Wednesbury Unreasonableness

Adam Perry★

Abstract. Administrative decisions are unlawful if they are unreasonable, in the

sense that Associated Provincial Picture Houses Ltd v Wednesbury Corporation made
famous. What is Wednesbury unreasonableness, precisely? Courts have not
clearly said, and existing academic answers are flawed. Here I propose a new
answer. My claim, roughly, is that a Wednesbury unreasonable decision is one
that a court is entitled, given the evidence before it, to conclude was wrong,
given the evidence before the authority when it made the decision. In a slogan:
Wednesbury unreasonableness is demonstrable wrongness.

I. WHAT IS UNREASONABLENESS?

In 1995, members of the Orange Lodge, a Northern Irish unionist group,

assembled at Drumcree parish church. From there they planned to
parade to the centre of Portadown, a route that would take them along
Garvaghy Road, a nationalist area. Before the parade could begin,
nationalist protesters arrived. A standoff ensued and the police
intervened. The conflict grew and spread; before long, there was
disorder and rioting across Northern Ireland.
In the aftermath, Parliament enacted the Public Processions
(Northern Ireland) Act 1998. The Act established a Parades Commission
to mediate parade-related disputes and set conditions for parades,
including in respect of routes. From its outset, the commission was
opposed by unionist groups. That opposition became entrenched when,
year after year, the commission prohibited unionists from parading

★
University of Oxford. For comments I thank Sandy Steel. For conversations about the
topic I thank Hasan Dindjer. This is a draft and some things are missing, including some
citations.

Electronic copy available at: https://ssrn.com/abstract=4207429

down Garvaghy Road. By the early 2000s, the government felt it had to
do something to gain the cooperation of unionists.
In 2005, several vacancies on the commission came open. In addition
to advertising these positions publicly, the Secretary of State for
Northern Ireland wrote to leaders of unionist groups asking them to
encourage their members to apply. Among those who applied were Mr
Burrows and Mr Mackay. Both men were active members of the
Portadown Orange Lodge. Both insisted in their applications that they
could decide on parade issues impartially. On the recommendation of
the selection panel, the Secretary of State appointed Burrows and
Mackay as commissioners.
These appointments were challenged by way of judicial review by Mr
Duffy, a member of the Garvaghy Road Residents’ Association. In re
Duffy1 was eventually heard by the House of Lords. Before their lordships
was a single issue: were the appointments unreasonable, in the sense
made famous by Associated Provincial Picture Houses Ltd v Wednesbury
Corporation2? Lord Bingham (with whom the other lords agreed) said
that for the commission to fulfil its functions its members must both be
impartial and appear to be impartial. Burrows and Mackay could be
neither:
No reasonable person, knowing of the two appointees’ backgrounds
and activities, could have supposed that either would bring an
objective or impartial judgment to bear on problems raised by the
parade in Portadown and similar parades elsewhere.3

1
In re Duffy [2008] UKHL 4.
2
Associated Provincial Picture Houses Ltd v Wednesbury Corporation [1948] 1 KB 223.
Lord Greene MR said that administrators act unlawfully if they (1) ‘have taken
into account matters which they ought not to take into account, or, conversely,
have refused to take into account or neglected to take into account matters which
they ought to take into account’; or (2) have ‘come to a conclusion so unreasonable
that no reasonable authority could ever have come to it’ (233-4). On the
distinction between (1) and (2), see e.g. R (Law Society) v Lord Chancellor [2018]
EWHC 2094 (Admin), [98] (Carr J); Braganza v BP Shipping Ltd [2015] UKSC 17,
[24] (Lady Hale). I am concerned only with (2) or with what is sometimes called
Wednesbury in the ‘narrow sense’.
3
In re Duffy [2008] UKHL 4, [27] (Lord Bingham).

Electronic copy available at: https://ssrn.com/abstract=4207429

Appointing representatives of unionist groups was one thing.
Appointing ‘hardline members of the very lodges whose activities had
been a focus … of the serious problems which had caused widespread
disorder’4 was quite another. The appointments were, his lordship held,
Wednesbury unreasonable (“unreasonable” hereafter).
I have no doubt that Lord Bingham’s conclusion was correct. The
Northern Ireland Secretary’s decision to appoint immoderates as
moderators was clearly unreasonable. It is far less clear, at least to me,
why Lord Bingham was correct and why the appointments were
unreasonable. It is all well and fine to be able to recognise decisions as
unreasonable; but, at some point, one wants to be able to say what one is
recognising. What is unreasonableness, precisely?
If one looks to the case law for the answer, I am afraid one will be
disappointed. Wednesbury told us that an unreasonable decision is one
that ‘no reasonable body could have come to’5. That tautology is now 75
years old.6 One might hope the law would be clearer by now. But the
formulations in later cases are not much better. A decision is said to be
unreasonable if it is ‘beyond the range of reasonable decisions’7, ‘beyond
rational justification’8, or beyond what a ‘sensible person’9 would do,
phrases that are hardly self-explanatory. An unreasonable decision is
supposed to be one that ‘lacks logic’10, lacks rationality11, or ‘lacks …
justification’12, as if logic, rationality, and justification are the same
thing. It is also said – and for sheer opacity this is my favourite – that an

