Volume 60, Issue 2
Page 503
S
tanford
Law Review
FOUR MODELS OF FOURTH AMENDMENT
PROTECTION
Orin S. Kerr
© 2007 by Orin S. Kerr and the Board of Trustees of the Leland Stanford Junior University,
from the Stanford Law Review at 60 S
TAN. L. REV. 503 (2007). For information visit
http://lawreview.stanford.edu.
503
ARTICLES
F
OUR MODELS OF FOURTH AMENDMENT
PROTECTION
Orin S. Kerr*
The Fourth Amendment protects reasonable expectations of privacy, but the
Supreme Court has refused to provide a consistent explanation for what makes
an expectation of privacy “reasonable.” The Court’s refusal has disappointed
scholars and frustrated students for four decades. This Article explains why the
Supreme Court cannot provide an answer: no one test can accurately and
consistently distinguish less troublesome police practices that do not require
Fourth Amendment oversight from more troublesome police practices that are
reasonable only if the police have a warrant or compelling circumstances.
Instead of endorsing one approach, the Supreme Court has recognized four
coexisting approaches. There are four models of Fourth Amendment protection:
a probabilistic model, a private facts model, a positive law model, and a policy
model. Using multiple models has a major advantage over using one model. It
allows the courts to use different approaches in different contexts depending on
which approach most accurately and consistently identifies practices that need
Fourth Amendment regulation. Explicit recognition of the four models would
advance this function, resulting in more accurate and consistent Fourth
Amendment rules.
INTRODUCTION.......................................................................................................504
I. THE FOUR MODELS.............................................................................................507
A. The Probabilistic Model..............................................................................508
B. The Private Facts Model.............................................................................512
C. The Positive Law Model..............................................................................516
© 2007 by Orin S. Kerr and the Board of Trustees of the Leland Stanford Junior University.
* Professor, George Washington University Law School. Thanks to Eve Brensike,
Yale Kamisar, Wayne LaFave, Saul Levmore, Doug Lichtman, Chip Lupu, Eric Muller,
Richard Myers, Eric Posner, Adam Samaha, Daniel Solove, Lior Strahilevitz, Cass Sunstein,
Peter Swire, and the participants in law school faculty workshops at the University of
Michigan, University of Chicago, Northwestern University, George Washington University,
the University of North Carolina, and Loyola University Chicago for helpful comments on
an earlier draft.
504 STANFORD LAW REVIEW [Vol. 60:503
D. The Policy Model ........................................................................................519
E. The Relationship Among the Models ...........................................................522
II. THE CASE FOR MULTIPLE MODELS OF FOURTH AMENDMENT PROTECTION......525
A. The Goals of the Reasonable Expectations of Privacy Test ........................526
B. Why the Proxy Models Cannot Provide Exclusive Guides to Fourth
Amendment Protection ...............................................................................531
1. The Probabilistic Model .......................................................................531
2. The Positive Law Model .......................................................................532
3. The Private Facts Model.......................................................................534
C. Why the Policy Model Cannot Provide an Exclusive Guide to Fourth
Amendment Protection ...............................................................................536
1. Lower Courts and the Reasonable Expectation of Privacy Test...........537
2. The Instability of the Policy Model in the Lower Courts ......................539
D. The Case for Multiple Models.....................................................................542
1. Supreme Court Selection Among the Four Models...............................543
2. Lower Court Use of the Four Models ...................................................545
3. The Need for Recognition of the Four Models......................................548
CONCLUSION..........................................................................................................549
INTRODUCTION
The reasonable expectation of privacy test is the central mystery of Fourth
Amendment law. According to the Supreme Court, the Fourth Amendment
regulates government conduct that violates an individual’s reasonable
expectation of privacy.
1
But no one seems to know what makes an expectation
of privacy constitutionally “reasonable.” The Supreme Court has repeatedly
refused to offer a single test.
2
The Court has noted that “concepts of real or
personal property law” might be relevant, as well as “understandings that are
recognized and permitted by society.”
3
But the Court has elsewhere rejected
property as a guide,
4
and no one knows when society might opt to “recognize”
or “permit” something. Who is “society,” and how do Supreme Court Justices
know what it thinks? Although four decades have passed since Justice Harlan
introduced the test in his concurrence in Katz v. United States,
5
the meaning of
1. See Smith v. Maryland, 442 U.S. 735, 740 (1979) (discussing Katz v. United States,
389 U.S. 347, 361 (1967) (Harlan, J., concurring)).
2. See, e.g., O’Connor v. Ortega, 480 U.S. 709, 715 (1987) (O’Connor, J., plurality
opinion) (“We have no talisman that determines in all cases those privacy expectations that
society is prepared to accept as reasonable.”); Oliver v. United States, 466 U.S. 170, 177
(1984) (“No single factor determines whether an individual legitimately may claim under the
Fourth Amendment that a place should be free of government intrusion not authorized by
warrant.”); see also 1 W
AYNE R. LAFAVE, SEARCH AND SEIZURE: A TREATISE ON THE FOURTH
AMENDMENT § 2.1(a), at 380 (3d ed. 1996) (“The Supreme Court . . . has never managed to
set out a comprehensive definition of the word ‘searches’ as it is used in the Fourth
Amendment.”).
3. Rakas v. Illinois, 439 U.S. 128, 143 n.12 (1978).
4. See, e.g., Warden v. Hayden, 387 U.S. 294 (1967).
5. 389 U.S. at 360 (Harlan, J., concurring).
November 2007] FOURTH AMENDMENT MODELS 505
the phrase “reasonable expectation of privacy” remains remarkably opaque.
Among scholars, this state of affairs is widely considered an
embarrassment. The Court’s handiwork has been condemned as “distressingly
unmanageable,”
6
“unstable,”
7
and “a series of inconsistent and bizarre results
that [the Court] has left entirely undefended.”
8
Treatises and casebooks
struggle to explain the test. Most simply announce the outcomes in the
Supreme Court’s cases,
9
and some suggest that the only way to identify when
an expectation of privacy is reasonable is when five Justices say so.
10
The
consensus among scholars is that the Supreme Court’s “reasonable expectation
of privacy” cases are a failure.
11
The chaos prompts an obvious question: why can’t the Supreme Court
settle on a single test for what makes an expectation of privacy “reasonable”?
Many areas of law require courts to apply vague standards, such as whether a
company was “negligent”
12
or a defendant’s awareness of risk deviated from
the standard of a reasonable person.
13
But the confusion over the reasonable
expectation of privacy test is much deeper. Supreme Court opinions cannot
even agree on what kind of test it is. Is it descriptive? Is it normative? Just what
does it measure? The cases are all over the map, and the Justices have declined
to resolve the confusion.
14
This Article explains why the Supreme Court has not and cannot adopt a
6. Richard G. Wilkins, Defining the “Reasonable Expectation of Privacy”: An
Emerging Tripartite Analysis, 40 V
AND. L. REV. 1077, 1107 (1987).
7. Sherry F. Colb, What Is a Search? Two Conceptual Flaws in Fourth Amendment
Doctrine and Some Hints of a Remedy, 55 S
TAN. L. REV. 119, 122 (2002).
8. Silas J. Wasserstrom & Louis Michael Seidman, The Fourth Amendment as
Constitutional Theory, 77 G
EO. L.J. 19, 29 (1988).
9. See, e.g., C
HARLES H. WHITEBREAD & CHISTOPHER SLOBOGIN, CRIMINAL
PROCEDURE: AN ANALYSIS OF CASES AND CONCEPTS (3d ed. 1993).
10. See, e.g., R
OBERT M. BLOOM, SEARCHES, SEIZURES, AND WARRANTS 46 (2003)
(“How do we know what society is prepared to accept as reasonable? Because there is no
straightforward answer to this question, ‘reasonable’ has largely come to mean what a
majority of the Supreme Court Justices says is reasonable . . . .”); P
HILIP E. JOHNSON, CASES
AND
MATERIALS ON CRIMINAL PROCEDURE 19 (3d ed. 2000) (“When the court refers to
society’s judgment, it is looking in a mirror.”). Wayne LaFave’s influential treatise suggests
that the best explanation for current doctrine is a tautology—“the fourth amendment protects
those interests that may justifiably claim fourth amendment protection”—with the important
caveat that most Supreme Court Justices do not understand what justifiably claims Fourth
Amendment protection. 1 L
AFAVE, supra note 2, § 2.1(d), at 393 (quoting Anthony G.
Amsterdam, Perspectives on the Fourth Amendment, 58 Minn. L. Rev. 349, 385 (1974)).
11. See, e.g., Donald R.C. Pongrace, Stereotypification of the Fourth Amendment’s
Public/Private Distinction: An Opportunity for Clarity, 34 A
M. U. L. REV. 1191, 1208 (1985)
(“[M]ost commentators have recognized that regardless of the political palatability of recent
decisions, fourth amendment doctrine is in a state of theoretical chaos that belies its
supposed objective legitimation of governmental intrusions into our ‘private affairs.’”).
12. See, e.g., United States v. Carroll Towing Co., 159 F.2d 169 (2d Cir. 1947).
13. See, e.g., MODEL PENAL CODE § 2.02(c) (1985).
14. See infra Part I.
506 STANFORD LAW REVIEW [Vol. 60:503
single test for what makes an expectation of privacy “reasonable.” Because
finding an expectation of privacy “reasonable” usually subjects the
government’s conduct to the warrant requirement, the doctrine needs to
distinguish less troublesome police practices permitted without a warrant from
more troublesome practices allowed only with a warrant or under special
circumstances such as exigent circumstances or consent. The Supreme Court
has not and cannot adopt a single test for when an expectation is “reasonable”
because no one test effectively and consistently distinguishes the more
troublesome police practices that require Fourth Amendment scrutiny from the
less troublesome practices that do not.
There are two basic ways that courts could try to develop such a test, and
neither approach works in practice. First, courts could identify a measurement
that serves as a reliable proxy for whether a police practice requires regulation.
Such proxies fail because the facts of police investigations prove too diverse;
no one measurement accurately draws the line in all cases. Alternatively, courts
could examine specific practices directly and decide whether they are
troublesome enough to require Fourth Amendment regulation. This approach
fails because it cannot be administered consistently by decentralized lower
courts. Because each Fourth Amendment case involves a single discrete set of
facts, courts must imagine each case as within a category of cases before
determining whether that category of police practices is troublesome enough to
require a warrant. This choice of category is entirely arbitrary, however,
meaning that no two lower courts would be likely to agree on any given Fourth
Amendment rule.
The failure of any one test to consistently distinguish police practices
needing Fourth Amendment regulation from those that do not has led to the
mixed system that exists today. Although the courts speak of a single
“reasonable expectation of privacy” test, the one label masks several distinct
but coexisting approaches. Four approaches predominate, together reflecting
four different models of Fourth Amendment protection. The first three rely on
proxies. The probabilistic model considers the likelihood that the subject’s
information would become known to others or the police. The lower the
likelihood, the more likely it is that a reasonable expectation of privacy exists.
The private facts model asks whether the government’s conduct reveals
particularly private and personal information deserving of protection. This
approach focuses on the information the government collects rather than how it
is collected. The positive law model considers whether the government conduct
interferes with property rights or other legal standards outside the Fourth
Amendment. When courts apply the positive law model, an expectation of
privacy becomes reasonable when it is backed by positive law such as trespass.
The fourth and final model, the policy model, reflects the direct approach.
Courts applying the policy model focus directly on whether the police practice
should be regulated by the Fourth Amendment.
Scholars and students of Fourth Amendment law find the current approach
November 2007] FOURTH AMENDMENT MODELS 507
frustrating because the courts routinely mix and match the four models. Most
Supreme Court opinions feature multiple models to varying degrees, and they
often switch from model to model without recognizing the change. It’s easy to
see why the current approach is so widely condemned: at the Supreme Court
level, the Justices pick from different arguments and can seem to justify almost
any result by picking the model that best suits it. And yet there is no recognized
meta-theory to the models, no single rationale that explains when some models
should be used and others should not be. The result is a body of law that seems
chaotic and confused and in need of major reworking.
But appearances can be deceiving. What at first looks like conceptual
confusion turns out to be a much-needed range of approaches. Specifically, the
use of multiple models has a critical advantage over the use of a single model:
it facilitates a decentralized Fourth Amendment in which different models
apply in different settings depending on which model best identifies practices
in need of constitutional regulation in that setting. Lower courts can then
incorporate the Supreme Court’s choice of model through analogy, resulting in
the predominance of particular models in particular types of cases. Indeed, it
turns out that the Supreme Court’s cases reflect this dynamic already, at least to
a modest degree. The Court’s emphasis on particular models seems to correlate
reasonably well with the contexts in which those models accurately help
identify police practices in need of constitutional regulation.
Greater awareness of the four models could facilitate this goal
considerably. The appearance of confusion in the Supreme Court’s cases partly
reflects the incorrect assumption that there must be a single test for when an
expectation of privacy is reasonable. Greater awareness of the need for multiple
approaches can help judges select models in each case to better accomplish the
goals of the reasonable expectation of privacy test. At the Supreme Court level,
Justices should pick models by considering which models best identify
practices in need of regulation in that setting. Lower court judges should in turn
apply Supreme Court precedents with the Court’s choice of model explicitly in
mind.
The Article will proceed in two Parts. Part I introduces the four models,
and it explains how they work and the Supreme Court cases in which each
model appears. Part II explains why the Court has embraced all four models
instead of one, and how a greater recognition of the models can help courts
better use them to accurately and consistently identify which police practices
should count as a Fourth Amendment “search.”
I.
THE FOUR MODELS
The reasonable expectation of privacy test distinguishes investigative steps
that the Fourth Amendment regulates from investigative steps that it does not
regulate. If government conduct violates a reasonable expectation of privacy,
then that conduct is a “search” and is legal only if justified by a search warrant
508 STANFORD LAW REVIEW [Vol. 60:503
or a specific exception to the warrant requirement such as consent or exigent
circumstances.
15
On the other hand, if government action does not implicate a
reasonable expectation of privacy, then the Fourth Amendment does not
regulate it and investigators can take that step at any time without constitutional
limitation.
16
As a result, the reasonable expectation of privacy test defines the
line between unregulated investigative steps that can be used at any time from
special investigative steps that must be used only sparingly and in specific
circumstances.
But what makes an expectation of privacy constitutionally “reasonable”?
This Part argues that the Supreme Court’s decisions include four equally viable
answers to the question. There are four different models of Fourth Amendment
protection—four relatively distinct categories of argument used to justify
whether a reasonable expectation of privacy exists.
17
Two are normative and
two are descriptive. Two are macro-scale and two are micro-scale.
18
Most
opinions mix and match the four approaches, relying on multiple models in
each opinion. As a result, observers often don’t see the distinct types of claims.
This Part clarifies the four existing models, and demonstrates that the Supreme
Court sometimes embraces and sometimes rejects each of the four models as a
guide to Fourth Amendment protection.
A. The Probabilistic Model
The first model of the Fourth Amendment is what I term the probabilistic
model. According to this approach, a reasonable expectation of privacy depends
on the chance that a sensible person would predict that he would maintain his
privacy. The inquiry is descriptive rather than normative: it tries to assess the
likelihood that a person will be observed or a place investigated based on
prevailing social practices.
19
Under the probabilistic approach, a person has a
reasonable expectation of privacy when the odds are very high that others will
15. See Illinois v. Rodriguez, 497 U.S. 177, 185 (1990).
16. This analysis assumes that the government conduct is not a seizure. Seizures are
regulated by the Fourth Amendment even if they do not violate a reasonable expectation of
privacy. See Soldal v. Cook County, 506 U.S. 56 (1992).
17. Cf. P
HILIP BOBBITT, CONSTITUTIONAL FATE: THEORY OF THE CONSTITUTION (1982).
