of violation is
deterrable.72 Many of the problems that lead to Brady violations and other constitutional
infringements arise, not from identifiable individual intentions, but from incentive structures
deeply imbedded in the culture of the office—incentive structures that exacerbate existing
tendencies toward self-protection and self-deception.73 Recent research on cognitive processing
helps explain how police and prosecutors can take (or fail to take) actions that violate rights and
lead to wrongful convictions without exactly “knowing” they are doing so. 74 For example, a
number of scholars have written about the problem of prosecutorial tunnel vision. Tunnel vision
can be explained as a species of cognitive bias that causes prosecutors to screen out information
70

Bandes. Patterns of Injustice, supra note 30 at 1329.
Daniel Medwed, The Prosecutor as Minister of Justice: Preaching to the Unconverted from the Post-Conviction
Pulpit, 84 Washington L. Rev. 35, 44 (2009).
72
In the exclusionary rule context, the Court has frequently assumed that negligent acts are unlikely to be
deterrable, and that intentional acts are the best candidates for deterrence. See e.g. United States v. Leon, 468
U.S. 897, 911 (1984); Herring v. United States, 129 S. Ct. 695, 702 (2009). For a critique of this view, see Susan A.
Bandes, The Roberts Court and the Future of the Exclusionary Rule, ACS Issue Brief, April 2009, available at
http://www.acslaw.org/files/Bandes%20Issue%20Brief.pdf In Connick, conversely, the “miscreant prosecutor
Gerry Deegan’s willful suppression of evidence he believed to be exculpatory” is treated as non-deterrable.
Connick, 131 S. Ct. at 1368. But as the Court seems to recognize in the exclusionary rule context, If Deegan had
foreseen any negative consequences from his intentional suppression of evidence, he may have behaved quite
differently. See Tony Mauro, Stevens Criticizes Ruling on Prosecutorial Immunity, The Blog of Legal Times, May 11,
2011, http://legaltimes.typepad.com/blt/2011/05/stevens-criticizes-ruling-on-prosecutorial-immunity.html
(reporting on Justice Stevens arguing that training and supervision can affect intentional as well as negligent
misconduct).
73
Robert Mosteller, Exculpatory Evidence, Ethics, and the Road to the Disbarment of Mike Nifong: The Critical
Importance of Full Open-File Discovery, 15 Geo. Mason L. Rev. 257, 309-311 (2008) (discussing the problems
arising from the focus on prosecutorial intent in determining discipline for Brady violations).
74
See generally Alafair S. Burke, Improving Prosecutorial Decision Making: Some Lessons of Cognitive Science, 47
William and Mary L. Rev. 1587 (2006), Barbara O'Brien, A Recipe for Bias: An Empirical Look at the Interplay
Between Institutional Incentives and Bounded Rationality in Prosecutorial Decision-Making, 74 Mo. L. Rev. 999
(2009).
71

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

that might cast doubt on the accuracy of their initial version of events. It “infects all phases of
criminal proceedings, beginning with the investigation of cases and then proceeding through the
prosecution, trial or plea-bargaining, appeal, and post-conviction stages.”75 Tunnel vision can be
a particular problem when a prosecutor must make decisions about what counts as Brady
material based on elastic concepts such as materiality, which require “weighing a single piece of
potentially exculpatory evidence against all inculpating evidence, the totality of which may seem
especially powerful in the investigative stage.”76 A prosecutor is likely to believe in the strength
of her own case and the guilt of the suspect,77 and this belief may color her judgment about the
relative importance of potentially exculpatory evidence in her file. As Robert Mosteller recently
described:
For a prosecutor who has reached the conclusion that the accused is guilty….there can be
no true exculpatory evidence. If it is truly exculpatory, the case should be dismissed, or
that thought must be seriously entertained. Otherwise, the evidence must be not really
exculpatory, and therefore, is simply useful ammunition for the defense in the adversary
battle of the criminal trial. 78
Tunnel vision and other cognitive biases may not operate on a wholly conscious level,
but that does not mean they are impervious to influence. On the contrary, even biases that are not
entirely conscious may be amenable to change when the incentive structures make change
desirable, or refusal to change undesirable. If assistant prosecutors discovered that