4
In re Duffy [2008] UKHL 4, [27] (Lord Bingham).
5
Associated Provincial Picture Houses Ltd v Wednesbury Corporation [1948] 1 KB 223, 230
(Lord Greene MR
6
A Lester and J Jowell, ‘Beyond Wednesbury: Substantive Principles of
Administrative Law’ [1987] Public Law 368, 372.
7
R (Johnson) v Secretary of State for Work and Pensions [2020] EWCA Civ 778, [48]
(Rose LJ); see also General Medical Council v Michalak [2017] UKSC 71, [21] (Lord
Kerr).
8
Kennedy v Charity Commission [2014] UKSC 20, [132] (Lord Toulson).
9
HMB Holdings v Antigua and Barbuda [2007] UKPC 37, [31] (Lord Hope) . See also
Secretary of State for Education and Science v Tameside Metropolitan Borough Council
[1977] AC 1014, 1064 (Lord Diplock).
10
R (A) v Liverpool City Council [2007] EWHC 1477 (Admin), [39] (Walker J).
11
CCSU v Minister for the Civil Service [1985] 1 AC 374 (HL), 410 (Lord Diplock).
12
R (Mackay) v Parole Board [2019] EWHC 1178, [38] (Kramer J).

Electronic copy available at: https://ssrn.com/abstract=4207429

unreasonable decision is whatever causes a judge to think “my goodness,
that is certainly wrong”’13.
I must believe that judges know how little help these brief, cryptic,
and divergent remarks provide.14 And I must believe that judges would
offer something more helpful if they could. The implication is not
reassuring. Judges, it would seem, do not know what unreasonableness
is, or at least not in a way they can articulate clearly.
So I have tried to work out for myself what unreasonableness is. I
think I have succeeded; at the least, I cannot tell where I have failed.
Although my analysis will make more sense with the background in
place, my basic idea is that unreasonableness is a function of two bodies
of material or evidence. The first is the material available to a court
when it reviews the decision. The second is the material available to the
authority when it made its decision. An unreasonable decision is one
that a court is entitled, given the former body of evidence, to consider
wrong, given the latter body of evidence. In a slogan: unreasonable
decisions are demonstrably wrong decisions. I hope to persuade you to
think of unreasonableness this way, too.
Here is how I proceed. Section II explains how we will recognise a
good analysis of unreasonableness when we see one. Section III sets out
my analysis in detail. In section IV I show that my analysis meets the
criteria for a good analysis. I show that no existing academic analysis
does as well in section V. Section VI concludes.

II. WHAT EVERYONE KNOWS

What is the test of a good analysis of unreasonableness? We start with

what everyone familiar with unreasonableness knows about it – that is,
with truisms about unreasonableness. We identify a candidate analysis.
Then we check if it accounts for what we know about unreasonableness.
If it does, then the analysis is successful and our work is done. Otherwise
we must start again, for an analysis that ‘flouts too many [of the truisms]’

13
R v Devon County Council ex p George [1989] AC 573, 583 (Lord Donaldson MR).
14
Cf Dudley Metropolitan Borough Council v Dudley Muslim Association [2014] EWCA
Civ 911, [11] (Sir Stephen Sedley J) (remarking on the ‘impoverished’ state of the
‘Wednesbury jurisprudence’).

Electronic copy available at: https://ssrn.com/abstract=4207429

is an analysis not ‘of the intended entity but something else entirely’15.
While there is no canonical list of truisms about unreasonableness, any
list would include at least the ones below.
First, unreasonableness concerns the balance of reasons for and against a
decision.16 In Tesco Stores v Environment Secretary17, Lord Keith said that ‘it
is entirely for the decision maker to attribute to the relevant
considerations such weight as he thinks fit, and the courts will not
interfere unless he has acted unreasonably in the Wednesbury sense’18. The
implication of the last clause is that unreasonableness regulates the
weight placed on relevant reasons. Later courts are more explicit. For
example, Lord Mance said in Kennedy v Charity Commission that
‘reasonableness review … involve[s] considerations of weight and
balance’.19
Second, unreasonableness is a high standard.20 That means it takes
more than the balance of relevant reasons opposing a decision to make it
unreasonable. It takes more than wrongness, in other words. Lord
Hailsham expressed this thought when he said that ‘[n]ot every
reasonable exercise of judgment is right, and not every mistaken exercise
of judgment is unreasonable’21. So did Lord Ackner when he said that the
Secretary of State’s decision to issue certain directives could be