For reasons that will become clear later in the Article, I will exclude stare decisis as an
independent rationale for or against Fourth Amendment protection. Precedent is often used
as a crutch in Fourth Amendment law; the Supreme Court will often say that the reasonable
expectation of privacy test just so happens to match pre-Katz interpretations of the Fourth
Amendment. See, for example, United States v. White, 401 U.S. 745, 750 (1971), reaffirmed
by On Lee v. United States, 343 U.S. 747 (1952), which had held that the police did not need
a warrant to go undercover and wear a “wire” that transmitted the defendant’s conversations
to a police observation post, and Oliver v. United States, 466 U.S. 170 (1984), reaffirmed by
Hester v. United States, 265 U.S. 57 (1924), retaining the “open fields” doctrine.
18. See infra Part I.E.
19. In part, this is the “understandings that are recognized and permitted by society,”
discussed in Rakas v. Illinois, 439 U.S. 128, 143 n.12 (1978).
November 2007] FOURTH AMENDMENT MODELS 509
not successfully pry into his affairs. As those odds drop, the individual’s
expectation of privacy becomes less and less reasonable.
20
As a result, the
Fourth Amendment protects citizens against unexpected invasions of privacy.
When government conduct collects evidence in a way that interferes with
customs and social expectations, revealing what a reasonable person might
expect would remain hidden, it violates a reasonable expectation of privacy
Bond v. United States offers an example of the probabilistic approach.
21
A
border patrol agent boarded a bus at the Texas-Mexico border and conducted a
brief search for narcotics by walking the length of the bus and squeezing soft
luggage placed in the overhead compartment. A squeeze of the defendant’s
canvas bag revealed what appeared to be a “brick-like” object stored inside,
and the agent then opened the bag and found drugs.
22
In an opinion by Chief
Justice Rehnquist, the Court held that the officer’s “probing tactile
examination” of the defendant’s luggage violated his reasonable expectation of
privacy.
23
The key was that the agent’s probing had exceeded the usual
handling common among bus passengers:
When a bus passenger places a bag in an overhead bin, he expects that other
passengers or bus employees may move it for one reason or another. Thus, a
bus passenger clearly expects that his bag may be handled. He does not expect
that other passengers or bus employees will, as a matter of course, feel the bag
in an exploratory manner.
24
The officer’s conduct was a search because it was contrary to the reasonable
expectations of bus passengers.
Minnesota v. Olson is another useful example.
25
Olson helped rob a gas
station, and after the robbery he returned to a friend’s duplex apartment where
he was staying as an overnight guest. The police searched the apartment
without a warrant and found Olson hiding in a closet. In an opinion by Justice
White, the Court held that Olson had a reasonable expectation of privacy in the
apartment. Such a rule “merely recognizes the everyday expectations of privacy
that we all share,”
26
Justice White explained. “Staying overnight in another’s
home is a longstanding social custom,”
27
and the customary practice is for
20. This appears to be the version of the reasonable expectation of privacy test that is
commonly used in the privacy torts. See, e.g., Lior Jacob Strahilevitz, A Social Networks
Theory of Privacy, 72 U. CHI. L. REV. 919 (2005). This test is also used in the professional
responsibility context to determine when the attorney-client privilege is retained. See, e.g.,
ABA Comm. on Ethics and Prof’l Responsibility, Formal Op. 99-413 (1999) (concluding
that e-mail supports a reasonable expectation of privacy because e-mail is likely to remain
private).
21. 529 U.S. 334 (2000).
22. Id. at 336.
23. Id. at 337.
24. Id. at 338-39.
25. 495 U.S. 91 (1990).
26. Id. at 98.
27. Id.
510 STANFORD LAW REVIEW [Vol. 60:503
hosts to respect the privacy interests of their guests: “The host may admit or
exclude from the house as he prefers, but it is unlikely that he will admit
someone who wants to see or meet with the guest over the objection of the
guest.”
28
Olson had a reasonable expectation of privacy in the apartment
because social customs and norms made it reasonable for him to expect others
would not be admitted there.
The majority and dissenting opinions in California v. Ciraolo provide
another illustration.
29
Ciraolo was growing marijuana in his backyard, and
constructed a 10-foot fence around the property to block others from seeing it.
The police borrowed an airplane, flew over the property at 1,000 feet, and took
photographs of the marijuana plants growing in the backyard. The majority
opinion by Chief Justice Burger suggested that the surveillance did not violate
Ciraolo’s reasonable expectation of privacy because aerial observation is
common: “In an age where private and commercial flight in the public airways
is routine, it is unreasonable for respondent to expect that his marijuana plants
were constitutionally protected from being observed with the naked eye from
an altitude of 1,000 feet.”
30
In dissent, however, Justice Powell disagreed on
the likelihood of observation: he argued that the chances were so low that
Ciraolo’s expectation was reasonable.
31
Although the two opinions disagreed
on the outcome, they agreed on the proper inquiry: both opinions considered
the likelihood that the suspect’s property would be subject to observation by
others.
Bond, Ciraolo, and Olson are just three among many Supreme Court cases
relying on the probabilistic model.
32
They teach that a reasonable expectation
of privacy is a descriptive expectation based on norms and prevailing social
practices that others will not observe what the individual seeks to protect as
private. Whether an expectation of privacy is reasonable depends on the
expectations of a reasonable person. Much like Learned Hand’s famous
negligence formula in Carroll Towing measured the probability that a harmful
event might occur,
33
so a reasonable person might measure the probability that
28.Id. at 99.
29. 476 U.S. 207 (1986).
30.Id. at 215. There are other explanations in the Ciraolo majority opinion; this is
only one among several.
31. Id. at 223 (Powell, J., dissenting) (“[T]he actual risk to privacy from commercial or
pleasure aircraft is virtually nonexistent. Travelers on commercial flights, as well as private
planes used for business or personal reasons, normally obtain at most a fleeting, anonymous,
and nondiscriminating glimpse of the landscape and buildings over which they pass.”).
32. For other examples, see O’Connor v. Ortega, 480 U.S. 709 (1987) (O’Connor, J.,
plurality opinion) (addressing government workplace privacy); California v. Carney, 471
U.S. 386 (1985) (concluding that a person has a lesser expectation of privacy in a car
because cars are heavily regulated and therefore drivers and passengers do not expect as
much privacy in them as they do in homes); and United States v. Dionisio, 410 U.S. 1 (1973)
(addressing grand jury subpoena).
33. United States v. Carroll Towing Co., 159 F.2d 169, 173 (2d Cir. 1947).
November 2007] FOURTH AMENDMENT MODELS 511
particular information will be revealed.
34
So far, so good. But there’s a wrinkle: for every case in which the Court
endorses the probabilistic model, you can find several others flatly rejecting it.
In many cases, the Supreme Court has dismissed the probabilistic model as
simply incorrect as a matter of basic Fourth Amendment law. Consider the
Supreme Court’s recent decision in Illinois v. Caballes.
35
Caballes was stopped
for speeding, and the officer brought a drug-sniffing dog to the scene. When the
dog alerted the officer to the presence of drugs in the trunk, the officer searched
the trunk and found marijuana. The Supreme Court held that the use of the dog
to alert for the presence of drugs was not a search. According to Justice
Stevens’ majority opinion, the chance that the police would find out about the
drugs in the trunk was completely irrelevant to the Fourth Amendment inquiry:
“[T]he expectation ‘that certain facts will not come to the attention of the
authorities’ is not the same as an interest in ‘privacy that society is prepared to
consider reasonable.’”
36
As the Court emphasized twenty years earlier in
United States v. Jacobsen, “The concept of an interest in privacy that society is
prepared to recognize as reasonable is, by its very nature, critically different
from the mere expectation, however well justified, that certain facts will not
come to the attention of the authorities.”
37
The Supreme Court has also rejected the probabilistic model in its many
cases on misplaced confidences. In these cases, the defendants gave private
information to a friend or business associate on the assumption that such
information would remain a secret. The friend then gave the information to the
police, either because he was a confidential informant,
38
he was wearing a wire
on the government’s behalf,
39
or investigators served him with a subpoena.
40
In
all of these cases, the Court held that providing information to the third party
eliminated any reasonable expectation of privacy no matter how unlikely it was
34. From this perspective, the reasonable expectation of privacy test is a bit circular:
whether a reasonable person would expect privacy arguably depends at least in part on the
Supreme Court’s cases construing the Fourth Amendment. See, e.g., J
EFFREY ROSEN, THE
UNWANTED GAZE: THE DESTRUCTION OF PRIVACY IN AMERICA 60 (2000) (“Harlan’s test was
applauded as a victory for privacy, but it soon became clear that it was entirely circular.”);
Michael Abramowicz, Constitutional Circularity, 49 UCLA
L. REV. 1, 60-61 (2001)
(“Fourth Amendment doctrine, moreover, is circular, for someone can have a reasonable
expectation of privacy in an area if and only if the Court has held that a search in that area
would be unreasonable.”). Of course, even under this approach, the circularity is modest.
Only a lawyer would think that a person’s chances of having privacy in a particular place
hinges in large part on whether the police can enter it legally without a warrant, consent, or
exigent circumstances. Fortunately, police searches are rarer than that whether or not a
warrant is required.
35. 543 U.S. 405 (2005).
36.Id. at 408-09 (quoting United States v. Jacobsen, 466 U.S. 109, 122 (1984)).
37. Jacobsen, 466 U.S. at 122.
38. Hoffa v. United States, 385 U.S. 293 (1966).
39. United States v. White, 401 U.S. 745 (1971).
40. United States v. Miller, 425 U.S. 435 (1976).
512 STANFORD LAW REVIEW [Vol. 60:503
that the friend would betray the suspect’s confidence. As the Court summarized
in United States v. Miller:
[T]he Fourth Amendment does not prohibit the obtaining of information
revealed to a third party and conveyed by him to Government authorities, even
if the information is revealed on the assumption that it will be used only for a
limited purpose and the confidence placed in the third party will not be
betrayed.
41
The defendant assumes the risk, no matter how small, that the information will
end up in the hands of the police.
42
Finally, consider United States v. Ross, which involved the search of a
paper bag found in the suspect’s car.
43
The Court’s opinion emphasized that the
ease of opening the bag was completely irrelevant to whether it supported a
reasonable expectation of privacy. According to the Court, “the central purpose
of the Fourth Amendment” foreclosed a distinction between containers that
were “worthy” and “unworthy” of protection:
For just as the most frail cottage in the kingdom is absolutely entitled to the
same guarantees of privacy as the most majestic mansion, so also may a
traveler who carries a toothbrush and a few articles of clothing in a paper bag
or knotted scarf claim an equal right to conceal his possessions from official
inspection as the sophisticated executive with the locked attaché case.
44
Whether a reasonable person would expect privacy in his bag is irrelevant,
much like every home receives protection regardless of whether it is an isolated
mountain retreat or a frequently burglarized apartment in an urban
neighborhood.
In sum, the probabilistic model provides only an occasional guide to
whether an expectation of privacy is reasonable. Sometimes the Court relies on
it, but often the Court flatly rejects it and looks elsewhere.
B. The Private Facts Model
The second model of Fourth Amendment protection is what I call the
private facts model. The private facts model focuses on the information the
government collects, and considers whether that information is private and
worthy of constitutional protection. If the government obtains information that
is particularly private, then the acquisition of that information is a search; if the
41.Id. at 443.
42. See, e.g., Smith v. Maryland, 442 U.S. 735, 745 (1979) (“We are not inclined to
make a crazy quilt of the Fourth Amendment, especially in circumstances where (as here)
the pattern of protection would be dictated by billing practices of a private corporation.”);
see also Anderson v. Pollard, 2006 U.S. Dist. LEXIS 81960, at *11-12 (E.D. Wis. Nov. 7,
2006) (arguing unsuccessfully that a prison cell retained a reasonable expectation of privacy
because it was not searched very often).
43. 456 U.S. 798 (1982).
44.Id. at 822 (footnote omitted).
November 2007] FOURTH AMENDMENT MODELS 513
information collected is not private or does not otherwise merit protection, then
no search has occurred. The key question becomes what information the
government collected rather than how it was obtained or whether the
government’s conduct was unexpected.
United States v. Jacobsen offers a helpful example.
45
In Jacobsen, a
cardboard box sent via Federal Express broke open during delivery. A white
powder seeped out, and an FBI agent performed a chemical field test of the
powder to determine if the powder was cocaine. The field test returned a
positive result, leading to criminal charges against the package recipient. In an
opinion by Justice Stevens, the Supreme Court held that a field test for
narcotics could not violate a reasonable expectation of privacy. A field test
“could disclose only one fact,”
46
whether the powder was cocaine. But that fact
could not be a “private” fact, the Court concluded.
47
If the test returned a
negative result, then “such a result reveals nothing of special interest.”
48
And if
the test returned a positive result, then it merely revealed that a crime had been
committed because the possession of cocaine is a crime.
49
Because the field
test could only reveal evidence of a crime, “and no other arguably ‘private’
fact,” it could not violate any expectation of privacy that was constitutionally
“legitimate.”
50
Dow Chemical Co. v. United States reflects a similar approach.
51
The
Environmental Protection Agency (EPA) hired a commercial photographer to
take aerial photographs of a chemical plant to identify violations of
environmental protection laws. The owner of the chemical plant, Dow
Chemical, brought a civil suit claiming that the photography violated the Fourth
Amendment. In an opinion by Chief Justice Burger, the Court rejected the
Fourth Amendment challenge based in part on the limited information the
photography revealed: “[T]he photographs here are not so revealing of intimate
details as to raise constitutional concerns. Although they undoubtedly give EPA
more detailed information than naked-eye views, they remain limited to an
outline of the facility’s buildings and equipment.”
52
The photographs revealed
some information but did not reveal anything important or intimate. Therefore
it was not a search.
Of course, like the other models, the private facts model works both ways:
it can be used either to deny Fourth Amendment protection or to justify it.
Consider United States v. Karo, in which Drug Enforcement Agency (DEA)
agents placed a tracking device inside a can of chemicals used to extract
45. 466 U.S. 109 (1984).
46. Id. at 122.
47. Id. at 123.
48. Id.
49. Id.
50. Id.
51. 476 U.S. 227 (1986).
52. Id. at 238.
514 STANFORD LAW REVIEW [Vol. 60:503
cocaine from materials imported into the United States.
53
The tracking device
revealed the location of the can, and this showed that the can had been brought
inside a private home. In an opinion by Justice White, the Court held that using
the device to obtain information about the inside of a home violates a
reasonable expectation of privacy because details about the inside of a home
are private facts:
The monitoring of an electronic device such as a beeper is, of course, less
intrusive than a full-scale search, but it does reveal a critical fact about the
interior of the premises that the Government is extremely interested in
knowing and that it could not have otherwise obtained without a warrant.
54
The government conduct was a search because it revealed a critical fact.
Jacobsen, Dow Chemical Co., and Karo are just a few of the many cases
invoking the private facts model. Their common theme is that a Fourth
Amendment search occurs and a reasonable expectation of privacy is therefore
violated when the government obtains particularly private and personal
information deserving of privacy protection. The approach seems to follow
from our sense that the disclosure of some particularly personal information is
an invasion of privacy while the disclosure of less personal information may
seem like no invasion of privacy at all.
55
If the Fourth Amendment reflects this
widely shared notion of privacy, perhaps a “reasonable expectation of privacy”
requires a normative assessment of the value of the information revealed to the
government. This interpretation becomes more plausible thanks to the
occasional alternative formulation of the test as looking for a “legitimate”
expectation of privacy rather than a “reasonable” expectation of privacy.
56
Whether an expectation of privacy is “legitimate” clearly depends on a
normative judgment of legitimacy, an intuition reflected in the private facts
model.
Once again, however, these cases tell only half of the story. The Supreme
Court’s search cases often ignore the private facts model, finding a search when
no private information is obtained and concluding that no search occurs when
even very invasive information is collected.
57
For example, in Arizona v. Hicks,
53. 468 U.S. 705 (1984).
54.Id. at 715.
55.See Daniel J. Solove, A Taxonomy of Privacy, 154 U.
PA. L. REV. 477 (2006).
56. See, e.g., Rakas v. Illinois, 439 U.S. 128, 143 n.12 (1978).
57. Christopher Slobogin and Joseph Schumacher emphasized this point in a 1993
article that tested the social expectations of 217 individuals (most of them students).