75

Keith A. Findley & Michael S. Scott, the Multiple Dimensions of Tunnel Vision in Criminal Cases, 2006 Wis. L. Rev.
291, 295; accord Bandes, Loyalty to One’s Convictions, supra note 69 at 481-83.
76
Amici Curiae Brief of the Center on the Administration of Criminal Law, NYU et al in the case of Connick v.
Thompson.
77
Indeed, the command to do justice suggests that the prosecutor should not proceed unless convinced on some
level of the suspect’s guilt, and thus to proceed without such a belief creates both an ethical dilemma and perhaps
some cognitive dissonance. Robert Mosteller, Exculpatory Evidence, Ethics, and the Road to the Disbarment of
Mike Nifong: The Critical Importance of Full Open-File Discovery, 15 Geo. Mason L. Rev. 257, 309 (2008, Bandes,
Loyalty to Convictions, supra note 68 at 488.
78
Robert Mosteller, Exculpatory Evidence, Ethics, and the Road to the Disbarment of Mike Nifong: The Critical
Importance of Full Open-File Discovery, 15 Geo. Mason L. Rev. 257, 309 (2008).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

unconstitutional conduct caused them to lose opportunities at promotion or salary increase, or to
lose the respect of their colleagues and the support of their superiors, they might weigh costs and
benefits differently or take additional steps to avoid misconduct. Or “if elected prosecutors found
unconstitutional conduct…subjected them to sanctions, or their offices to litigation, they might
take a hard look at the incentive structure of the office and whether it provides a meaningful
check on such misconduct.”79
In oral argument in Connick, members of the Court declared themselves perplexed by the
question of what kind of training prosecutors could have been provided that might have made a
difference.80 Instead of insisting on a precise account of the substantive content of the training
that should have been offered, the Court should have focused on a much more important point
about the absence of training in Connick‟s office. Training, not just in its content but in the very
fact that it occurs, communicates important messages about the expectations and culture of the
office. As one amicus brief argued:
Connick plainly recognized, while he was D.A., that law school graduates had to be
trained in numerous areas of responsibility before they would be qualified to handle
significant criminal cases. Indeed, Connick, in his brief, congratulates himself on the
extensive training and supervision his new prosecutors received in virtually every type of
function. There was just one area missing: Brady. That Connick would recognize the
need to instruct prosecutors in virtually every facet of prosecution, but provide no
training about Brady, was a powerful piece of evidence before the jury proving his
79

Susan A. Bandes, Framing Wrongful Convictions 2008 Utah L. Rev. 5 at 21-22, Scott E. Sundby, Fallen
Superheroes and Constitutional Mirages: The Tale of Brady v. Maryland, 33 McGeorge L. Rev. 643, 653-54 (2002).
80
Transcript of Oral Argument at 29-36 , Connick v. Thompson, 2010 WL 3907898 (U.S. Oct. 6, 2010) (No. 09-571).
The majority opinion, having declared that only a lack training on the specific type of claim or scenario at issue
should matter, Connick v. Thompson, 131 S. Ct. 1350, 1363 (2011), went on to say that it would be inappropriate
for the Court to micromanage prosecutors’ offices by telling them exactly what to cover in their training regimes.
Connick v. Thompson, 131 S. Ct. 1350, 1363 (2011). But the specter of micromanagement is a function of the
Court’s own hyper-literal way of dicing up the training failure at issue. Another way to think about what happened
is the way Justice Ginsburg described it: “Members of the District Attorney’s Office, including the District Attorney
himself, misperceived Brady’s compass and therefore inadequately attended to their disclosure obligations.”
Connick v. Thompson, 131 S. Ct. 1350, 1370 (2011) (Ginsburg, J., dissenting).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