15
S Shapiro, Legality (Harvard University Press 2011) 14.
16
This point has been made persuasively by P Craig, ‘The Nature of Reasonableness
Review’ (2013) 66 Current Legal Problems 131, 135-142.
17
Tesco Stores v Environment Secretary [1995] 1 WLR 759.
18
Tesco Stores v Environment Secretary [1995] 1 WLR 759, 764 (Lord Keith).
19
Kennedy v Charity Commission [2014] UKSC 20, [54] (Lord Mance); See also Pham v
Secretary of State for the Home Department [2015] UKSC 19, [60] (Lord Carnwath).
Lords Mance and Carnwath both endorse Craig’s analysis, cited at n 16.
20
R v Ministery of Defence, ex p Smith [1996] QB 517 (CA) 558 (Lord Bingham); R v
IRC, ex p Unilever Plc [1996] STC 681 (CA), 692 (Lord Bingham MR); R (Iran) v
Secretary of State for the Home Department [2005] EWCA Civ 982, [11]-[12] (Brooke
LJ); R (Sandiford) v Secretary of State for Foreign and Commonwealth Affairs [2014]
UKSC 44, [66] (Lord Carnwath and Lord Mance).
21
Re W (An Infant) [1971] AC 682 (HL), 700E (Lord Hailsham). See also Secretary of
State for Education and Science v Tameside MBC [1977] AC 1014 (HL), 1074-5 (Lord
Russell);

Electronic copy available at: https://ssrn.com/abstract=4207429

reasonable ‘whether the Secretary of State was right or wrong to issue
[them]’22.
Third, unreasonableness review is neither an appeal nor a form of merits
review.23 A decision is not unreasonable just because a court, based on its
own assessment of the relevant reasons, considers it wrong. A court
must not, in the guise of unreasonableness review, ‘substitute its, the
judicial view, on the merits’24 for the authority’s.
Fourth, unreasonableness is context-dependent: a decision is more or
less likely to be unreasonable depending on a range of contextual factors.
Two factors are especially important. One is an authority’s expertise on
the matter decided relative to the court’s. The greater the authority’s
relative expertise, the less likely its decision is unreasonable, other things
being equal.25 For instance, the British Union for the Abolition of
Vivisection claimed that an investigator’s findings as to the suffering
caused by certain animal experiments were unreasonable.26 In rejecting
that claim, the reviewing court said that it should be careful not to
‘substitute its own inexpert view of the science’27 for that of an ‘expert
and experienced’28 scientist.
The other main contextual factor is the impact of the decision. A
decision that impacts a fundamental right or interest is subject to
‘anxious scrutiny’29, meaning that a court is ‘entitled to start from the

22
R v Secretary of State for the Home Department, ex p Brind [1991] 2 WLR 588, [1991] 1
AC 696, 757 (Lord Ackner).
23
R (Khatun) v London Borough of Newham [2004] EWCA Civ 55, [40] (Laws LJ); R v
Ministry of Defence, ex p Walker [2000] 1 WLR 806 (HL), 812 (Lord Slynn).
24
R v Secretary of State for the Home Department, ex p Brind [1991] 2 WLR 588, [1991] 1
AC 696 (HL), 757 (Lord Ackner).
25
R v Ministery of Defence, ex p Smith [1996] QB 517 (CA), 556. See also International
Transport Roth GmbH v Secretary of State for the Home Department [2002] EWCA Civ
158 (Laws LJ); R (CENTRO) v Secretary of State for Transport [2007] EWHC 2729
(Admin), [36] (Beatson J).
26
R (British Union for the Abolition of Vivisection) v Secretary of State for the Home
Department [2008] EWCA Civ 417.
27
R (British Union for the Abolition of Vivisection) v Secretary of State for the Home
Department [2008] EWCA Civ 417, [1] (May LJ)
28
R (British Union for the Abolition of Vivisection) v Secretary of State for the Home
Department [2008] EWCA Civ 417, [1] (May LJ)
29
Bugdaycay v Secretary of State for the Home Department [1987] AC 514 (HL), 531
(Lord Bridge). See also R (Yogathas) v Secretary of State for the Home Department