Christopher Slobogin & Joseph E. Schumacher, Reasonable Expectations of Privacy and
Autonomy in Fourth Amendment Cases: An Empirical Look at “Understandings Recognized
and Permitted by Society,” 42 D
UKE L.J. 727 (1993). The subjects of the study were asked to
rank the “intrusiveness” of various law enforcement practices featured in Fourth Amendment
decisions. Slobogin and Schumacher found little correlation between the scope of Fourth
Amendment protection and the intrusiveness of different steps: “[T]he Supreme Court’s
conclusions about the scope of the Fourth Amendment are often not in tune with commonly
held attitudes about police investigative techniques.” Id. at 774.
November 2007] FOURTH AMENDMENT MODELS 515
the police entered an apartment looking for a gunman who had fired shots from
the apartment moments earlier.
58
An officer came across expensive audio
equipment in what was otherwise a ramshackle apartment, and picked up a
turntable to check its serial number for a match with stolen equipment. In an
opinion by Justice Scalia, the Court held that moving the stereo equipment to
reveal the serial number violated the defendant’s reasonable expectation of
privacy. The fact that moving the turntable revealed only a serial number was
irrelevant:
It matters not that the search uncovered nothing of any great personal value to
respondent—serial numbers rather than (what might conceivably have been
hidden behind or under the equipment) letters or photographs. A search is a
search, even if it happens to disclose nothing but the bottom of a turntable.
59
The Court has also rejected the private facts model in cases involving the
opening of packages. Here the rule is simple: when a government agent opens a
suspect’s sealed package or other container, opening the container is a “search”
regardless of what is inside the box.
60
Note the tension with Caballes, the dog
sniff case: the use of a drug-sniffing dog to find drugs without opening a
package is no search because the possession of drugs is not protected as a
private fact, but opening a package and finding drugs is a search even if the
opening reveals nothing private except the drugs themselves. In the former
case, the private facts model applies; in the latter case, it does not.
61
Finally, the Court has rejected the private facts model in the misplaced
confidences cases.
62
Recall that these are the cases in which a suspect gives
private information to a friend or business associate on the assumption that
such information will remain a secret, and the friend or business associate then
gives the information to the police. The Court’s opinions teach that the act of
disclosure to a third party eliminates protection: having surrendered control
over the information and given it to a third party, the suspect no longer retains
any privacy rights in it. In these cases, however, the nature of the information
disclosed is irrelevant: the mere mechanism of disclosure eliminates protection,
even if the information disclosed is deeply private. A defendant who discloses
information to a third party simply has “assumed the risk” that the third party
will turn around and disclose the information to the police.
Much like the probabilistic model, the private facts model provides only an
occasional guide to when an expectation of privacy is reasonable. In some
cases, whether police conduct disclosed private facts through the use of a
practice is essential to whether it constitutes a “search,” while in others it is
irrelevant.
58. 480 U.S. 321 (1987).
59.Id. at 325.
60.See United States v. Jacobsen, 466 U.S. 109, 120 n.17 (1984).
61. See id. (“A container which can support a reasonable expectation of privacy may
not be searched, even on probable cause, without a warrant.”).
62.See supra notes 38-44.
516 STANFORD LAW REVIEW [Vol. 60:503
C. The Positive Law Model
The positive law model offers a third model of Fourth Amendment
protection. When courts apply the positive law model, they look at whether
there is some law that prohibits or restricts the government’s action (other than
the Fourth Amendment itself). If the government broke the law in order to
obtain the information it did, the government conduct violated a reasonable
expectation of privacy. This approach often focuses on whether the information
collected was legally available to the public. If a member of the public could
have accessed the information legally, then it does not violate a reasonable
expectation of privacy for the government to do the same. The positive law
approach is descriptive, not normative: it asks whether the government’s access
to the suspect’s information was achieved legally based on preexisting legal
doctrine.
The positive law model has deep roots in Fourth Amendment history. The
Fourth Amendment began as a mechanism for protecting property rights,
63
and
the property-based contours of the Fourth Amendment remain strong today.
64
As a result, “protection for property under the Fourth Amendment”
65
remains a
major theme of the post-Katz era: If a person owns property or has a close
relationship to the owner, access to that property usually violates his reasonable
expectation of privacy. On the other hand, lack of a property right or a
relationship to the property owner often renders an expectation of privacy
unreasonable. Descriptively speaking, interference with property rights
provides a surprisingly helpful guide to the scope of Fourth Amendment
protection.
66
Rakas v. Illinois provides an example of the positive law model.
67
The
defendants had helped rob a clothing store, and were passengers in the getaway
car when it was stopped by a pursuing police officer. The officer searched the
car and found a gun under the front passenger seat and a box of shells in the
locked glove compartment. In an opinion by Justice Rehnquist, the Court held
that searching the car did not violate the passengers’ Fourth Amendment rights.
According to the Court, “One of the main rights attaching to property is the
right to exclude others, and one who owns or lawfully possesses or controls
property will in all likelihood have a legitimate expectation of privacy by virtue
of this right to exclude.”
68
The passengers lacked a property right, however,
and also lacked a mere possessory right in the car or the items seized. Because
they “asserted neither a property nor a possessory interest in the automobile
63. See Entick v. Carrington, 19 Howell’s State Trials 1029, 1030 (C.P. 1765).
64. See Orin S. Kerr, The Fourth Amendment and New Technologies: Constitutional
Myths and the Case for Caution, 102 M
ICH. L. REV. 801, 809-15 (2004).
65. Soldal v. Cook County, 506 U.S. 56, 64 (1992).
66. See Kerr, supra note 64.
67. 439 U.S. 128 (1978).
68. Id. at 143 n.12 (internal citations omitted).
November 2007] FOURTH AMENDMENT MODELS 517
searched nor an interest in the property seized,” they could not establish a
reasonable expectation of privacy.
69
The positive law model extends beyond property, however. Consider
Justice White’s plurality opinion in Florida v. Riley.
70
The facts of Riley are
similar to those of Ciraolo: investigators flew a helicopter over the defendant’s
property at an altitude of 400 feet, and from that vantage point observed
marijuana growing in his greenhouse. The government’s brief invoked the
positive law model, pointing out that Federal Aviation Administration (FAA)
regulations banning fixed-wing aircraft from traveling below an altitude of 500
feet do not apply to helicopters. Because FAA regulations permitted helicopters
to fly above Riley’s property at that altitude, doing so did not violate his
reasonable expectation of privacy. Justice White’s plurality opinion agreed:
We would have a different case if flying at that altitude had been contrary to
law or regulation. But helicopters are not bound by the lower limits of the
navigable airspace allowed to other aircraft. Any member of the public could
legally have been flying over Riley’s property in a helicopter at the altitude of
400 feet and could have observed Riley’s greenhouse. The police officer did
no more. . . . [I]t is of obvious importance that the helicopter in this case was
not violating the law . . . .
71
Like the other models, the positive law model cuts both ways; it can either
deny protection or justify it. An example of positive law justifying protection is
Justice Powell’s dissent for four Justices in Dow Chemical Co.
72
The EPA
hired an aerial photographer to snap pictures of a Dow industrial plant to
determine whether Dow had complied with environmental regulations. Dow’s
brief argued that this surveillance divulged Dow’s trade secrets, and that this
violation of trade secret law infringed Dow’s reasonable expectation of privacy.
In his dissent, Justice Powell agreed, joined by Justices Brennan, Marshall, and
Blackmun. Trade secrets laws “constitute society’s express determination that
commercial entities have a legitimate interest in the privacy of certain kinds of
property,” Justice Powell wrote. Because Dow’s efforts to shield its plant from
view allegedly triggered trade secret protections, it also triggered Dow’s
“reasonable expectation of privacy in its commercial facility in the sense
required by the Fourth Amendment.”
73
Rakas, Riley, and the dissent in Dow Chemical Co. are only three of the
many Supreme Court opinions invoking the positive law model.
74
Their
69. Id. at 129.
70. 488 U.S. 445 (1989).
71. Id. at 451 (White, J., plurality opinion, joined by Rehnquist, C.J., Scalia, J., and
Kennedy, J.).
72. Dow Chemical Co. v. United States, 476 U.S. 227 (1986).
73. Id. at 249 (Powell, J., concurring in part, dissenting in part).
74. Perhaps the most remarkable example of this is the discussion of Bahamian bank
secrecy law in United States v. Payner, 447 U.S. 727 (1980). Payner argued that Bahamian
bank secrecy laws created a reasonable expectation of privacy in his bank records in the
Bahamas. The Court disagreed, finding that “[t]he statute is hardly a blanket guarantee of
518 STANFORD LAW REVIEW [Vol. 60:503
common theme is that a reasonable expectation of privacy derives from values
expressed and affirmed by positive law, such that government action that
violates the standards of existing law triggers Fourth Amendment protection.
Once again, however, these cases tell only half the story. For every opinion
applying the positive law model, there is another opinion rejecting it. (You
knew that was coming, didn’t you?)
Consider Oliver v. United States,
75
which held that the 1920s-era “open
fields” doctrine survived the establishment of the Katz “reasonable expectation
of privacy” framework.
76
Under the open fields doctrine, the police can roam
around and investigate crime on open property belonging to the suspect so long
as they don’t get too close to the suspect’s house.
77
The Oliver court justified
this curious doctrine by rejecting the positive law model in an opinion by
Justice Powell. According to Justice Powell, the fact that the officers violated
trespass laws did not mean that it triggered a reasonable expectation of privacy:
“[T]respass law extends to instances where the exercise of the right to exclude
vindicates no legitimate privacy interest. Thus, in the case of open fields, the
general rights of property protected by the common law of trespass have little
or no relevance to the applicability of the Fourth Amendment.”
78
California v. Greenwood is also helpful.
79
Greenwood left her trash in a
bag near the border of her property for pickup by local trash collectors. A
police officer instructed the trash collectors to pick up the trash and turn it over
to the authorities, and a subsequent search of the trash revealed amounts of
cocaine and hashish. The defendant’s brief tried to invoke the positive law
model: it noted that the California Supreme Court had held that such conduct
violated the state constitution, and reasoned that state constitutional law
privacy. Its application is limited; it is hedged with exceptions; and we have been directed to
no authority construing its terms.” Id. at 732 n.4. Another example is Alderman v. United
States, 394 U.S. 165 (1969), in which the Court held that wiretapping a home telephone line
implicates the homeowner’s reasonable expectation of privacy even if he was not a party to
the call intercepted. According to the Court, the homeowner has Fourth Amendment rights
even if his own privacy rights are not invaded: the fact that the surveillance occurs inside the
owner’s home is enough to confer Fourth Amendment protection. Id. at 179-80 (“The rights
of the owner of the premises are as clearly invaded when the police enter and install a
listening device in his house as they are when the entry is made to undertake a warrantless
search for tangible property . . . .”).
75. 466 U.S. 170 (1984).
76. This doctrine was first established in Hester v. United States, 265 U.S. 57, 59
(1924) (citing W
ILLIAM BLACKSTONE, 4 COMMENTARIES *223, *225-226).
77. If the police get too close to the home they will cross on to the “curtilage,” which
is the area near the home that is essentially home-like and thus protected by the Fourth
Amendment. See United States v. Dunn, 480 U.S. 294 (1987).
78. Oliver, 466 U.S. at 183-84. It is possible to reconcile these outcomes: trade secrets
are about secrets, whereas trespass law is not. So it could be consistent with a theory that
some types of positive law should be recognized under the positive law model while others
types should not be. Courts haven’t drawn these lines, though, in part because trespass law
often is about privacy.
79. 486 U.S. 35 (1988).
November 2007] FOURTH AMENDMENT MODELS 519
rendered Greenwood’s expectation of privacy constitutionally reasonable. The
Court disagreed. Justice White’s majority opinion bristled at the notion that
state law could govern Fourth Amendment protection. “We have never
intimated” at such a notion, Justice White explained.
80
“We have emphasized
instead that the Fourth Amendment analysis must turn on such factors as ‘our
societal understanding that certain areas deserve the most scrupulous protection
from government invasion.’”
81
Much like the probabilistic model and the private facts model, the positive
law model is only an occasional guide to Fourth Amendment protection. Some
opinions embrace it and others reject it.
D. The Policy Model
The fourth and final model of Fourth Amendment protection is the policy
model. Under the policy model, the reasonable expectation of privacy inquiry
poses a policy question: should a particular set of police practices be regulated
by the warrant requirement or should those practices remain unregulated by the
Fourth Amendment? If the consequences of leaving conduct unregulated are
particularly troublesome to civil liberties, then that conduct violates a
reasonable expectation of privacy. On the other hand, if the practical
consequences of regulating such conduct unnecessarily restrict government
investigations given the gain to civil liberties protection, then any expectation
of privacy is constitutionally unreasonable. Whether an expectation of privacy
is reasonable hinges on a normative value judgment. Judges must consider the
consequences of regulating a particular type of government activity, weigh
privacy and security interests, and opt for the better rule.
82
As legal realists, we may rightly suspect that the policy model often drives
outcomes in Supreme Court decisions. As we will see in the next Part, it is
widely agreed that something akin to the policy model helps frame the basic
goals of Fourth Amendment law and the reasonable expectation of privacy
test.
83
As a result, the policy model presumably plays a guiding hand in many
cases even when an opinion itself is framed in terms of the probabilistic model,
private facts model, and/or positive law model. At the same time, explicit
reliance on the policy model turns out to be very common in the Supreme
Court’s Fourth Amendment decisions. This willingness is on display in Katz v.
United States itself.
84
The majority’s rationale for why bugging a phone booth
triggered the Fourth Amendment was primarily instrumental: “To read the
80. Id. at 43.
81. Id. at 43 (quoting Oliver, 466 U.S. at 178 (emphasis added in Greenwood)).
82. See Amsterdam, supra note 10, at 403; see also Colb, supra note 7, at 124 (arguing
that the reasonable expectation of privacy test forces “[d]ecisions . . . [to] rest on normative
choices”).
83. See infra Part II.
84. 389 U.S. 347 (1967).
520 STANFORD LAW REVIEW [Vol. 60:503
Constitution more narrowly is to ignore the vital role that the public telephone
has come to play in private communication.”
85
In the wake of Katz, the Justices
have regularly employed explicit normative arguments to justify outcomes in
“search” cases.
Kyllo v. United States offers an illustration of the policy approach to the
reasonable expectation of privacy test.
86
Acting on the suspicion that Kyllo was
growing marijuana in his home, federal agents pointed a thermal imaging
device towards the exterior of his home to check for high temperatures that
might indicate the use of heat lamps. The imaging device measured infrared
radiation emanating from the walls and roof of the home; because the amount
of infrared radiation emanating from a surface depends on its temperature, the
device created a temperature profile of the exterior of a home. Sitting in a
parked car across the street, the agents used the device and found that one wall
and the roof of Kyllo’s garage were unusually hot. In his opinion for the Court,
Justice Scalia reasoned that use of the device to monitor a home was a search
because in “the long view”
87
the use of sense-enhancing devices presented a
major threat to privacy in the home. “To withdraw protection . . . would be to
permit police technology to erode the privacy guaranteed by the Fourth
Amendment,”
88
Scalia reasoned. “[O]btaining by sense-enhancing technology
any information regarding the interior of the home that could not otherwise
have been obtained without physical ‘intrusion”
89
must be a search, as such a
holding “assures preservation of that degree of privacy against government that
existed when the Fourth Amendment was adopted.”
90
Justice Marshall’s dissent in Smith v. Maryland provides another
example.
91
The police had asked the telephone company to install a pen
register on Smith’s home to confirm that he was guilty of harassing a woman.
The pen register recorded the outgoing numbers dialed from Smith’s home
telephone, and confirmed that he was indeed the caller.
92
In a dissent joined by
Justice Brennan, Justice Marshall urged that this monitoring violated Smith’s
reasonable expectation of privacy. Marshall’s inquiry was explicitly normative:
he asked whether permitting the monitoring was consistent with “a free and
85. Id. at 352.
86. 533 U.S. 27 (2001).