deliberate indifference to whether such compliance actually occurred. Obtaining
convictions obviously mattered more.81
Certainly explicit instruction sends important signals, and Connick‟s instruction to work
to rule—to give only what was absolutely required—sent an important signal about the attitude
of the office, particularly when coupled with a lack of accurate information about exactly what
was required.82 So did the fact that the office rarely if ever disciplined anyone for violating
Brady, 83 and that compliance with norms of fair play appeared to have no connection to, or even
a negative impact upon, professional advancement. A culture is communicated through deed,
gesture, attitude, intonation and all sorts of intangibles. It is communicated by the decision not to
spend valuable office time on training about discovery obligations.
The Court in Connick held that “a district attorney is entitled to rely on prosecutors‟
professional training and ethical obligations in the absence of specific reason, such as a pattern of
violations, to believe that those tools are insufficient to prevent future constitutional violations in
the usual and recurring situations with which the prosecutors must deal.” 84 Yet when all the
incentives align to encourage the prosecutor to ignore his role as a minister of justice, the result
is entirely predictable, and the policymaker, quite simply, should not be “entitled to rely on
prosecutors‟ professional training and ethical obligations.” Such reliance under the
circumstances amounts to willful blindness, and unfortunately the Connick opinion makes the
choice of willful blindness cost-free.

81

Brief of Amicus Curiae the National Association of Criminal Defense Lawyers in Support of Respondent at 17.
See generally Mosteller, supra note 79 (arguing that mistakes are inevitable in the absence of open file
discovery).
83
See Connick, 131 S. Ct. at 1382 (Ginsburg, J., dissenting) (asserting that nobody had ever been disciplined by
Connick for a Brady violation).
84
131 S. Ct. at 1359.
82

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

The Section 1983 Compensatory Scheme as Shell Game
Most problematic about the majority opinion in Connick is that, quite simply, it never
once addresses the goals of Section 1983, which are commonly held to be compensation and
deterrence. The entire burden of the majority opinion is to establish the narrowness of failure to
train entity liability. Like a C+ law school exam, Justice Thomas‟s majority opinion treats this as
a doctrinal question that can be discussed in a vacuum, without ever addressing the underlying
purposes of the statute the Court is charged with construing.85
It is instructive to compare the current Supreme Court‟s approach to that in Owen v. City
of Independence, the 1980 case deciding that under Sec. 1983, municipal entities were not
entitled to immunity from suit. The language of the opinion, and indeed the legislative history on
which the opinion relies, sound sadly quaint today. The Court quoted Representative
Shallabarger (the bill‟s author), who said about the proper construction of the statute: “The act is
remedial, and in aid of the preservation of human liberty and human rights. All statutes and
constitutional provisions authorizing such statutes are liberally and beneficently construed. It
would be most strange and, in civilized law, monstrous were this not the rule of interpretation.”86
The Owen Court went on to observe that given the statute‟s central aim to provide protection to
85

Justice Scalia’s concurrence is similarly focused on the dangers of an expansive view of failure to train liability. It
goes further by apparently claiming that Brady was not violated in any event, because prosecutors never followed
up to determine what Thompson’s blood type was, and therefore remained ignorant of whether the blood
evidence would in fact be exculpatory. Connick v. Thompson, 131 S. Ct. 1350, 1369 (2011) (Scalia, J., concurring).
Confusingly, Justice Scalia also indicates that Deegan’s failure to turn over evidence that he himself believed
exculpatory did violate Brady, evidently by virtue of its bad faith nature. Although Brady makes the good or bad
faith of the State irrelevant, Justice Scalia is apparently treating Deegan’s failure as a failure to preserve evidentiary
material rather than a failure to turn over exculpatory evidence, on the theory that absent knowledge of
Thompson’s blood type, the evidence should not be considered exculpatory. He notes that bad faith is relevant to
claims of failure to preserve evidence that, if subjected to tests, might lead to results that exonerate the
defendant. But see Justice Ginsburg’s dissent at 7 n6, describing this as a “don’t ask, don’t tell” version of the
prosecutor’s obligation under Brady. Connick v. Thompson, 131 S. Ct. 1350, 1373 n.6 (Ginsburg, J., dissenting).
86
Owen v. City of Independence, 445 U.S. 622, 636 (1980).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