Electronic copy available at: https://ssrn.com/abstract=4207429

premise’ that the decision ‘requires to be justified’30 and that only an
important justification will do. Thus, it was unreasonable for the Bloody
Sunday inquiry to publish the full names of soldiers who had fired live
rounds during the massacre given that (a) the decision risked ‘the most
fundamental of all human rights’31 – the lives of ‘the former soldiers and
their families’32; and (b) there was no ‘compelling justification’33 for
publishing the names.
Fifth, unreasonableness is evidence-dependent. The court’s task is to
test the authority’s ‘conclusion against the evidence before [the
authority]’ and ‘ask whether the conclusion can … be safely justified on
the basis of that evidence’34. One implication is that a decision is
unreasonable if there is clearly no support for that decision in the
evidence before the authority.35 Another implication is that a decision is
not unreasonable if evidence that would have counted decisively against
it becomes available only after the decision has been made.36 In one
recent case, it was reasonable for a parole board panel to direct the
release of a prisoner who had been smuggling large amounts of steroids
and alcohol into a prison, given that the panel did not learn of the
smuggling until after it made its decision.37

[2002] UKHL 36, [9] (Lord Bingham), [58]-[59] (Lord Hope); R (YH) v Secretary of
State for the Home Department [2010] EWCA Civ 116, [22]-[24] (Lord Carnwath).
30
R v Secretary of State for the Home Department, ex p Brind [1991] 1 AC 696, 748-749
(Lord Bridge). See also R v Secretary of State ex p Moon [1997] 1 INLR 165 (QB), 170
(Sedley J).
31
R v Lord Saville of Newdigate, ex p A [2000] 1 WLR 1855 (CA), 1877 (Lord Woolf
MR).
32
R v Lord Saville of Newdigate, ex p A [2000] 1 WLR 1855 (CA), 1877 (Lord Woolf
MR).
33
R v Lord Saville of Newdigate, ex p A [2000] 1 WLR 1855 (CA), 1877 (Lord Woolf
MR).
34
R (Wells) v Parole Board [2019] EWHC 2710 (Admin), [32].
35
Champion v Chief Constable of Gwent [1990] 1 WLR 1, 6-7 (HL); R v Birmingham City
Council ex p Sheptonhurst Ltd [1990] 1 All ER 1026, 1038 (CA); R v Secretary of State
for the Home Department, ex p Zakrocki (1998) 1 CCLR 374 (QB).
36
R (Dickins) v Parole Board for England and Wales [2021] EWHC 1166 (Admin).
37
R (Dickins) v Parole Board for England and Wales [2021] EWHC 1166 (Admin). On
the relevance of late evidence, see A Beetham, ‘The Parole Board as “Functus
Officio”’ (2021) 85 Journal of Criminal Law 406, 407).

Electronic copy available at: https://ssrn.com/abstract=4207429

 The sixth truism concerns moral legitimacy or justification. Judges
generally believe that unreasonableness review is morally justified.
Coke’s words still inspire:
… notwithstanding the words of the commission give authority to the
commissioners to do according to their discretions, yet their
proceedings ought to be limited and bound with the rule of reason
and law. For discretion is a science or understanding to discern
between falsity and truth, between wrong and right, between
shadows and substance, between equity and colourable glosses and
pretences, and not to do according to their wills and private
affections ….38
It could be that judges are mistaken, and they should not engage in
unreasonableness review. Perhaps they should engage in a more or less
intrusive form of review instead. Judges are, however, typically
intelligent people, knowledgeable in the law, acting in good faith. If they
say that unreasonableness review is justified, then it is at least plausible
that it is justified. So, an analysis of unreasonableness does not need to
explain why it is justified, but it does need to explain why it ‘might
plausibly be thought to be’39 so.
Finally, there are indicia of unreasonableness, that is, features of
decisions reliably associated with unreasonableness.40 There are at least
four indicia:
1.—Oppressiveness. The fact that a decision imposes costs on an
individual disproportionate to its benefits suggests unreasonableness.41
For example, Ms F had been raped by three soldiers while living in
Kosovo. She suffered lasting psychiatric injury as a result. When F
applied for asylum in the United Kingdom, she was refused. F appealed.

38
Rooke’s Case, (1598) 5 Co Rep 99b, 77 ER 209 (KB), 210. For many other examples,
see HWR Wade and CF Forsyth, Administrative Law (11th edn, OUP 2014), 293-5.
39
S Smith, Contract Theory (OUP 2004) 20. This is Smith’s ‘moderate’ version of his
morality criterion for interpretive theories or rational reconstructions.
40
For similar lists, see P Daly, ‘Wednesbury’s Reason and Structure’ [2011] PL 238, 242-
247. H Dindjer, ‘What Makes an Administrative Decision Unreasonable?’ (2021) 84
MLR 265, 293-4.
41
Kruse v Johnson [1898] 2 QB 91 (DC), 99-100 (Lord Russell CJ); R v Barnsley MBC, ex
p Hook [1976] 1 WLR 1052 (CA); R (Khatun) v London Borough of Newham [2004]
EWCA Civ 55, [41] (Laws LJ).