87. Id. at 40.
88. Id. at 34.
89. Id. (quoting Silverman v. United States, 365 U.S. 505 (1961)).
90. Id. The Court invoked a similar normative judgment in United States v. Karo, 468
U.S. 705 (1984), the case involving use of an electronic locating device. According to the
majority, using a locating device was a “search” because the unregulated use of such devices
would be too dangerous: “Indiscriminate monitoring of property that has been withdrawn
from public view would present far too serious a threat to privacy interests in the home to
escape entirely some sort of Fourth Amendment oversight.” Id. at 716.
91. 442 U.S. 735 (1979).
92. Id. at 736 & n.1, 737.
November 2007] FOURTH AMENDMENT MODELS 521
open society” and “the basic values underlying the Fourth Amendment.”
93
Permitting the use of pen registers without Fourth Amendment oversight was
too dangerous, Justice Marshall argued, as it would “impede certain forms of
political affiliation and journalistic endeavor that are the hallmark of a truly
free society.”
94
As with the other three models, the policy model cuts both ways. It can be
used to justify protection or its absence. Consider a prison search case, Hudson
v. Palmer.
95
Hudson, a prison guard, searched the cell of Palmer, a prison
inmate. Palmer sued Hudson, claiming that the search violated his Fourth
Amendment rights. In an opinion by Chief Justice Burger, the Court held that
the inmate had no reasonable expectation of privacy in his cell because a
contrary rule would interfere with sound corrections policy. According to
Burger, the expectation of privacy analysis “necessarily entails a balancing of
interests,” in this case a balance between “the interest of society in the security
of its penal institutions and the interest of the prisoner in privacy within his
cell.”
96
The opinion then simply announced that the Court had decided to
“strike the balance in favor of institutional security.”
97
Palmer did not have a
reasonable expectation of privacy in his cell because “recognition of privacy
rights for prisoners in their individual cells simply cannot be reconciled with
the concept of incarceration and the needs and objectives of penal
institutions.”
98
Kyllo, Hudson, and the dissent in Smith v. Maryland are only three of
the many cases relying on the policy model.
99
Their common theme is that the
“reasonable expectation of privacy” test reflects a normative policy choice
between regulating government conduct and leaving such conduct unregulated.
Different opinions approach the policy question in different ways, but the basic
idea is that whether an expectation of privacy is reasonable depends on the
desirability of a legal rule holding it to be so.
100
Unlike the other three models,
93. Id. at 750-51 (Marshall, J., dissenting).
94. See id. at 751.
95. 468 U.S. 517 (1984).
96. Id. at 527.
97. Id.
98. Id. at 526.
99. See, e.g., Delaware v. Prouse, 440 U.S. 648, 662-63 (1979) (justifying the
conclusion that searching a car violates a reasonable expectation of privacy on the ground
that “[w]ere the individual subject to unfettered governmental intrusion every time he
entered an automobile, the security guaranteed by the Fourth Amendment would be seriously
circumscribed”); United States v. White, 401 U.S. 745, 778 (1971) (Harlan, J., dissenting)
(arguing that the reasonable expectation of privacy framework should be decided in light of
policy considerations).
100. This focus on policy interests in Fourth Amendment law arguably goes back to
the first Fourth Amendment decision, Boyd v. United States, 116 U.S. 616 (1886). The
Court’s description of the Fourth Amendment’s purpose in that case is notable for its
potpourri of themes: according to Justice Bradley, the “essence” of the Fourth Amendment
was protection against “the invasion of [a person’s] indefeasible right of personal security,
522 STANFORD LAW REVIEW [Vol. 60:503
the policy model is never rejected in Supreme Court opinions.
101
At the same
time, many Supreme Court opinions ignore the policy model on their face,
102
and most opinions that discuss it do so only alongside other models.
103
Decisions rely heavily on the policy approach in a few cases, moderately in
some cases, and not at all in other cases. Although we may suspect that policy
plays an important role even when it is not invoked explicitly, the Supreme
Court’s opinions themselves treat the policy model much like the other three
approaches. The policy approach provides one of four possible arguments that
the Justices use to justify the scope of a reasonable expectation of privacy.
E. The Relationship Among the Models
Now that the basic principles of the four models have been developed, it
may help to step back and compare them. At the broadest level, the four models
provide four distinct ways of identifying when an expectation of privacy is
personal liberty, and private property.” Id. at 630. Following Boyd, the Court’s focus on
weighing of privacy interests was a common theme in early cases. In 1932, for example, the
Court warned in United States v. Lefkowitz, that the Fourth Amendment must be “construed
liberally to safeguard the right of privacy.” 285 U.S. 452, 464 (1932). And in Wolf v.
Colorado, Justice Frankfurter urged that at the “core” of the Fourth Amendment was “[t]he
security of one’s privacy against arbitrary intrusion by the police.” 338 U.S. 25, 27 (1949).
See, e.g., United States v. Jeffers, 342 U.S. 48 (1951); see also McGuire v. United States,
273 U.S. 95 (1927).
101. A few cases suggest that the policy concerns should go beyond a mere balancing
of interests to include pragmatic questions such as institutional competence, see Kyllo v.
United States, 533 U.S. 27, 51 (2001) (Stevens, J. dissenting), and the administrability of
rules, see Smith v. Maryland, 442 U.S. 735, 744-45 (1979).
102. See, e.g., California v. Ciraolo, 476 U.S. 207 (1986). Ciraolo involved aerial
surveillance of marijuana plants growing in the defendant’s backyard, and the opinion relied
on the probabilistic and positive law models but ignored the policy model. According to
Chief Justice Burger, aerial surveillance from public airspace does not violate a reasonable
expectation of privacy because any member of the public could legally conduct such
surveillance (positive law), and in an age of routine air travel expecting privacy is
unreasonable (probabilistic). See id. at 213-14. But the opinion never explores the policy
consequences of its holding, never balances the privacy and law enforcement interests, and
never suggests that such inquiries might be relevant. Although we can imagine such
arguments, none appear in the opinion.
103. Justice Scalia’s opinion for the Court in Kyllo, 533 U.S. at 29, is a good example.
Although it relies heavily on the policy model, it also discusses and attempts to conform to
both the private facts model and the probabilistic model. The government’s brief relied
heavily on the private facts model, and claimed that merely obtaining the temperature of the
home was not sufficiently “intimate” to constitute a search. Justice Scalia disagreed,
contending that the temperature of the home was “intimate” because all facts about the inside
of the home are “intimate” facts. Id. at 37. Scalia also defined the scope of the Court’s
holding using the probabilistic model: the prohibition against the warrantless use of sense-
enhancing devices applied only to those devices “not in general public use.” Id. at 34. Use of
the thermal imaging device was a search because “thermal imaging is not ‘routine’” under
the probabilistic model. Id. at 39 n.6. If using such devices had been routine, the Court
suggested, the outcome might very well be different.
November 2007] FOURTH AMENDMENT MODELS 523
reasonable. The probabilistic model looks to prevailing social practices; the
private facts model looks to the privacy invasion itself; the positive law model
looks to positive law; and the policy model looks to consequences. These
differences map across two important dichotomies: normative versus
descriptive, and micro-scale versus macro-scale. The normative/descriptive
dichotomy is easy to understand. Descriptive approaches key the legal test to
existing law or social practice, whereas normative approaches rely on judicial
assessments of preferred values and practices.
104
The micro-scale/macro-scale
dichotomy is equally fundamental.
105
A micro-scale inquiry focuses on the
exact facts of the case before the court, such as what the officer did or what
information he obtained. In contrast, a macro-scale inquiry looks to
characteristics of the general type of government conduct, of which the case at
hand is merely one example. This requires some arbitrary classification, of
course: cases can be grouped in different ways, depending on what facts seem
most important.
106
But the basic idea is that each case is an example of a
broader subset of cases.
Classifying the four models using these two dichotomies reveals how the
four models provide the courts with four different approaches to measuring
whether an investigatory step should be regulated under the Fourth
Amendment. First, the private facts model is normative and micro-scale. It
looks at the information obtained in the precise case before the Court, and
requires a normative assessment of the “privateness” of the information. In
contrast, the policy model is normative and macro-scale: it calls for a normative
assessment of the privacy stakes and competing law enforcement needs in a
broader category of cases, such as prison searches,
107
use of sense-enhancing
devices,
108
or use of bugging devices.
109
What matters is the dynamics of a
typical case in a generally defined category, not those of the specific one before
the court. Next, the positive law model is descriptive and micro-scale. It asks
whether the facts of that particular case involve a government violation of a
preexisting legal prohibition. Finally, the probabilistic model is descriptive and
macro-scale: it considers whether a person with an accurate sense of social
practices generally would expect privacy in the exposed information. Put
another way, the probabilistic model considers the overall ex ante likelihood of
104.See supra notes 19 and 82 and accompanying text.
105.Cf. Amsterdam, supra note 10, at 367 (distinguishing between “atomistic” and
“regulatory” approaches to the Fourth Amendment).
106. For example, imagine a police officer breaks into a post office and rifles through
the mail of a suspect he is investigating; when he finds a private letter, the officer opens the
letter and reads it. A micro-scale inquiry might look at whether the officer violated any laws,
or what information he obtained in that particular case. A macro-scale inquiry might see the
case as one example of government access to postal letters, and seek to base a rule on the
general and typical characteristics of postal mail and postal mail privacy.
107. Hudson v. Palmer, 468 U.S. 517 (1984).
108. Kyllo, 533 U.S. 27.
109. Katz v. United States, 389 U.S. 347 (1967).
524 STANFORD LAW REVIEW [Vol. 60:503
government access to information resulting from a wide range of possible
practices, rather than the specific case before the court.
110
The four basic categories map as follows:
Micro-Scale Macro-Scale
Descriptive Positive Law Probabilistic
Normative Private Facts Policy
The results under these different approaches will often overlap, which no
doubt explains why most opinions mix and match different models. For
example, positive law often tracks social norms. If conduct is illegal, we can
expect few people to do it. If most people follow the law, the two descriptive
models will support the same result. Similarly, the disclosure of private facts
will tend to have significant civil liberties implications; if government
surveillance reveals unusually private information, letting investigators access
that information without legal restriction is likely to result in a substantial
injury to privacy. Thus we might expect the two normative models to reach the
same result in many cases as well. More broadly, in easy cases we would
expect all four models to point to the same result. For example, breaking into a
suspect’s house would be a search under all four models, while walking down
the street in a crowded public place wouldn’t be a search under any of them.
The fact that the different approaches overlap both explains why the different
models are used simultaneously and why Fourth Amendment scholars have not
seen the arguments as distinct. Most opinions jump from model to model
without acknowledging the jump. This is possible because the models usually
are used as general tools rather than clear and specific doctrinal tests. In many
cases there is considerable wiggle room in terms of how each model applies.
Nonetheless, the four models do frame four distinct sets of approaches to
answering when an expectation of privacy is reasonable. In some cases they
will overlap, but in other cases they will point in different directions. The hard
cases tend to be those in which the different models point judges to different
conclusions. In those cases, courts must choose which model applies to that
particular case.
110. Consider whether a person has a reasonable expectation of privacy in her postal
mail letters. Under the probabilistic model, the question is whether, taking into account all of
the possible ways that mail privacy might be violated, discounted by their probability, a
person could reasonably expect privacy in her postal mail. This is a macro-scale inquiry, in
that it concerns the general question of the odds of an invasion of mail privacy rather than an
estimate of the chances that the exact invasion that did occur was likely to occur. The
chances that the exact invasion that did occur was likely to occur ex ante is irrelevant under
the probabilistic model. For example, if an FBI agent breaks into a post office and opens a
letter addressed to a suspect, the probabilistic model would not focus attention on the
likelihood that the FBI agent would break in exactly as he did, but on the likelihood of the
letter being opened by the government in a range of hypothetical scenarios.
November 2007] FOURTH AMENDMENT MODELS 525
II.
THE CASE FOR MULTIPLE MODELS OF FOURTH AMENDMENT PROTECTION
The Supreme Court’s use of four models of Fourth Amendment protection
prompts an obvious question: why hasn’t the Court selected a single model and
applied it in every case? The existing approach seems unsatisfying, as it allows
the Justices to justify nearly any outcome in any case by picking and choosing
among the models. If a majority of the Justices doesn’t like the result that
seems to follow from one model, it can simply focus on a different model to
generate the desired outcome.
Why has the Court failed to embrace a single model? This Part argues that
the Court has failed to generate a single test because no one test accurately and
consistently distinguishes police practices that are reasonable per se from those
practices that are reasonable only if the government has a countervailing
interest such as probable cause. The Fourth Amendment imposes reasonable
restrictions on investigative practices, and the exclusionary rule forces courts to
distinguish police practices that are reasonable even if not regulated by the
Fourth Amendment from those practices that are reasonable only in special
circumstances. The function of the “reasonable expectation of privacy” doctrine
is to demarcate this line, and the exclusionary rule requires courts to implement
it by announcing clear rules for each discrete police practice.
The Supreme Court cannot settle on a single model because no model
draws that line accurately in every case. Three of the models try to draw the
lines using proxies for whether a police practice demands regulation; no one
model works in all cases because the proxies are imperfect. The probabilistic
model uses the proxy of social practice and everyday experience; the private
facts model uses the proxy of information obtained; the positive law model uses
the proxy of law outside the Fourth Amendment. Each of these models
accurately identifies police practices in need of regulation in some cases. But
the key point is that no one model fits every case: the range of cases is
incredibly diverse, and as a result each proxy works in some cases but fails in
others. No one size fits all.
Further, the direct approach of the policy model can’t provide an exclusive
guide because it cannot be administered consistently in a system of lower court
rulemaking. Although the Supreme Court grabs the headlines, the job of
announcing rules that govern the reasonable expectation of privacy test falls
mostly to the hundreds of decentralized lower courts. Because the police often
must rely on lower court precedents from other jurisdictions, consistency across
jurisdictions is essential. And yet the policy model cannot create such certainty
as it depends entirely on arbitrary classifications of what police “practice” the
court must evaluate. Fourth Amendment cases always involve a specific set of
facts, and the policy model requires courts to imagine those facts as one
example of a broader category of cases. But the choice of category is
completely arbitrary: courts can pick along a continuum from extremely
specific to very broad, and no point along the continuum is clearly better than
526 STANFORD LAW REVIEW [Vol. 60:503
another. In a system of lower court rulemaking, use of the policy model alone
can only result in doctrinal chaos and confusion.
Although no one model of Fourth Amendment protection can provide an
exclusive guide, the coexistence of multiple models has an important advantage
over a single approach. It allows the localized use of specific models to more
accurately identify police practices needing regulation. Fourth Amendment
decision making relies heavily on analogies, which means that legal reasoning
adopted in one factual setting will tend to be adopted in similar factual settings.
In a system of coexisting models, this dynamic allows the Supreme Court to
pick the model most likely to accurately identify police practices in need of
constitutional regulation in the setting of that particular case. Lower courts
faced with new cases will reason by analogy to the Supreme Court’s
precedents, and through analogical reasoning will incorporate and adopt the
models used in the Court’s most similar decisions. The result is a system of
localized models: different models will control different types of cases
depending on which model best draws the line in each particular type of case.
Something like this happens already, even though judges and Justices are
not fully aware of the four models. Reviewing the cases shows that the choices
of models are not random; the need to justify a decision using a sensible
rationale will normally push Justices to focus on the model that provides the
most sensible results in related cases. In the lower courts, the usual process of
reasoning by analogy will push them to adopt the same models in factually
similar cases. More importantly, a better appreciation of the four models can
help courts achieve the goals of the Fourth Amendment more effectively. An
awareness of the four models and their function suggests a meta-theory of sorts
as to which models should be used: the Supreme Court should try to explain
outcomes using the model that best draws the needed line in that type of case.
Lower courts should then mirror that decision in factually similar cases,
resulting in different models governing different types of cases in ways that
most accurately identify police practices in need of constitutional regulation.