those wronged by the misuse of power, it would be “‟uniquely amiss‟ if the government itself
were permitted to disavow liability for the injury it has begotten.”87
One major factor cited by the Owen court in favor of rejecting municipal immunity was
its concern that if the entity could block suit by asserting immunity the statute‟s compensatory
aims would be thwarted. “Many victims of municipal malfeasance would be left remediless if the
city were also allowed to assert a good-faith defense.”88 Under the Owen Court‟s approach to
interpreting Section 1983, Van de Kamp‟s grant of absolute immunity to individual prosecutors
should be viewed as a reason for entity liability, not against it. Without entity liability,
Thompson is left remediless, exactly the unjust situation Owen warned against. Likewise, letting
Connick‟s office off the hook does violence to the deterrent aims of the statute and remove any
incentive for prosecutors to institute rules and programs designed to minimize the likelihood of
violating rights. One searches the majority opinion and the Scalia concurrence in vain for any
discussion of the need to allocate the costs of constitutional harm, or any recognition of the
perverse incentives created by the Court‟s holding.
Since Monroe v. Pape, the Court‟s Section 1983 jurisprudence has recognized that
explicit written law-- “the law on the books”--is not the only actionable source of liability. It has
recognized that law can be adequate in theory but not in practice—as indeed it was with the
failures to prosecute the Klan that were a central impetus for the statute.89 The Court has
struggled to steer clear of respondeat superior liability90 while fleshing out the contours of a

87

Owen v. City of Independence, 445 U.S. 622, 651 (1980).
Owen v. City of Independence, 445 U.S. 622, 651 (1980).
89
Monroe v. Pape, 365 U.S. 167, 180 (1961).
90
Though it is beyond the scope of this article, the question of whether the refusal to apply respondeat superior
liability is yet another misreading of the ambiguous statutory history, specifically the meaning of the 1871
88

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

municipal liability jurisprudence that does not simply immunize policymakers for having a
facially constitutional policy in place and proceeding to allow every subordinate official to
ignore it at will. Connick v. Thompson comes uncomfortably close to endorsing precisely this
latter course of action. At oral argument members of the Court saw no problem with a policy that
instructed prosecutors simply to turn over what was required and nothing more. If assistant
prosecutors misconstrued such a policy, the fault would be assumed to lie with them until the
number of violations of Brady by subordinates reached the level of custom. The policy itself,
despite its lack of accompanying training and its direction to err on the side of withholding
evidence, could not be at fault because, as the majority holds,
A district attorney is entitled to rely on prosecutors‟ professional training and ethical
obligations in the absence of specific reason, such as a pattern of violations, to believe
that those tools are insufficient to prevent future constitutional violations in the usual and
recurring situations with which the prosecutors must deal. 91
The failure to train cases arose from the realization that unless there is an incentive to do
otherwise, policymakers are likely to choose a facially legal but widely ignored policy, or simply
adopt no policy at all. 92 In light of Connick, the incentives point toward adopting no policy on

Congress’s rejection of the Sherman Amendment is an important one. See City of Oklahoma City v. Tuttle, 471 U.S.
808, 834-41(1985) (plurality opinion) (Stevens, J., dissenting). (arguing for the adoption of respondeat superior
liability in Sec. 1983 cases, and asserting that the rejection of the Sherman amendment, on which the Court relies
in rejecting respondeat superior liability, establishes only that the 1871 Congress did not mean to hold
governmental entities liable for the acts of private parties of which it had no notice, an entirely separate
proposition from holding government liable for the acts of its own employees). See also Tony Mauro, Stevens
Criticizes Ruling on Prosecutorial Immunity, supra note 73, reporting on a speech by retired Justice Stevens
criticizing the Connick decision and arguing that Sec. 1983 municipal liability ought to be extended to permit
respondeat superior liability, either by judicial interpretation or by an act of Congress amending the statute. See
also Don Doernberg’s article for this symposium, arguing that the rejection of the Sherman amendment should not
have been construed to preclude respondeat superior liability.
91
131 S. Ct. at 1359.
92
See Susan Bandes, Monell, Parratt, Daniels and Davidson: Distinguishing a Custom or Policy from a Random,
Unauthorized Act, 72 Iowa L. Rev. 101, 155 and n399 (discussing skewed governmental incentives toward
inaction).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