Electronic copy available at: https://ssrn.com/abstract=4207429

A psychiatrist would be able to submit a report on F’s injuries, but not
until a few days after the appeal hearing was scheduled. The special
adjudicator refused to adjourn the hearing. In R (F) v A Special
Adjudicator42, the refusal was quashed as unreasonable. On the one hand,
the psychiatrist’s report would have assisted F’s asylum application. On
the other, the delay to the hearing was minor. Overall, it was clear that
the costs of the refusal far exceeded its benefits.
2.—Incoherence. That an authority adopted an end but not the
necessary means to that end indicates unreasonableness.43 In R (Cawser) v
Secretary of State for the Home Department44, for example, the Secretary of
State conceded ‘that it would be irrational to have a policy of making
release [from prison] dependent upon the prisoner undergoing a [sex
offender rehabilitation] treatment course without making ... provision
for such courses’45.
3.—Inconsistency. That a present decision is inconsistent with a prior
decision also suggests that the present decision is unreasonable.46 For
instance, in 2005 the Home Office adopted a scheme for highly skilled
migrants under which they would be eligible to obtain indefinite leave
to remain (ILR) after four years’ residence. The Home Secretary later
changed the rules of the scheme, making it more difficult to obtain ILR.
Some migrants left the United Kingdom as a result. Others who
happened to be abroad when the changes were announced chose not to
return. The rule changes were then declared unlawful. The question thus
arose: should a migrant’s absences due to the unlawful rule changes be

42
R (F) v Special Adjudicator [2002] EWHC 777 (Admin).
43
R (ABCIFER) v Defence Secretary [2003] EWCA Civ 473, [40] (Dyson LJ); R (Law
Society) v Legal Services Commission [2010] EWHC 2550 (Admin). For a helpful
discussion, see R Williams, ‘Structuring Substantive Review’ [2017] Public Law 99,
107-8.
44
R (Cawser) v Secretary of State for the Home Department [2003] EWCA Civ 1522.
45
R (Cawser) v Secretary of State for the Home Department [2003] EWCA Civ 1522, [30]
(Simon Brown LJ). The concession was accepted as appropriate in R (James) v
Secretary of State for the Home Department [2009] UKHL 22, [44] (Lord Carswell).
46
R (Middlebrook Mushroom Ltd.) v Agricultural Wages Board [2004] EWHC 1447
(Admin), [74] (Stanley Burton J); R (Gallaher Group Ltd.) v The Competition and
Markets Authority [2018] UKSC 25, [26] (Lord Carnwath). See also J Randhawa and
M Smyth, ‘Equal Treatment and Consistency Before and After Gallaher’ (2018) 23
Judicial Review 159.

Electronic copy available at: https://ssrn.com/abstract=4207429

counted against them when they applied for ILR? In Patel v Secretary of
State for the Home Department47, it was held unreasonable for the Home
Secretary to count the absences of those who had stayed abroad against
them. An important reason was that (i) the Home Secretary had earlier
decided not to count the absences of those who left the UK against them
and (ii) the two groups were very similar.
4.—Lack of stated reasons. Lord Pearce said in Padfield v Minister of
Agriculture, Fisheries, and Food48 that if ‘all of the prima facie reasons seem
to point in favour’49 of a decision, and if the authority ‘gives no reason
whatever for taking a contrary course, the court may infer that he has no
good reason’50. The use of the power is unreasonable, as a result.
I do not claim that a decision is reasonable unless it displays one of
these markers. The decision in Duffy does not exhibit any of them but
was still unreasonable, for example. All I claim is that a good analysis
will explain why, despite their diversity, all of these factors indicate the
same thing, namely, unreasonableness.
None of these truisms are optional. None are less vital than any of
the others. What we need is an analysis that tells us what
unreasonableness is such that all of these things are true of it.

III. DEMONSTRABLE WRONGNESS

In the course of writing this article, I have embraced and abandoned a

number of analyses of unreasonableness. Among my many mistakes, the
most time-consuming was to assume that a confusing ground of review
needs a complicated analysis. It does not, of course. The analysis I want
to persuade you of is simple. It says: a decision is unreasonable if and only if
it is demonstrably wrong. I explain the analysis in this section. I argue for
it in the next two sections.

47
Patel v Secretary of State for Home Department [2012] EWHC 2100 (Admin)
48
Padfield v Minister of Agriculture, Fisheries, and Food [1968] 1 AC 997 (HL).
49
Padfield v Minister of Agriculture, Fisheries, and Food [1968] 1 AC 997 (HL), 1053
(Lord Pearce).
50
Padfield v Minister of Agriculture, Fisheries, and Food [1968] 1 AC 997 (HL), 1053-4
(Lord Pearce). See also R v Secretary of State for Trade and Industry ex p Lonhro Plc
[1989] 1 WLR 525 (HL), 539-540 (Lord Keith).