I will develop this argument in three Parts. The first Part explains why the
courts need to generate thousands of context-specific rules to identify when an
expectation of privacy is reasonable. The second Part explains why no single
model can provide these rules. The third Part demonstrates how the use of
multiple models of Fourth Amendment protection facilitates the regime of
localized models that helps identify police practices in need of constitutional
regulation.
A. The Goals of the Reasonable Expectations of Privacy Test
Understanding why the Supreme Court has not embraced a single model of
Fourth Amendment protection must begin with a functional role of the
reasonable expectation of privacy test and how its rules are created. The Fourth
Amendment prohibits unreasonable searches and seizures, and in the Katz era it
November 2007] FOURTH AMENDMENT MODELS 527
has been understood as an essential tool for requiring reasonable police
practices when the police collect evidence.
111
The ultimate touchstone, the
Supreme Court often reminds us, is reasonableness.
112
In an abstract sense,
reasonableness sets up a familiar balancing of interests: the courts must devise
a set of legal commands that limit police powers to deter abuses and yet also
give the police the needed powers to investigate criminal activities.
113
However, this balancing occurs in the shadow of the exclusionary rule, which
provides the primary remedy to enforce Fourth Amendment rules. Under the
exclusionary rule, the remedy for Fourth Amendment violations is severe:
when the police violate the Fourth Amendment in the course of an
investigation, the evidence obtained as a fruit of the violation ordinarily cannot
be used in court.
114
Courts impose this remedy only at the end of the case: the
government will charge a defendant with an offense, and the defendant will file
a motion to suppress that, if successful, leads to suppression and potentially
dismissal of all of the charges.
The Fourth Amendment’s suppression remedy exerts a profound influence
on the shape of Fourth Amendment doctrine. In particular, it generates
tremendous pressure on the courts to implement the Fourth Amendment using
clear ex ante rules rather than vague ex post standards.
115
Unlike other
remedies, suppression is a blunt instrument. It acts as an on-off switch;
evidence either is excluded or it is not.
116
When the police err, courts must
impose the high social cost of suppressing evidence and letting a potentially
111. See Katz, 389 U.S. at 373-74 (Black, J., dissenting) (noting that the Supreme
Court was interpreting the Fourth Amendment less as a restriction on unreasonable searches
and seizures and more as a general mechanism for limiting police investigations). I should
add a caveat here: the Fourth Amendment covers evidence collection other than the
statements of individual people. That area is addressed by the Fifth Amendment.
112. Brigham City v. Stuart, 126 S. Ct. 1943, 1947 (2006) (“[T]he ultimate touchstone
of the Fourth Amendment is ‘reasonableness’ . . . .” (quoting Flippo v. West Virginia, 528
U.S. 11, 13 (1999) (per curiam))); United States v. Knights, 534 U.S. 112, 118 (2001) (“The
touchstone of the Fourth Amendment is reasonableness . . . .”).
113. See Knights, 534 U.S. at 118-19 (“The touchstone of the Fourth Amendment is
reasonableness, and the reasonableness of a search is determined ‘by assessing, on the one
hand, the degree to which it intrudes upon an individual’s privacy and, on the other, the
degree to which it is needed for the promotion of legitimate governmental interests.’”
(quoting Wyoming v. Houghton, 526 U.S. 295, 300 (1999))).
114.See Wong Sun v. United States, 371 U.S. 471, 485 (1963).
115. Cf. Gerald G. Ashdown, Good Faith, the Exclusionary Remedy, and Rule-
Oriented Adjudication in the Criminal Process, 24 W
M. & MARY L. REV. 335, 336-37 (1983)
(“Although the case-by-case approach may retain needed flexibility, its amorphous nature
provides little guidance to the police as to what investigatory practices will be viewed
retrospectively as reasonable on balance with the individual interest affected.”).
116. Further, every defendant can file a motion to suppress on federal Fourth
Amendment grounds, and most defendants have a very strong incentive to file claims even if
they border on frivolous. Although most claims will fail, defendants do not pay the cost of
filing the motion; defendants facing jail time are provided with attorneys free of charge
under the Sixth Amendment. See generally Gideon v. Wainwright, 372 U.S. 335 (1963).
528 STANFORD LAW REVIEW [Vol. 60:503
guilty man go free. To minimize this cost while ensuring the enforcement of
Fourth Amendment rules, courts turn where possible to clear rules police can
readily follow.
117
Clear rules announce ex ante what the police can and cannot
do; so long as the police comply with the clear rules, the police will know that
the evidence cannot be excluded. The Supreme Court recognized this point
most clearly in New York v. Belton, quoting Fourth Amendment scholar Wayne
LaFave: “Fourth Amendment doctrine, given force and effect by the
exclusionary rule, is primarily intended to regulate the police in their day-to-
day activities,”
118
the Court noted. The protection of the Fourth Amendment
“can only be realized if the police are acting under a set of rules which, in most
instances, makes it possible to reach a correct determination beforehand as to
whether an invasion of privacy is justified in the interest of law
enforcement.”
119
The need for clear rules explains both the existence and the contours of the
reasonable expectation of privacy test. First, it explains why the Supreme Court
has divided the Fourth Amendment into two discrete steps. Fourth Amendment
doctrine includes two questions: the threshold question of what is a “search”—
that is, what violates a reasonable expectation of privacy—and the second
question of the circumstances when a search is “reasonable.”
120
This
bifurcation is sensible textually, as the Fourth Amendment prohibits
“unreasonable searches and seizures.”
121
But it also serves a critical functional
role in light of the Fourth Amendment’s suppression remedy. If the Fourth
Amendment simply asked whether police behavior were “reasonable,” the
police would have little way to know ex ante whether the evidence they
collected would end up being suppressed.
122
Dividing the Fourth Amendment
into two stages provides considerable certainty by carving out a set of
investigative steps that cannot lead to suppression. So long as the police stick to
conduct that does not violate a reasonable expectation of privacy, the Fourth
Amendment does not apply and the suppression remedy is unavailable.
123
In
contrast, conduct that does violate a reasonable expectation of privacy can lead
to suppression. The police must be very careful before engaging in such steps:
117. See generally Wayne R. LaFave, “Case-By-Case Adjudication” Versus
“Standardized Procedures”: The Robinson Dilemma, 1974 S
UP. CT. REV. 127.
118. 453 U.S. 454, 458 (1981) (quoting LaFave, supra note 117, at 141).
119. Id. (quoting LaFave, supra note 117, at 142). Clear rules will not be possible in
every case, of course. For example, they are not feasible in contexts involving emergencies,
such as those that implicate the exigent circumstances exception. However, rules are
preferable to standards where feasible.
120. See, e.g., Flippo v. West Virginia, 528 U.S. 11, 13-14 (1999) (per curiam) (“A
warrantless search by the police is invalid unless it falls within one of the narrow and well-
delineated exceptions to the warrant requirement . . . .”).
121. U.S.
CONST. amend. IV.
122. See Amsterdam, supra note 10, at 414-15.
123. I am excluding Fourth Amendment “seizures” here, which are covered by
separate rules.
November 2007] FOURTH AMENDMENT MODELS 529
suppression of any evidence will result unless the police have a warrant or a
fact-specific exception to the warrant requirement applies.
Now we can see the key inquiry the reasonable expectation of privacy test
must resolve. The results of the test must somehow carve out specific police
practices that are reasonable per se—that is, reasonable if allowed in all
circumstances—and distinguish them from practices that are reasonable only
when justified by a warrant or exception to the warrant requirement.
124
Some
practices are not very invasive or not very likely to be abused; these practices
are reasonable per se and can be free from Fourth Amendment regulation. For
example, walking down a public street is reasonable per se, as giving the police
the ability to walk down the street without justification is not likely to lead to
civil liberties abuses. On the other hand, other practices are highly invasive or
likely to be abused if left unregulated. These practices are reasonable only
when the government can identify countervailing government interests such as
probable cause or exigent circumstances in that specific case that render that
particular search reasonable. For example, it would be unreasonable for the
police to break into your house unless they had a very good reason such as a
warrant, a pressing emergency, or your consent. Whatever the meaning of the
phrase “reasonable expectation of privacy” in the abstract, the test must be
interpreted in ways that make such distinctions in a relatively consistent and
readily administered way.
The need for clear ex ante rules also pushes courts to announce when
government conduct violates a reasonable expectation of privacy using rules
rather than standards.
125
The courts may justify outcomes using models, but the
outcomes themselves tend to be articulated as rules that provide clear ex ante
guidance to law enforcement as to whether police practices violate a reasonable
expectation of privacy. For example, when the Supreme Court applies the
policy model, it usually balances privacy and security and announces a clear
rule. Consider the prison case, Hudson v. Palmer:
126
the Court balanced
privacy and security in the context of a prison and then announced the rule that
prison inmates have no Fourth Amendment rights. Similarly, when courts apply
the private facts model, they typically create a rule by defining the scope of
what does or does not reveal private facts in a rule-like way. In Jacobsen,
127
for
124. This function is sometimes obscured by Fourth Amendment doctrine, which of
course labels the question as whether the conduct is a “search” at all. This bifurcation allows
the courts to carve out police conduct from reasonableness review. From a practical
perspective, however, it is easier to understand the difference as being whether the police
conduct is reasonable per se or needs to be assessed for reasonableness based on the
circumstances.
125. See generally F
REDERICK SCHAUER, PLAYING BY THE RULES: A PHILOSOPHICAL
EXAMINATION OF RULE-BASED DECISION-MAKING IN LAW AND IN LIFE (1991); Louis
Kaplow, Rules Versus Standards: An Economic Analysis, 42 D
UKE L.J. 557 (1992); Cass R.
Sunstein, Problems with Rules, 83 C
AL. L. REV. 953 (1995).
126. 468 U.S. 517 (1984).
127. 466 U.S. 109 (1984).
530 STANFORD LAW REVIEW [Vol. 60:503
example, the Court’s conclusion that a chemical field test reveals no private
facts was used to generate a rule that field testing does not trigger Fourth
Amendment protection. The probabilistic model is probably most conducive to
standards, although it too is usually used to justify a rule. In California v.
Ciraolo,
128
for example, the alleged frequency of aerial surveillance was used
to justify a rule that aerial surveillance does not violate a reasonable
expectation of privacy. Even when the Supreme Court promulgates a standard,
such as the “open fields” doctrine’s four-factor test to distinguish “open fields”
from “curtilage,”
129
application of the test by lower courts ends up creating a
doctrine that begins to look very rule-like.
130
This pressure has led to specific Fourth Amendment rules for almost every
common way of collecting evidence. A few moments leafing through Wayne
LaFave’s magisterial Search and Seizure treatise confirms the point: the treatise
devotes about 370 pages to summarizing the rules.
131
However complex or
vague the “reasonable expectation of privacy” test may be in theory, it takes on
firm shape in the thousands of decisions implementing it in very specific
factual contexts. For example, detailed case guidance tells the police how the
reasonable expectation of privacy test applies to searching stolen goods,
132
opening screen doors,
133
searching hotel rooms after checkout time,
134
searching rental cars,
135
accessing computer files,
136
entering temporary
housing set up by campers and squatters,
137
peering into houses through open
128. 476 U.S. 207 (1986).
129. See United States v. Dunn, 480 U.S. 294 (1987).
130. See, e.g., United States v. Seckman, 175 F. App’x 193, 196 (10th Cir. 2006)
(noting that under the four-factor open fields doctrine, the police can go “places where
visitors could be expected to go, i.e. walkways, driveways or porches”).
131. See 1 W
AYNE R. LAFAVE, SEARCH AND SEIZURE: A TREATISE ON THE FOURTH
AMENDMENT 321-780 (4th ed. 2004).
132. See United States v. Caymen, 404 F.3d 1196 (9th Cir. 2005) (finding no
reasonable expectation of privacy in the contents of a stolen computer).
133. United States v. Arellano-Ochoa, 461 F.3d 1142 (9th Cir. 2006). The opinion
begins, “We publish to clarify Fourth Amendment law regarding the police opening a screen
door without a search warrant.” Id. at 1143. Relying on the probabilistic model, the court
ruled that the answer hinges on whether the solid door behind the screen door is open or
closed. Opening a screen door backed by a closed solid door does not violate a reasonable
expectation of privacy, whereas opening a screen door violates a reasonable expectation of
privacy if the solid door happens to be open. Id. at 1145.
134. See, e.g., United States v. Dorais, 241 F.3d 1124, 1128 (9th Cir. 2001) (“[A]
defendant has no reasonable expectation of privacy in a hotel room when the rental period
has expired and the hotel has taken affirmative steps to repossess the room.”); United States
v. Nerber, 222 F.3d 597, 600 n.2 (9th Cir. 2000) (“For Fourth Amendment purposes, a hotel
room is treated essentially the same, if not exactly the same, as a home.”).
135. United States v. Thomas, 447 F.3d 1191, 1196-99 (9th Cir. 2006).
136. See, e.g., United States v. David, 756 F. Supp. 1385 (D. Nev. 1991).
137. See Amezquita v. Hernandez-Colon, 518 F.2d 8 (1st Cir. 1975) (finding no
reasonable expectation of privacy for squatters).
November 2007] FOURTH AMENDMENT MODELS 531
windows,
138
and many other more or less common investigatory techniques.
The question is, what set of principles can best distinguish police practices
that are reasonable per se from practices that are reasonable only with a warrant
or under special circumstances? What principles can help courts answer this
question in a case with specific facts so that the court can announce whether
that practice does or does not violate a reasonable expectation of privacy?
B. Why the Proxy Models Cannot Provide Exclusive Guides to Fourth
Amendment Protection
Three of the four models of Fourth Amendment protection rely on proxies
to draw the line between police practices that are reasonable per se and
practices that are reasonable only with a warrant or special circumstances. The
probabilistic model relies on the proxy of probability that an event will occur;
the private facts models relies on the proxy of information exposure; and the
positive law model relies on the proxy of positive law regulating access to
places and information. In this Part, I explain why none of these models can
provide the exclusive guide to Fourth Amendment protection. The reason is
straightforward: each model provides an accurate proxy in some situations but
fails in important types of cases. As a result, no one proxy model can provide
the exclusive guide to Fourth Amendment protection.
1. The Probabilistic Model
Let’s start with the probabilistic approach, which teaches that an
individual’s reasonable expectation of privacy is violated if the government
accesses information in an unusual or unexpected way. In many cases, this test
successfully distinguishes less troublesome police practices from more
troublesome practices. The main reason is that individuals often control their
privacy, and when they control their privacy they can decide to take or reject
protective measures to make privacy invasions more or less difficult.
Individuals will tend to hide their most private information unusually carefully,
and they won’t bother hiding information that is not private. As a result, a
highly unexpected intrusion will often correlate with a particularly severe
intrusion on privacy that is reasonable and desirable only if justified by exigent
circumstances or a warrant.
A few simple examples confirm the point. If you have a personal diary and
you want to keep it safe from prying eyes, you might hide it in a locked box
tucked away in your bedroom closet. You’ll protect your privacy by taking
measures that render access unlikely. In contrast, you won’t take strong
measures if you don’t fear the consequences of observation. You might leave a
138. United States v. Barajas-Avalos, 377 F.3d 1040 (9th Cir. 2004) (holding that
approaching a window of a home and peering inside is a search).
532 STANFORD LAW REVIEW [Vol. 60:503
book on the backseat of your car where anyone can see it: onlookers can look
through your window and see what you’re reading, but you’re unlikely to care.
(And if you do care, you can put the book in the trunk.) The probabilistic model
offers a helpful proxy in such cases because the likelihood of privacy seems to
correlate reasonably well with the invasiveness of its deprivation. Government
access to frequently accessed places will usually be reasonable per se, while
government access to rarely accessed places will usually be unreasonable
unless justified by specific circumstances such as a warrant.