training or supervision.93 An office can take this route with impunity, and chalk up every
deviation from law as the isolated act of a rogue prosecutor until the deviations reach the level of
custom. This is the “gaping hole” problem the Court skirted in City of St. Louis v. Praprotnik.94
If all responsibility is delegated to mid level or street level personnel, the policymaker may be
insulated from liability as long as he has no unconstitutional policy in place. The Praprotnik
Court optimistically assumed that “custom” would fill the liability gap.95 But Connick drives
home how easily a court can disaggregate conduct so that a series of wrongful acts is construed
as a random assortment of isolated incidents rather than the sort of pattern that should put a
policymaker on notice.
The Court in Owen v. City of Independence observed that Section 1983 was meant to
“encourage those in a policymaking position to institute internal rules and programs designed to
minimize the likelihood of unintentional infringements on constitutional rights,”96 and to
encourage individual officials to err on the side of protecting rights. The threat of liability should
also “encourage those in a policymaking position to institute internal rules and programs
designed to minimize the likelihood of unintentional infringements on constitutional rights.”97
For concrete examples of the costs of the other sort of regime—the sort that does not err on the
side of protecting rights-- one need look no further than the New Orleans Parish D.A.‟s office.
The office ethos was to err on the side of non-disclosure. When it came to Brady, the top-down
93

Justice Kagan asked in oral argument whether “the failure to train or supervise in any way and setting up a
structural system that’s pretty much guaranteed to produce Brady violations…would be enough” absent any actual
violations. The attorney for Connick’s office responded that it would not be enough absent a pattern of violations,
and this seems to be the correct answer under the Connick decision.
94
City of St. Louis v. Praprotnik, 485 U.S. 112, 131 (1988) (plurality opinion).
95
See Praprotnik, 485 U.S. at 131. But see Praprotnik, 485 U.S. at 144 (Brennan, J., dissenting) (arguing that custom
and usage will not fill the gaping hole left by the decision’s reliance on official policy and refusal to recognize
implicit delegation of policymaking authority).
96
Owen v. City of Independence, 445 U.S. 622, 652 (1980).
97
Owen v. City of Independence, 445 U.S. 622, 652 (1980).

Susan A. Bandes, The Lone Miscreant, The Self-Training Prosecutor, and other Fictions: A Comment on
Connick v. Thompson, Fordham L. Rev. (forthcoming) (draft of May 16, 2011).

model was a failure to train, supervise and discipline, and every incentive was skewed toward
failure to act. The resulting failures were therefore predictable—failure to follow up on a
suspect‟s blood type in the face of blood evidence, failure to turn over evidence, failure to
disclose a deathbed confession that the wrong man was facing execution, failure to uphold the
“minister of justice” aspect of the prosecutor‟s dual role. For a Court concerned with the
compensatory, deterrent and declaratory aims of Section 1983, placing its imprimatur on such a
regime should be difficult to justify. But as Justice Scalia tellingly reveals in his concurrence, the
Court considers constitutional violations inevitable, and is haunted by the specter of too much
liability. 98 The opposing nightmare scenario—a remedial vacuum for egregious constitutional
violations—is acknowledged only in dissent.

98

“Brady mistakes are inevitable. So are all species of error routinely confronted by prosecutors…The District
Court’s instructions cover every recurring situation in which citizens’ rights can be violated.” 131 S. Ct. at 1367
(Scalia, J., concurring).