10

Electronic copy available at: https://ssrn.com/abstract=4207429

 Let me start with what makes a decision “wrong”. There are reasons
for and against decisions. These reasons compete by strength or weight,
with the weightier reasons prevailing over the less weighty ones. If the
reasons against a decision are collectively weightier than the reasons for
it, then there is most reason or decisive reason against that decision.51 We
can also say that the decision is contrary to the balance of reasons. If a
decision is contrary to the balance of reasons, one ought not to take that
decision. Not taking it is right; taking it is wrong.
These claims are commonplace. Things get more controversial when
it comes to the relationship between reasons and evidence. According to
some philosophers, one can have a reason that is not part of one’s
evidence. If late last night someone snuck into your garage and secretly
disabled the brakes on your car, then you have a reason not to drive your
car this morning. It does not matter that the fact the brakes were
disabled is not part of your evidence. Other philosophers think that a
fact can be a reason for one only if it is part of one’s evidence. Since the
fact that your brakes were disabled was not part of your evidence when
you decided to drive, that fact was no reason for you at that time. Since
there were – we can assume – no other reasons not to drive, it was not
wrong to drive. Following Benjamin Kiesewetter, let us call these views
objectivism and perspectivism, respectively.52
I prefer perspectivism, for two reasons. One is that it yields more
intuitive consequences in a range of cases. Consider an example of Frank
Jackson’s:
Jill is a physician who has to decide on the correct treatment for her
patient, John, who has a minor but not trivial skin complaint. She
has three drugs to choose from: drug A, drug B, and drug C. Careful
consideration of the literature has led her to the following opinions.
Drug A is very likely to relieve the condition but will not completely
cure it. One of drugs B and C will completely cure the skin
condition; the other though will kill the patient, and there is no way

51
D Parfit, On What Matters (OUP 2011) 32.
52
B Kiesewetter, The Normativity of Rationality (OUP 2017) 195. The formulations of
these views are also based on Kiesewetter.

11

Electronic copy available at: https://ssrn.com/abstract=4207429

 that she can tell which of the two is the perfect cure and which the
killer drug. What should Jill do?53
Obviously the answer is: Jill ought to prescribe drug A. That is the
answer perspectivism gives, too. Let us say that a reason is available to
one at some time just if it is part of one’s evidence at that time.54 The
reasons that are available to Jill are (i) that drug A is very likely to
relieve the condition, (i) that drug B has a 50-50 chance of killing or
curing the patient, and (iii) that drug C has a 50-50 chance of killing or
curing the patient 55 The balance of the available reasons clearly favours
giving John drug A.
By contrast, the objectivist must say that the fact that drug B (or
drug C, as the case may be) will cure John is a reason for Jill to prescribe
drug B, even though it is not part of Jill’s evidence. That reason will be
decisive. So, according to the objectivist, Jill ought to prescribe drug B
and not drug A. But that is very counterintuitive: Jill ought not to give
John a drug that, as far as her evidence indicates, has a 50% chance of
killing him just to treat a minor skin condition. So, in this case and
others like it, perspectivism is more intuitive than objectivism.
The second argument for perspectivism is that reasons and oughts
are supposed to help guide our deliberation and decision-making, but
we cannot be guided by what is not is not accessible to us. One of the
motivations for the principle that “ought” implies “can” is that practical
requirements must not be, as Jonathan Dancy puts it, mere ‘shouting in
the dark’56. Likewise, the facts that constitute our reasons ‘must lie
within our capacities for recognition, if they are to be capable of being
practically relevant for us’57. In Jill’s case, the fact of the matter as to
whether drug B or drug C will cure the patient cannot guide Jill’s
deliberation or decision-making, since it is not accessible to her. It

53
F Jackson, ‘Decision-Theoretic Consequentialism and the Nearest and Dearest
Objection’ (1991) 101 Ethics 461, 462-3.
54
B Kiesewetter, The Normativity of Rationality (OUP 2017) 199.
55
B Kiesewetter, The Normativity of Rationality (OUP 2017) 202.
56
J Dancy, Practical Reality (OUP 2002) 59. Dancy is here discussing moral
requirements and moral reasons. Similar points are made by J Oberdiek, Imposing
Risk (OUP 2017) 50-52.
57
J Dancy, Practical Reality (OUP 2002) 59.