However, the correlation often will not hold. In many cases, modest
intrusions on privacy that are per se reasonable will also be highly unexpected,
and common intrusions may trigger very severe government practices that are
reasonable only if justified by a warrant or exigent circumstances. Individuals
often lack control over their privacy. As a result, they cannot take steps to make
such intrusions more or less likely. For example, imagine the government
announced that the FBI is tapping every single telephone call in the United
States to listen for evidence of criminal activity. The invasions of privacy
would be extraordinarily severe, but no reasonable person would expect
privacy in their calls after learning of this fact.
139
The probabilistic model no
longer works, as the fact of widespread surveillance would defeat Fourth
Amendment rights against it. Similarly, consider a burglar plying his trade in a
vacant home far removed from any surveillance cameras or any other
observers.
140
The burglar could quite rationally expect privacy under the
circumstances, and under a probabilistic approach would have a reasonable
expectation of privacy in the burglarized home. The low chances of being
watched would require the police to obtain a warrant before observing the
burglar, even if only from a public street. This result makes no sense, however,
as monitoring from a public street seems to be a per se reasonable police
practice.
As these examples show, the probabilistic model is only an occasionally
useful guide to distinguish police practices that are per se reasonable from
police practices that will be reasonable only if justified by a warrant or an
exception to the warrant requirement. It is often true that invasive steps are rare
and modest steps are frequent. But the reverse is common as well. As a result,
the probabilistic model cannot provide the exclusive guide to Fourth
Amendment protection.
2. The Positive Law Model
The positive law model has similar flaws as a universal guide. Like the
139. Cf. Amsterdam, supra note 10, at 384.
140. Cf. Rakas v. Illinois, 439 U.S. 128, 143 n.12 (1978) (noting that a burglar who
has entered a home to commit a crime cannot have a reasonable expectation of privacy
there).
November 2007] FOURTH AMENDMENT MODELS 533
probabilistic model, the positive law model provides an accurate proxy for the
reasonableness of police practices in many cases. Positive law that restricts
access to information and places often reflects widely shared notions of which
accesses cause significant harms and which do not.
141
Courts, legislatures, and
agencies will tend to promulgate rules restricting access to a person’s private
material when unfettered access will be harmful, which is a major factor in
determining whether a particular government practice is reasonable per se or
reasonable only in context. This means that government practices that violate
positive law will often be reasonable only in context, and those practices that
do not will often be reasonable per se.
Physical intrusions into the home provide an obvious example. Breaking
into to your house interferes with your property rights; it also reflects a deeply
invasive affront to your privacy and security. On the other hand, watching your
house from a public street does not violate positive law; while it may be a bit
creepy, it does not amount to a severe invasion of your privacy. In this context,
positive law nicely matches our intuitions as to which kinds of police practices
should be permitted without a warrant. The positive law model provides a
useful proxy to determine when a government invasion violates a reasonable
expectation of privacy.
But much like the probabilistic model, the positive law model does not
work in every case. In many circumstances positive law will not accurately
capture whether police practices are per se reasonable. This is true for two
primary reasons. First, positive laws are enacted for a wide range of reasons
that may have nothing to do with whether access by criminal investigators
would be reasonable per se. Consider the FAA regulations analyzed in Florida
v. Riley, the helicopter flyover case.
142
The FAA presumably drafted those
regulations to minimize noise and deter accidents, not to limit the police.
Whether the police happened to fly over or under FAA airspace limits has no
significant connection to whether particular police flyovers are reasonable only
if justified by a warrant. Similarly, imagine a person lives on a 100-acre farm
with their small house in the center. Trespassing a few feet onto that person’s
property will violate his property rights but not infringe on his privacy.
143
In
that setting, positive law does not accurately track the reasonableness of the
government’s investigation.
The positive law model may also be insufficient when technology or social
141. Cf. HENRY M. HART, JR. & ALBERT M. SACKS, THE LEGAL PROCESS 148 (1994)
(“Law is a doing of something, a purposive activity, a continuous striving to solve the basic
problems of social living . . . . Legal arrangements (laws) are provisions for the future in aid
of this effort.”).
142. 488 U.S. 445 (1989).
143. See R
ICHARD A. POSNER, THE ECONOMICS OF JUSTICE 314 n.8 (1981) (noting that
under the open fields doctrine, the Fourth Amendment permits invasions of property without
triggering Fourth Amendment protection because the property invasion does not trigger a
privacy invasion).
534 STANFORD LAW REVIEW [Vol. 60:503
practice changes rapidly. The story is a familiar one, as it tracks the standard
explanation for the move from Olmstead to Katz. Technology tends to shift the
privacy implications of different law enforcement steps.
144
New technologies
can divorce privacy and social norms from property law and other statutory and
regulatory protections. New technological surveillance tools make it possible to
invade privacy without interfering with property or other laws, and they can
also make it possible to establish privacy without harnessing positive law.
145
As a result, technological change can make the function of positive law largely
arbitrary; it no longer serves the same function it might if the technology
remained stable. A test that focuses entirely on positive law such as property
will be underprotective with technological surveillance techniques and may
prove overprotective with other techniques.
Like the probabilistic model, positive law accurately distinguishes per se
and contextually reasonable practices in some cases but not others. As a result,
it cannot provide an exclusive guide to what makes an expectation of privacy
reasonable.
3. The Private Facts Model
Finally, we can tell a similar tale about the private facts model. Like the
probabilistic and positive law models, the private facts model offers an accurate
guide to the line between per se reasonable and contextually reasonable police
practices in some cases. The nature of the information obtained by the
government is obviously a critical aspect of its invasiveness and need for legal
regulation. It is extremely invasive if the police learn that you are HIV-positive;
it is not invasive at all if the police recognize that you are having a bad hair
day.
At the same time, the private facts model likewise does not work well in
every case. Whether a government practice is reasonable per se often hinges on
how the government obtains the information rather than what information they
collect. For example, imagine the police break into a home in order to steal the
daily newspaper. The fact that they only collect a public newspaper does not
render the break-in a reasonable law enforcement practice. Surely no one would
want the police to break into homes without a warrant so long as the cops
obtained no particularly private information.
146
On the flip side, imagine a
police officer turns on a TV set and watches a news program announcing that a
basketball star is HIV-positive. Watching a broadcast is reasonable per se:
144. See Kerr, supra note 64, at 864-67.
145. Cf. Olmstead v. United States, 277 U.S. 438, 479 (1928) (Brandeis, J.,
dissenting).
146. In Kyllo, Justice Scalia tried to reconcile this by saying that any information about
the inside of the home is automatically private. 533 U.S. 27, 34 (2001). But this is surely
overbroad, as cases like Smith v. Maryland, 442 U.S. 735 (1979) (holding that pen register
information sent from a home is not covered by Fourth Amendment), would suggest.
November 2007] FOURTH AMENDMENT MODELS 535
surely no one would suggest that the officer should have to obtain a warrant
before watching TV. In these two examples, how the police collected
information mattered more than what information they collected. As these
examples illustrate, whether a particular police practice is reasonable per se or
only contextually reasonable often depends on much more than merely the
information obtained.
Further, focusing only on the information obtained can make police
guidance difficult. The private facts model demands an ex post inquiry into
whether a past search ended up disclosing information that courts can consider
private or non-private. This can lead to clear rules when the police know with
certainty what information their technique will yield. Under Illinois v.
Caballes,
147
for example, the Court’s consideration of dog sniffs follows a
private facts model. Because the only fact disclosed is whether drugs are
present, a nonprivate fact by judicial decree, the technique does not violate a
reasonable expectation of privacy. The rules are clear ex ante because the dog
sniff will never reveal a private fact: either the dog won’t alert, revealing
nothing, or the dog will alert to the presence of narcotics, revealing only
information deemed not deserving of privacy protection in Caballes. Caballes
thus leads to a simple rule: the Fourth Amendment does not regulate dog sniffs.
But no clear rule emerges when the private facts model is applied to
physical searches of packages and car trunks. The officer wouldn’t know what
he’ll see until he opens the package, which means that the officer wouldn’t
know if his opening was going to be a forbidden search until he opened the
package and viewed its contents. If an officer physically opened a sealed
package and the only private thing he saw were drugs, the search would be
constitutional, but if the officer saw something private such as a private letter or
personal item the opening would violate the owner’s reasonable expectation of
privacy. The legality of the search would seem to depend on the happenstance
of how the narcotics were packaged, which savvy offenders could control to
trigger Fourth Amendment protection. And imagine if the same rule were
applied to searches in public places: A drug dealer could guarantee Fourth
Amendment protection of his drugs even in public by taping his stash to the
pages of his personal diary. The officer could not search for the drugs without
opening the diary, presumably revealing private facts from the pages of the
diary.
148
If the private facts model were applied to every case, officers would
never know ex ante what steps they could take, and clever felons could
manipulate the rules to guarantee Fourth Amendment protection even when
conducting their offenses in public.
147. 543 U.S. 405 (2005).
148. This problem doesn’t arise with dog sniffs because a well-trained dog either finds
nothing or alerts to the presence of narcotics. Those are the only options, and neither option
involves the possible disclosure of private facts.
536 STANFORD LAW REVIEW [Vol. 60:503
C. Why the Policy Model Cannot Provide an Exclusive Guide to Fourth
Amendment Protection
If the proxies cannot provide exclusive guides, the obvious alternative is
the direct approach. At first blush, the policy model looks like a promising test
to determine in all cases when an expectation of privacy is reasonable. This
model restates the basic goal of the reasonable expectation of privacy test,
which must identify police practices that are reasonable per se and separate
them out from more invasive police practices that could only be reasonable in
some specific contexts.
149
If the policy model accurately frames the basic goals
of the reasonable expectation of privacy test, doesn’t the policy model provide
an exclusive guide to whether an expectation of privacy is reasonable?
Shouldn’t the courts apply the policy model in every case?
150
This Part explains that the answer is “no.” The policy model cannot
provide the exclusive guide because lower courts cannot administer it
consistently. The Fourth Amendment implications of most police practices
never reach the Supreme Court; rather, they are decided bottom up from lower
federal courts and state courts.
151
For this decentralized regime to work, the
reasonable expectation of privacy test must be interpreted in a way that
provides reasonably consistent results across different jurisdictions. If the law
is to take shape without centralized guidance, the test must be sufficiently
predictable that results in one jurisdiction are likely to lead to the same results
in another.
149. See supra Part II.A.
150. Professor Amsterdam believed courts could and perhaps should do so. See
Amsterdam, supra note 10. Professor Amsterdam concluded that courts could not be honest
that policy concerns were driving the scope of the Fourth Amendment because police
officers could not implement such policy concerns in the field. See id. at 403-05. This
explanation is weak, however, because the rationale for a court’s decision is distinct from its
resulting rule. That is, courts could easily apply the policy model and then announce a rule;
the officers would then follow the rule and could safely ignore the policy argument that led
to it. The difficulty with exclusive reliance on the policy model has a different source.
151. For the distinction between a “top down” and “bottom up” regulatory system, see
generally Richard A. Posner, Legal Reasoning from the Top Down and from the Bottom Up:
The Question of Unenumerated Constitutional Rights, 59 U.
CHI. L. REV. 433 (1992). As
Judge Posner explains:
In top-down reasoning, the judge or other legal analyst invents or adopts a theory about an
area of law—perhaps about all law—and uses it to organize, criticize, accept or reject,
explain or explain away, distinguish or amplify the existing decisions to make them conform
to the theory and generate an outcome in each new case as it arises that will be consistent
with the theory and with the canonical cases, that is, the cases accepted as authoritative
within the theory. . . . In bottom-up reasoning, which encompasses such familiar lawyers’
techniques as “plain meaning” and “reasoning by analogy,” one starts with the words of a
statute or other enactment, or with a case or a mass of cases, and moves from there-but
doesn’t move far . . . .
Id. at 433. See also Ronald J. Allen & Ross M. Rosenberg, The Fourth Amendment and the
Limits of Theory: Local Versus General Theoretical Knowledge, 72 S
T. JOHN’S L. REV. 1149
(1998).
November 2007] FOURTH AMENDMENT MODELS 537
The policy model fails to provide that consistency because it provides no
objective standards for what “practice” a court must evaluate. Fourth
Amendment cases involve discrete sets of facts in which particular officers did
particular things in particular ways. But the policy model requires balancing
over an arbitrary set of hypothetical facts. When a court applies the policy
model, it evaluates whether a particular range of policy practices should be
regulated. No method exists for identifying the proper range, however, which
means that no two courts are likely to pick the same range. This leads to
unusually inconsistent rules: different ranges will produce different balances
and result in different rules of different scope. As a result, the policy model
cannot provide an exclusive guide to Fourth Amendment protection.
1. Lower Courts and the Reasonable Expectation of Privacy Test
Police practices are, in Anthony Amsterdam’s phrase, “a perpetual
Pandora’s box.”
152
The Fourth Amendment applies to all of the eight million
government employees in the United States,
153
including the more than eight
hundred thousand law enforcement officers who have powers to make arrests in
criminal cases.
154
Every year, those officers conduct hundreds of millions of
criminal investigations, leading to about fourteen million arrests every year.
155
Most of those investigations will involve multiple steps potentially subject to
Fourth Amendment regulation. And the steps the police may take in those
investigations are incredibly diverse, offering every imaginable fact pattern
under the sun.
Consider an officer “walking the beat” in an urban neighborhood. The
officer might walk down the street, knock on doors, enter stores, look in shop
windows, check out an alleyway, rattle a door handle, look through trash, peer
in a parked car, talk to witnesses, inquire of suspicious activity, ask someone
for identification, or put someone in handcuffs. Similarly, consider an officer
surveilling a home for evidence of illegal activity. The officer might watch the
street from his parked squad car, use a flashlight to peer in a window, use a
“shotgun” microphone to hear what is inside, get electricity and telephone
records from the home, knock on the door in an undercover capacity, or find
152. See Amsterdam, supra note 10, at 386-87 (“[T]he police engage in a vast range of
activities affecting a broad spectrum of citizens’ interests in a complex variety of ways.”).
153. According to the U.S. Census Bureau, in 2006 there were 5.1 million state
government employees, see State Government Employment Data: March 2006,
http://ftp2.census.gov/govs/apes/06stus.txt, and 2.7 million federal government employees,
see Federal Government Civilian Employment by Function: December 2006,
http://ftp2.census.gov/govs/apes/06fedfun.pdf.
154. Bureau of Justice Stats., U.S. Dep’t of Justice, Law Enforcement Statistics,
http://www.ojp.usdoj.gov/bjs/lawenf.htm
155. According to FBI figures, there were about fourteen million arrests in the United
States in 2006. F
ED. BUREAU OF INVESTIGATION, CRIME IN THE UNITED STATES tbl.29 (2006),
available at http://www.fbi.gov/ucr/cius2006/data/table_29.html.
538 STANFORD LAW REVIEW [Vol. 60:503
out what mail is being delivered. An FBI agent investigating a CEO for insider
trading might tap the CEO’s phone, obtain his bank records, access his
computer files, or search his desk drawers. In all of these settings, the
government’s investigation will be governed primarily by a single
constitutional command: the Fourth Amendment’s prohibition on unreasonable
searches and seizures. The Fourth Amendment must contain rules to govern all
of these steps.
Determining which of these police practices infringe a reasonable
expectation of privacy begins—and in most cases ends—with the lower courts.
When the police begin using a new law enforcement practice that leads to
evidence in criminal cases, decentralized lower courts will begin applying
established precedents from analogous contexts and will announce rules that
govern the new practice. If a government practice is covered by a preexisting
rule, then courts will simply apply that rule. But if the practice is not already
covered by a rule, individual lower courts will draw analogies to existing
precedents and will announce rules as to whether and when the technique
violates a reasonable expectation of privacy. A federal district court in Ohio
might start with one decision; a state appellate court in Alabama might be next.
Lower courts’ decisions will begin to dot the landscape. If the rules announced
in the early decisions are consistent across decisions, investigators nationwide
will come to rely on them. Treatises, police manuals, and other doctrinal
guidance will consider them established law even if no binding guidance exists
in a particular officer’s jurisdiction.
156
The Supreme Court cannot generate more than a small fraction of these
rules. The high Court is authoritative when it speaks, but it speaks only
rarely.
157
In recent years, the Court has agreed to review only about eighty
cases every year from the 8,000 or so petitions filed in all areas of federal
law.
158
In a typical year, a handful of those cases concern some aspect of the
Fourth Amendment. And on average, only about one case per year concerns
some aspect of the reasonable expectation of privacy test.
159
Further, the
Supreme Court’s occasional guidance tends to be very narrow. For example,
Illinois v. Caballes created a rule for dog sniffs by well-trained dogs, but only
156. Further, heavy reliance on treatises such as Wayne R. LaFave’s Search and
Seizure, supra note 131, means that as soon as a rule is settled enough to appear in LaFave’s
treatise, many courts will simply treat it as “the law.”
157. Cf. Paul M. Bator, What Is Wrong with the Supreme Court?, 51 U. PITT. L. REV.
673, 678 (1990) (“And what is the capacity of the Supreme Court? Like it or not, it is finite.
Over the years, we have learned that the United States Supreme Court cannot be asked to
decide more than 150 to 175—200 at the most—cases with full opinion in a year.”).
158. See generally Philip Allen Lacovara, The Incredible Shrinking Court, AM. LAW.,
Dec. 2003, at 53, 54.
159. A Westlaw search for the phrase “reasonable expectation of privacy” in the SCT
database yields 81 hits as of March 1, 2007. About half of these cases are decisions that
actually apply the test as compared to those that simply mention the phrase in the course of
discussing another topic.
November 2007] FOURTH AMENDMENT MODELS 539
dog sniffs by well-trained dogs.
160
Kyllo v. United States created a rule for
sense-enhancing devices not in general public use directed at a home, but its
rule doesn’t cover sense-enhancing devices directed at a car or person nor
devices that are in general public use.
161
Because the Supreme Court’s cases
are so few and so narrow, the Supreme Court’s decisions cover only a tiny
sliver of fact patterns common in police investigations. Absent substantial
conflicts in the lower courts, the Supreme Court normally will stay away and
let the lower courts have the final word.
162
In all but extraordinary
circumstances, the meaning of the reasonable expectation of privacy test
evolves case by case in lower court decisions without any Supreme Court
involvement.
The rarity of Supreme Court intervention places a premium on consistency
among the lower courts. Lower court agreement effectively settles the rule for a
particular technique, inducing reliance among law enforcement and making it
unlikely the Supreme Court will ever address the issue.
163
In contrast,
inconsistent lower court decisions leave investigators in most jurisdictions
unsure of whether particular techniques can be used.
164
If a district court in
New Jersey says that a technique violates a reasonable expectation of privacy
but a state appellate court in New Hampshire disagrees, those decisions will
give no guidance to police officers in California or Texas or New York.
Investigators need rules, and inconsistent rules from other jurisdictions are
equivalent to no rules at all. As a result, the reasonable expectation of privacy
test must provide sufficient guidance that an officer in one jurisdiction can
normally expect that an open issue of Fourth Amendment law in his own
jurisdiction will be decided the same way as precedents in other jurisdictions.
2. The Instability of the Policy Model in the Lower Courts
The policy model cannot provide this consistency because it requires courts
to identify a particular “practice,” and that practice always can be defined more
narrowly or more broadly. A Fourth Amendment case always involves specific
facts. But the policy model does not: it asks courts to assess whether a
particular set of practices require regulation, inviting a balancing of interests
160. 543 U.S. 405 (2005).
161. 533 U.S. 27 (2001).
162. See H.W. PERRY, JR., DECIDING TO DECIDE: AGENDA SETTING IN THE UNITED
STATES SUPREME COURT 246, 251 (1991). The Supreme Court agreed to hear only about one
percent of the cases it is petitioned to review. By far the most important factor in whether the
Supreme Court agreed to take a case is whether the lower courts have divided on how the
law applies to that set of facts.
163. Absent a lower court division, the chance that the Supreme Court would agree to
hear a case to resolve whether a police practice violated a reasonable expectation of privacy
is extremely low.
164. If a split exists in foreign jurisdictions, the officer will not have any assurance that
a court would rule in the government’s favor.
540 STANFORD LAW REVIEW [Vol. 60:503
over the range of those facts that fall within the defined practice. Courts can
select a very narrow range of facts that define the practice under consideration,
such as only those facts exactly like the case before them. Alternatively, they
can pick a very broad range of facts or some arbitrary subset in between. No
one choice is “correct,” as law enforcement practices are not self-defining. As a
result, each court that applies the policy model must make an essentially
arbitrary decision as to the scope of the law enforcement practice that it is
considering.
This renders the policy model highly unstable. Judges can reach whatever
result they like by simply manipulating the realm of cases for which the
balance will be drawn.
165
A judge committed to ruling for the government
could apply the policy balance over a realm of facts that happen to share strong
government interests and lesser privacy concerns, of which the facts before the
court happen to be at the outer bounds. On the other hand, a judge eager to rule
for the defense in the same case could apply the same balance over a realm of
facts with lesser government interests and greater privacy concerns, albeit one
that also happens to include the case before it. The result would be two
completely different rules with different results and different scope even though
the two judges followed the same framework and shared the same assessments
of the privacy and security interests.
Consider an example. Imagine that a lower court must decide whether a
suspect has a reasonable expectation of privacy in his work e-mail account, a
question that is highly uncertain under existing Fourth Amendment doctrine.
166
Let’s say that the suspect works in the Manhattan office of a Canadian
company that has its computer servers in Toronto, and that the company’s
workplace e-mail policy states that employees should not expect any privacy in
their work e-mail. Let’s also say that despite the formal policy, most employees
at the company use their work e-mail for personal reasons and treat their work
e-mail as fully private. The police believe that the suspect committed serious
crimes and left evidence in his work e-mails, and they obtain copies of the
suspect’s e-mails from the server in Canada without first obtaining a warrant.
The issue before our hypothetical court is whether the suspect has a reasonable
expectation of privacy in his e-mails.
In a world governed exclusively by the policy model, the correct answer
depends entirely on the category of cases the court chooses to group together. If
the court defines the category as all cases involving the contents of e-mails,
then in all likelihood it will conclude that e-mail does merit a reasonable
expectation of privacy. Most people consider their e-mail as highly private and
deserving of strong privacy protection, so a balance of interests will yield the
conclusion that an expectation of privacy in e-mail is constitutionally
165. This is limited by the possibility that the court’s rule will trample on other
holdings, however.
166. See generally O
RIN S. KERR, COMPUTER CRIME LAW 394-445 (2006).
November 2007] FOURTH AMENDMENT MODELS 541
reasonable. But changing the category changes the balance. For example,
imagine the court picks a narrower category such as e-mail stored outside the
United States, or employee e-mail covered by a formal policy announcing that
employees have no privacy rights. Or imagine the category only includes e-
mail both stored abroad and covered by the workplace monitoring policy. The
policy balance now shifts: the privacy interests are lesser within the subset of
cases carved out by the narrower category, and a strong argument exists that
within that category government interests outweigh privacy interests. Under
that approach, the suspect should have no reasonable expectation of privacy in
his e-mail because his case falls within a grouping that triggers lesser privacy
interests.
Further, the categories are almost infinitely malleable. In the previous
paragraph, the narrower range of facts led to no protection. But we could easily
reverse this in the same hypothetical and find a broad category that is less
protective than a narrow one. For example, a court could define the appropriate
category with ever greater scope as all Internet communications generally,
which would suggest that there should be no expectation of privacy. Or else it
could draw the category narrowly as only e-mails that a user believes are
private, a narrow category that would suggest that an expectation of privacy
would in fact be reasonable. The outcome of the balance depends almost
entirely on the selection of what facts are balanced.
If the Supreme Court decided every Fourth Amendment case, the
malleability of the policy model might be a feature rather than a bug. It would
let the Justices define the precise set of cases they want covered by a particular
legal rule or standard.
167
The rule might be that there are no Fourth
Amendment rights in prison (as in Hudson v. Palmer),
168
or that the police
violate a reasonable expectation of privacy when they use sense-enhancing
devices not in general public use to obtain details of the home not otherwise
available without physical penetration of the home (as in Kyllo v. United
States).
169
Whether these rules are correct or not, they impose a settled rule or
standard for a given set of cases. At the Supreme Court level, the policy model
lets the Court define an arbitrary range of facts and impose a rule that covers
those facts.
But in a world of decentralized lower court rulemaking, reliance on the
policy model is a recipe for doctrinal chaos. No two lower courts would be
likely to pick the same category, resulting in a blizzard of inconsistent rules and
167. Consider the precision of Justice Scalia’s rule in the thermal imaging case, Kyllo
v. United States, 533 U.S. 27 (2001). Scalia’s rule announces that the police violate a
reasonable expectation of privacy when they use (a) sense-enhancing devices, (b) not in
general public use, (c) to obtain details of the home not otherwise available without physical
penetration of the home. See id. at 34. This is a carefully articulated rule, reflecting a
carefully chosen category under the policy model.
168. 468 U.S. 517 (1984).
169. 533 U.S. 27.
542 STANFORD LAW REVIEW [Vol. 60:503
no predictability in any courts where the rule was not already settled. Different
panels of different courts would pick different categories, leading to different
rules in different jurisdictions. Officers in jurisdictions without a binding rule
would have no way of predicting which way the courts would rule, defeating
the primary interest in settling rules clearly. And because cases within a
jurisdiction would build on other cases in that jurisdiction, the law in each
jurisdiction would evolve separately. Even assuming that every judge shares
exactly the same values and beliefs, each court would pick a different realm,
leading to tremendous uncertainty from state to state and circuit to circuit.
D. The Case for Multiple Models
Multiple models provide a critical benefit over a single model. Whereas no
one model suffices in every case, the use of multiple models permits the
Supreme Court to generate localized guidance.
170
Different models accurately
draw the line between reasonable per se and contextually reasonable police
practices in different settings. Using multiple models lets the Justices explain
decisions using the models that best draw the line in that particular setting.
Further, certainty among lower courts can result because lower courts naturally
adopt the Supreme Court choice of model when reasoning by analogy.
Common techniques of analogical reasoning lead lower courts to reflect the
Supreme Court’s choices of models in analogous cases, leading both to the
adoption of similar rules across decentralized courts and localized guidance that
helps accurately distinguish per se from contextually reasonable law
enforcement practices.
The result is a system of localized control of the various models, with
different models controlling outcomes in different contexts depending on which
models best draw the line in each context. While no one model works in every
case, the use of multiple models allows the courts to use the model that best
identifies police practices in need of regulation in that context. This goes on to
a modest extent already, albeit without judicial recognition of the four models.
170. In an important article, Ronald Allen and Ross Rosenberg have argued that
Fourth Amendment law consists of “localized knowledge,” individual rules that are very
clear in context but that lack a single top-down theory. See Allen & Rosenberg, supra note
151. I agree in part. Like Allen and Rosenberg, I see the development of Fourth Amendment
law as largely a bottom-up rather than top-down process. I also agree that localized rules are
a critical feature of Fourth Amendment doctrine. My approach differs in two essential ways,
however. First, unlike Allen and Rosenberg, I do see a unifying goal that drives judges and
Justices in Fourth Amendment cases. Allen and Rosenberg envision Fourth Amendment
rules as organic, as the products of a grown system rather than a made one. See id. at 1198-
99. In contrast, I think the system is made, but simply in a piece-by-piece way; it is made to
create reasonable police practices in a system of exclusionary remedies. Second, I use the
phrase “bottom-up” somewhat differently than do Allen and Rosenberg. Allen and
Rosenberg focus, like most commentators, on the Supreme Court. In contrast, I mean
“bottom-up” to refer literally to cases coming from the decentralized lower courts.
November 2007] FOURTH AMENDMENT MODELS 543
And it would work even more effectively if both judges and Justices recognized
the four models and chose them consciously with the ends of Fourth
Amendment protection in mind.
1. Supreme Court Selection Among the Four Models
The fact that different models provide accurate proxies in different contexts
points to the primary benefit of using multiple models. When confronted with a
new case, the Supreme Court can pick the model that best tracks the line
between less troublesome and more troublesome police practices in that
particular context. If the positive law model provides an accurate and consistent
proxy, the Court can apply the positive law model; if the private facts model
provides an accurate and consistent proxy, it can rely on the private facts
model. The coexistence of multiple models allows the Supreme Court to select
the model that works best in the context of that particular case. Indeed, we can
see this already in the Supreme Court’s Fourth Amendment decisions.
Although the data points are too few to develop a strong empirical claim, it
appears that the choices of models have not been not random. Reviewing the
Court’s cases reveals that its choice of models correlate reasonably well with
the contexts in which those models accurately help distinguish police practices
that are reasonable in every case from practices that are reasonable only if the
police have a warrant.
For example, the private facts model appears particularly often in cases
involving new technologies. The private facts model was used to regulate
chemical tests for drugs in United States v. Jacobsen,
171
taking photographs
made during aerial surveillance in Dow Chemical Co. v. United States,
172
and
the use of tracking devices in United States v. Karo
173
and United States v.
Knotts.
174
In all of these cases, the government used tools to obtain information
that may not have been perceptible with unassisted human observation.
175
Why
did the Justices pick the private facts model in technological surveillance
cases? It seems unlikely that they thought consciously in terms of different
“models.” However, they no doubt looked at the many existing precedents with
their many inconsistent rationales and picked the set of arguments that made
the most sense in context. Relying on probabilistic or positive law approaches
in cases involving technological surveillance would have an obvious drawback:
technological evolution renders the probabilistic model unpredictable, and
technology destabilizes the link between privacy and positive law. In contrast,
171. 466 U.S. 109 (1984).
172. 476 U.S. 227 (1986).
173. 468 U.S. 705 (1984).
174. 460 U.S. 276 (1983).
175. In contrast, the Court has rejected the private facts approach in cases involving
the physical moving and opening of physical things such as packages. See supra notes 58-59
and accompanying text (discussing Arizona v. Hicks, 480 U.S. 321 (1987)).
544 STANFORD LAW REVIEW [Vol. 60:503
the private facts model works independently of the technology and thus permits
a stable rule that remains constant as technology changes. Given that, it should
be unsurprising that the Supreme Court gravitated towards the private facts
model in cases that involve technological surveillance.
The probabilistic model appears to surface mostly in investigations that
occur in group settings. The Court relied on the probabilistic model in Bond v.
United States, the case in which an officer walked the length of the bus and
manipulated all of the passengers’ luggage.
176
The Court also relied on the
probabilistic model in Minnesota v. Olson, the case involving Fourth
Amendment rights of overnight guests.
177
In that case the Court considered the
social practices and expectations of guests together with hosts, and reached a
rule relying on the probabilistic model. Why might the Justices favor the
probabilistic model in group settings? The likely reason is that practices in
group settings tend to hinge on social norms that are difficult for the
government to manipulate. In a group setting, social practice will tend to reflect
the invasiveness of a particular technique relatively directly: if an unusual
technique leads to the discovery of evidence, it is likely that the technique was
also unusually invasive. Further, as social practices and actual expectations of
privacy change, the rules can change along with them under the probabilistic
model. In that context, the probabilistic model likely will prove a more stable
and clear basis for distinguishing regulated from unregulated law enforcement
techniques.
The positive law model tends to govern physical access to houses,
packages, letters, and automobiles.
178
If a person owns a package, letter or car,
physically opening it up will violate his reasonable expectation of privacy; if a
police officer merely looks at these objects from the outside, but does not
physically access them, it will not.
179
The selection of the positive law model
in physical entry cases makes sense: it provides clear and familiar ex ante
guidance for police, and in this context it resonates with our intuitions as to
what kind of investigative steps are only modestly invasive and what steps are
highly invasive. It is common to hide private materials in our property; access
to our property seems highly invasive. In contrast, merely observing the outside
of our property seems much less invasive; in the case of physical access,
176. 529 U.S. 334 (2000).
177. 495 U.S. 91 (1990).
178. See, e.g., Rakas v. Illinois, 439 U.S. 128 (1978) (applying positive law model to
search of an automobile).