12

Electronic copy available at: https://ssrn.com/abstract=4207429

cannot play the role that reasons are supposed to play, which suggests it
is not a reason at all.
I do not claim that these arguments show that perspectivism is
correct. There are counters to these arguments, counters to those
counters, and so on.58 All I claim is that these arguments show that
perspectivism is a plausible way to think about reasons, oughts, and
wrongness. Given that this article is not a contribution to the
philosophy of normativity, I hope that entitles me to treat perspectivism
as true in what follows. Assuming perspectivism is correct, when I say
that an unreasonable decision is demonstrably “wrong”, I mean that it is
demonstrably contrary to the balance of reasons available to the
authority when it made its decision.
I turn now to what makes a decision “demonstrably” wrong. The
wrongness of a decision is demonstrable, in the relevant sense, when a
reviewing court would be entitled to believe (conclude, consider, etc.)
that it is wrong, based on all the evidence before it. Although the idea is
simple enough, I need to explain what “all of the evidence” includes.
Let us say that an authority is a court’s epistemic superior on an issue
just if the authority’s conclusion on that issue is more reliable – more
likely true – than the court’s.59 If the court has reason to believe that an
authority is its epistemic superior on an issue, then it should treat the
authority’s conclusion on that issue as evidence of what it concluded. It
should attach weight to the authority’s conclusion. It should, in other
words, defer to the authority.60 The stronger the reason to believe that
the authority is the court’s superior, the more weight the court should

58
See P Graham, Subjective versus Objective Moral Wrongness (CUP 2021) for an
overview.
59
A Elga, ‘Reflection and Disagreement’ (2007) 41 Nous 478, 484.
60
For weight-based understandings of deference, see, e.g., Huang v Secretary of State
for the Home Department [2007] UKHL 11, [16] (Lord Bingham) (the court should
give ‘appropriate weight to the judgment of a person with responsibility for a
given subject matter and access to special sources of knowledge and advice’); A
Kavanagh, ‘Deference or Defiance? The Limits of the Judicial Role in
Constitutional Adjudication’ in G Huscroft (ed), Expounding the Constitution (CUP
2008).

13

Electronic copy available at: https://ssrn.com/abstract=4207429

attach to the authority’s conclusion, and the more it should defer to the
authority.61
Authorities often have greater ‘knowledge and experience’62 of policy
issues than courts. In other words, courts are often relatively ‘ill-
equipped to take decisions in place of the … authority’.63 It is plausible
that the Secretary of State for Northern Ireland, for example, knows
more about Northern Irish political dynamics than the judges of the
House of Lords. So, there is typically at least some reason for a court to
believe that an authority is its epistemic superior.64 If an authority took
a decision, it will have thought that decision was right. So, a court
should typically treat an authority’s decision on a policy issue as
evidence of at least some weight that its decision is right.
There will be other evidence available to the court, in addition to the
authority’s decision (e.g., testimony, affidavits, documents). Some of that
other evidence may give the court reason to believe that the authority’s
decision was wrong. In that case the court will need to take into account
the total evidence – the authority’s view plus the other evidence – to
work out what to believe. Perhaps it should overall agree with the
authority; perhaps not. It all depends on how much deference the
authority is owed and the strength of the other evidence.65 When I say
that an unreasonable decision is “demonstrably” wrong I mean that a
court ought to consider it wrong based on all of its evidence – including
the fact that the authority believed that the decision was right.
Putting these points together, the idea is that unreasonableness is a
function of two bodies of evidence: (i) the evidence that could be

61
SE Bokros, ‘A Deference Model of Epistemic Authority’ (2021) 198 Synthese 12041,
12064-5.
62
Associated Provincial Picture Houses Ltd v Wednesbury Corporation [1948] 1 KB 223, 230
(Lord Greene MR). See also R v Lord Saville of Newdigate, ex p A [2000] 1 WLR
1855, 1866 (Lord Woolfe MR);
63
Lord Irvine, ‘Judges and Decision Makers: The Theory and Practive of Wednesbury
Review’ [1996] PL 59, 61.
64
I do not claim that this reason is typically decisive, or that the overall balance of
reasons typically favours the court believing that the authority is its epistemic
superior.
65
A Kavanagh, ‘Defending Deference in Public Law and Constitutional Theory’
(2010) 126 LQR 222, 225; A Young, ‘Deference, Dialogue and the Search for
Legitimacy’ (2010) 30 OJLS 815, 818.