179. Once again, the trend has exceptions. Most notably, the open fields doctrine
permits a great deal of trespass on private property, justified under the private facts model
instead of the positive law model. See Oliver v. United States, 466 U.S. 170 (1984) (arguing
that walking on open fields did not violate a reasonable expectation of privacy because open
fields were not likely to be places in which the police would uncover private facts). It’s
actually somewhat unclear whether this rationale counts as the private facts model or the
policy model: it uses the language of the private facts model, but at a macro scale it seems
most at home in the policy model.
November 2007] FOURTH AMENDMENT MODELS 545
property law seems like a reasonably accurate guide to the invasiveness of
government conduct.
Finally, when none of these three approaches provides a sensible
mechanism for drawing lines between regulated and unregulated police
practices, the Court has tended to embrace the policy model to achieve the
desired results directly. Consider Hudson v. Palmer, the case about searching
prison cells.
180
The Supreme Court created the sensible rule that inmates do not
have Fourth Amendment rights in their cells under any circumstances. To do
this, the Court relied on a broad application of the policy model: society’s
interest in order in penal institutions outweighed the inmates’ privacy
interests.
181
That result would have been more difficult to reach under the other
three models, as each would have called for a more fact-sensitive inquiry that
would have led to uncertainty and frequent litigation. Although many prison
searches are common, presumably some are not, creating uncertainty under the
probabilistic model. Prisoners may have some property in their cells, creating
uncertainty under the positive law model, and a prison cell search may uncover
private items such as a diary, creating ex ante uncertainty under the private
facts model. Instead of becoming bogged down with these difficulties, the
Court simply bypassed them and created the desired clear rule directly under
the policy model.
To be clear, I’m not suggesting that these categories are firm or clear. It
seems that the Justices have been at most only vaguely aware of the different
models, so it would be quite surprising if their past decisions neatly lined up
based on specific categories. My point is more modest: different models fit
better or worse in different contexts, and past Supreme Court decisions appear
to reflect this even without expressly acknowledging the different models.
2. Lower Court Use of the Four Models
The fact that the Supreme Court can pick whatever model works best in
context is not by itself a particular strength of the multiple models approach.
However, it does become a particular strength when paired with lower court
applications of the reasonable expectation of privacy test. Recall that the
Supreme Court decides very few applications of the reasonable expectation of
privacy test; for the most part, rules evolve in the lower courts. The use of
multiple models in Supreme Court opinions provides a way that the law can
develop consistently and accurately in the lower courts. When novel issues
arise in the lower courts, lower courts will naturally reason by analogy to the
Supreme Court’s decided cases.
Analogical reasoning will focus lower court judges on the reasoning of the
Supreme Court’s analogous decisions, which will in turn push the lower courts
180. 468 U.S. 517 (1984).
181. Id. at 526.
546 STANFORD LAW REVIEW [Vol. 60:503
to apply the same models that the Supreme Court chose in that context. The
models will propagate to analogous sets of facts. When this process is repeated
over time, the Fourth Amendment tapestry evolves into a system of regional
dominance—regional in the sense of categories of cases, not geography—of
different models of when an expectation of privacy is reasonable. Different
models will apply in different contexts based on the models that the Supreme
Court picked in analogous cases. This leads to both increased accuracy and
increased consistency in the evolution of the reasonable expectation of privacy
test. Lower court cases will often draw the same analogies and then pick the
same models, leading to greater consistency and predictability in the lower
courts.
Importantly, this process can occur without conscious recognition of the
models. When a Fourth Amendment challenge involves a new police practice
unsettled by existing doctrine, litigants and courts will naturally turn to the
facts—and then to the reasoning—of the most analogous Supreme Court
decisions. If the Supreme Court applied the private facts model in the
analogous case, the lower court will look to that opinion and mirror the
reasoning of the private facts model. If the Supreme Court relied on a particular
model, the lower court will do the same. Through the process of analogical
reasoning, the Supreme Court’s choice of model will guide lower court judges
by providing different styles of reasoning that should apply in different types of
cases. If the Supreme Court’s chosen model accurately identifies police
practices in need of regulation, lower court applications of the same model in
similar cases should also draw the line accurately in the lower courts. Further,
the Supreme Court’s selection of model will ensure that lower courts all draw
the same or similar lines. Although the selection of the model does not
guarantee a result, it narrows the range of discretion and greatly increases the
likelihood that different jurisdictions will reach the same outcome.
The Supreme Court’s certiorari practice completes the feedback loop. If the
lower courts all apply the models in the same way to new facts, then the
Supreme Court is unlikely to intervene. When lower courts disagree on which
model applies or how it applies, leading to different rules, the Supreme Court
will usually take the case. The Court will address the circuit split, and resolve
how the four models apply to that case. The Supreme Court’s certiorari
practices naturally police lower court applications of the four models, and
uncertainty as to a model or its application tends to trigger Supreme Court
review.
The road to the Supreme Court’s thermal imaging decision, Kyllo v. United
States,
182
demonstrates how this works. The first use of thermal imaging
devices to detect the temperature of outside walls dates back to the early
1990s,
183
and lower court cases appeared regularly from that time until the late
182. 533 U.S. 27 (2001).
183. The first recorded decision on the Fourth Amendment implications of a thermal
November 2007] FOURTH AMENDMENT MODELS 547
1990s. The facts of most of the cases resemble those of Kyllo: narcotics
investigators used the devices to measure the temperature of exterior walls, and
used that information to help obtain a warrant. The defendants moved to
suppress, arguing that use of the imaging devices had violated their reasonable
expectations of privacy.
Such challenges raised a very difficult question not readily answered by
existing precedents, so lower courts relied on two analogies to find their way. A
few decisions analogized use of a thermal imaging device to garbage left on the
street as in California v. Greenwood.
184
These courts applied the positive law
model and ruled that the devices did not violate a reasonable expectation of
privacy. Just as Greenwood had left her trash out in the open where anyone
could get it, so had homeowners exposed their heat to the public.
185
Other cases
instead analogized thermal imaging devices to other technological surveillance
cases, such as cameras in aerial surveillance in Dow Chemical Co. v. United
States
186
and electronic tracking devices in United States v. Knotts
187
and
United States v. Karo.
188
These Supreme Court decisions relied on the private
facts model, so the lower courts applied the private facts model to thermal
imaging devices. The federal courts of appeals that addressed the issue
concluded that thermal imaging devices did not reveal sufficiently intimate
details of the home to violate a reasonable expectation of privacy.
189
At least in federal court, the two sets of analogies led to two different
models that happened to lead to the same result. If no courts had taken a
different path, the Supreme Court presumably would have not intervened and
the law surrounding thermal imagining devices would have become settled by
the lower courts. The constitutionality of using thermal imaging devices
became a rare subject for Supreme Court review after a few courts took a
different view. Several state courts applied the private facts model and
concluded that thermal imaging devices did in fact reveal sufficiently private
facts to implicate a reasonable expectation of privacy.
190
The split between
federal and state cases prompted Supreme Court review. The Court then
imaging device appears to be United States v. Penny-Feeney, 773 F. Supp. 220 (D. Haw.
1991), which involved use of a thermal imaging device in April 1990.
184. 486 U.S. 35 (1988); see, e.g., United States v. Myers, 46 F.3d 668 (7th Cir. 1995);
State v. Siegal, 934 P.2d 176 (Mont. 1997).
185. See, e.g., Myers, 46 F.3d 668.
186. 476 U.S. 227 (1986).
187. 460 U.S. 276 (1983).
188. 468 U.S. 705 (1984).
189. See United States v. Robinson, 62 F.3d 1325 (11th Cir. 1995) (upholding
warrantless use of a thermal imager); United States v. Ishmael, 48 F.3d 850 (5th Cir. 1995);
Myers, 46 F.3d 668; United States v. Pinson, 24 F.3d 1056 (8th Cir. 1994).
190. See, e.g., Commonwealth v. Gindlesperger, 743 A.2d 898, 906 (Pa. 1999) (“[W]e
conclude that Appellee met the requirements of Katz and thus established that a search
implicating the Fourth Amendment occurred when law enforcement agents scanned his
home with the WASP device.”).
548 STANFORD LAW REVIEW [Vol. 60:503
resolved the question with a rule that settled the question and provided
additional guidance for how to resolve cases in analogous contexts.
191
3. The Need for Recognition of the Four Models
Multiple models of Fourth Amendment protection provide a major
advantage over a single model. They allow the Supreme Court to direct the use
of particular models in different contexts, depending on which model most
accurately identifies police practices in need of constitutional regulation. When
judges and Justices are unaware of the four models, as they generally have been
in the past, this process works modestly well. In the course of writing opinions,
Supreme Court Justices naturally rely on precedents based on theories that
create sensible results in analogous cases. Lower court decisions naturally will
analogize to Supreme Court opinions, adopting the models and using them in
new and similar cases. As a result, the four models will help courts draw the
needed lines even when courts and litigants are not fully aware of the models
and how they work.
Explicit recognition of the different models would help the process work
more effectively. The Supreme Court’s Fourth Amendment decisions
sometimes have a shotgun quality: the Justices use any model they can to
justify the selected result. The Court should realize that when it applies the
reasonable expectation of privacy test, it faces a critical choice as to which
model or models to use. The Supreme Court’s choice of models in one case
won’t bind future courts to the same choice; rather, the Court must always
make a context-sensitive selection of which model to use to explain the Court’s
result in a particular case. The Court should pick the model that in context best
distinguishes less troublesome police practices that need not be regulated from
more troublesome practices that are only reasonable when accompanied by a
warrant or special circumstances. Furthermore, the Court’s choice of model
should be as clear as possible. By choosing the model or models, the Court’s
opinion will help frame the choice of models in analogous cases decided by the
lower courts. To facilitate lower court decision making, Supreme Court
opinions should pick models clearly.
Lower courts faced with novel Fourth Amendment questions should apply
analogies to the Supreme Court’s cases with the four models explicitly in mind.
Lower court judges should study the most analogous Supreme Court opinions
and note the models they used; the lower courts should then apply the Supreme
Court’s choice of models in those specific cases (with the possible exception of
the policy model, which is almost universal in Supreme Court opinions but is
used only relatively rarely in lower court decisions). This would mean a
modified approach among those lower courts that have tried to encapsulate the
“reasonable expectation of privacy” inquiry as a multifactor test. In the Fifth
191. United States v. Kyllo, 533 U.S. 27 (2001).
November 2007] FOURTH AMENDMENT MODELS 549
Circuit, for example, the Katz test has been reduced to a multifactor test with
“factors” drawn from explanations in the case law.
192
The Fifth Circuit has
noted that different factors are more or less important in different contexts,
193
which is a step towards recognizing that the Fourth Amendment test includes
localized rules where different models predominate. But the Fifth Circuit and
other courts should drop the “multifactor test” entirely. The reasonable
expectation of privacy test is simply not one test: it takes on different meanings
in different contexts based on the direction of the Supreme Court’s most
analogous decisions in that context. To apply the test in a new set of
circumstances, courts should apply the reasoning and model selection of the
most analogous precedents rather than look to all of the various “factors” that
have been used in the case law taken as a whole.
C
ONCLUSION
The reasonable expectation of privacy doctrine is the most often
misunderstood aspect of Fourth Amendment law. Many imagine the phrase
“reasonable expectation of privacy” as a test itself, and they mull over the
words “reasonable,” “expectation,” and “privacy” as if a deep understanding of
these three words will explain how the law applies. Their error is in assuming
that the reasonable expectation of privacy doctrine is a single test, one top-
down principle that applies to different facts to produce the correct answer. The
reasonable expectation of privacy instead is decentralized; it takes on different
meanings in different contexts. The exclusionary rule and the diverse facts of
investigations prevent the adoption of a single top-down test. Although the
single goal of distinguishing per se reasonable and contextually reasonable
police practices drives the doctrine, implementing that goal through
decentralized courts makes adoption of a single test impossible.
The use of multiple models appears at first blush to be inexcusable.
Multiple models seem like a recipe for mass confusion. From a bottom-up
perspective, however, the use of multiple models makes perfect sense. The law
evolves gradually, case by case, in a decentralized layering process of settled
rules building on settled rules. Confusion in theory doesn’t amount to confusion
in practice, as the justifications for the rules are distinct from the rules
themselves. Once the rules are settled, the justifications for them no longer
matter: what matters is how the Court’s choice of models will help influence
192. See, e.g., United States v. Runyan, 275 F.3d 449, 457 (5th Cir. 2001) (“[W]hether
an interest is protected by the Fourth Amendment depends on five factors: (1) whether the
defendant has a [property or] possessory interest in the thing seized or the place searched, (2)
whether he has the right to exclude others from that place, (3) whether he has exhibited a
subjective expectation of privacy that it would remain free from governmental intrusion, (4)
whether he took normal precautions to maintain privacy, and (5) whether he was legitimately
on the premises.” (internal quotations omitted)).
193. See id.; see also Kee v. City of Rowlett, 247 F.3d 206, 213 (5th Cir. 2001).
550 STANFORD LAW REVIEW [Vol. 60:503
courts in new cases not yet covered by rules. Using multiple models helps
lower courts adopt new rules that better track the line between per se
reasonable and contextually reasonable police practices in particular contexts. It
facilitates decentralized Fourth Amendment rules that are tailored to the needs
of different contexts.
The decentralized nature of the reasonable expectation of privacy test has
important consequences for the development of Fourth Amendment law. First,
it means that existing doctrine, once settled, is quite difficult to change. When
the Supreme Court hands down a new decision applying a particular model,
that decision is likely to have only very narrow importance. It may influence
decisions in closely related cases, but it is unlikely to impact rules in other
areas. This presumably explains why the Supreme Court’s creation of the
reasonable expectation of privacy test itself had little effect on the scope of
Fourth Amendment protection. As I have noted elsewhere, the Katz revolution
was more a revolution on paper than in practice.
194
And it also explains why
symbolic cases like Katz and Kyllo have had little impact on Fourth
Amendment law as a whole: the real fight is over the new fact patterns not
settled by existing rules, and the same need to distinguish between per se and
contextually reasonable police practices exists regardless of the Supreme
Court’s latest decision. What the Supreme Court says in individual cases
matters, but it matters much less than most commentators assume it does.
Second, the decentralized nature of the reasonable expectation of privacy
test suggests a potential new direction for legal scholarship. In the past, most
scholars have assumed that the Katz test is a single test. As a result,
commentators unsatisfied with existing law routinely urge the Supreme Court
to adopt a meaning for Katz (or a replacement for it, depending on what they
think Katz means) based on whatever normative principles they think would
work best. Under the decentralized Fourth Amendment, however, such calls
will necessarily fall on deaf ears. The Justices simply cannot embrace a single
top-down approach regardless of how protective it may be. Instead of offering
new top-down models, scholars should recognize that the principles guiding
what is a “search” must necessarily be decentralized. Scholarship should focus
on discrete law enforcement practices in the lower courts rather than grand
theories at the Supreme Court and should make the case for applying different
models to different practices.
Finally, the four models of Fourth Amendment protection should have
important implications for how the courts apply and explain the reasonable
expectation of privacy test. Currently the Justices mix and match models with
only a modest awareness that they are shifting from one argument to another.
The Justices should recognize and identify the models clearly, and should
consider directly which models are likely to best distinguish police practices
that are reasonable per se from police practices that are reasonable only with a
194. Kerr, supra note 64, at 816.
November 2007] FOURTH AMENDMENT MODELS 551
warrant or an exception in that category of case. Explicit use of the different
models will increase the likelihood that the Justices select models that draw the
needed line accurately and will also increase certainty ex ante for the police and
lower courts as to which model will apply in what context. Lower courts will
be more attuned to the Court’s choice of models, furthering the signaling
function, and the police will have greater certainty as to how the courts might
apply the Fourth Amendment to new techniques not covered by existing law.
Recognition of the four models will permit the Justices to better frame their
decisions, both serving the goals of the Fourth Amendment and providing
better guidance to the police and the public.
552 STANFORD LAW REVIEW [Vol. 60:503