14

Electronic copy available at: https://ssrn.com/abstract=4207429

considered by a reviewing court, and (ii) the evidence before the
authority when it made its decision. A decision is unreasonable if and
only if a court would be entitled, relative to the former body of
evidence, to find the decision wrong, relative to the latter body of
evidence. More concisely, an unreasonable decision is one that from a
court’s perspective is wrong from the authority’s perspective.
To illustrate, in Duffy the Secretary of State believed that Burrows
and Mackay could be impartial and be perceived as impartial. He might
have been correct: it is possible that Burrows and Mackay would have
been exemplary members of the Parades Commission. But the evidence
before the Secretary of State was very much to the contrary. Both men
had been ‘very prominent and committed proponents of the loyalist
parade from Drumcree along the Garvaghy Road’66. Neither had
resigned from the loyalist bodies to which they belong or given ‘any
recorded indication that he had changed his allegiance’67. Mackay even
planned to march in the parade from Drumcree along Garvaghy Road.68
The case against the appointments was so overwhelming that the House
of Lords was entitled to consider the appointments wrong,
notwithstanding the deference due to the Secretary of State. That is the
same as saying that the appointments were demonstrably wrong.

IV. UNREASONABLENESS IS DEMONSTRABLE WRONGNESS

My analysis is simple, and so is the reason to accept it. The reason is that
the analysis accounts for the self-evident truths about unreasonableness
described in section II.
First, why is unreasonableness concerned with the weight and
balance of the relevant reasons? Because – says my analysis –
unreasonableness is concerned with whether a decision is wrong, and
wrongness is understood in terms of the balance of reasons. Other
grounds of review (e.g., proper purposes, relevancy) are concerned with
the reasons for which the authority acted and the reasons it included in
its deliberations. Unreasonableness, by contrast, is not about motivation

66
In re Duffy [2008] UKHL 4, [27] (Lord Bingham).
67
In re Duffy [2008] UKHL 4, [27] (Lord Bingham).
68
In re Duffy [2008] UKHL 4, [44] (Lord Carswell).

15

Electronic copy available at: https://ssrn.com/abstract=4207429

or deliberation. It is only about whether a decision can be considered
contrary to the balance of reasons.
Second, why is unreasonableness a high standard? That is, why does
wrongness not suffice for unreasonableness? Because unreasonableness is
a matter of demonstrable wrongness and not every wrong decision is
demonstrably wrong. Demonstrable wrongness is a function of the
evidence before the court. Wrongness is a function of the evidence
before the authority. The two sets of evidence are different; only the
former includes the authority’s judgment that its decision was right.
That additional piece of evidence can make a difference to the
conclusion the court should reach. When it does, a decision is wrong but
reasonable. Suppose that the evidence before the Secretary of State gave
him a reason to appoint Burrows with a weight of 7. The evidence also
gave him a reason not to appoint Burrows with a weight of 10. If his view
that it was right to appoint them carries a weight of 4, then the
appointment is wrong (7 < 10), but the court should not consider it
wrong (11 > 10). The appointment would be wrong but reasonable.
For similar reasons, unreasonableness review is not a form of merits
review. In merits review, a court relies solely on its own judgment of the
merits. It does not attach weight to anyone else’s judgment as to whether
a decision is wrong (e.g., the Secretary of State’s). By contrast, a decision
is unreasonable only if a court should consider it wrong based on all the
evidence, including the fact of the authority’s judgment. Whereas merits
review is non-deferential, unreasonableness review is inherently
deferential.
Fourth, why is unreasonableness context-dependent? Some
contextual factors indicate how much weight the court should give to
certain evidence. For example, the greater an authority’s expertise, the
greater the weight a court should attach to the authority’s view. At the
extreme, so much weight is given to the authority’s view that it is hard
to imagine what contrary evidence could entitle the court to adopt a
different view. In the vivisection case, for example, an investigator’s
scientific expertise was so great that a challenge to his findings was at
the ‘further boundary of that which is suitable for judicial review’69.

69
R (British Union for the Abolition of Vivisection) v Secretary of State for the Home
Department [2008] EWCA Civ 417, [87] (May LJ).

16

Electronic copy available at: https://ssrn.com/abstract=4207429

Other contextual factors are evidence that an authority’s decision was
wrong. If a claimant can show that a decision (e.g., to publish the names
of the soldiers who fired live rounds at the Bloody Sunday massacre)
impacted on a fundamental right or interest (e.g., a right to life), they
are halfway home. They have already shown to the court that there is a
weighty reason against the decision. The decision is unreasonable unless
the court should find an even weightier reason – a ‘compelling
justification’70 – for the decision.
Although my analysis favours expertise-based deference, it does not
favour deference on democratic grounds. Deference, as I think of it, is
about treating others’ beliefs as evidence of what they believe when one
forms one’s own view on the matter. Deference, in this sense, is justified
by the reliability of others’ judgments. That a body is democratically
constituted does not, in itself, warrant deference to its views because it
does not, in itself, indicate anything about its reliability. Of course
democratic credentials may indicate reliability. A large and
representative body may be well suited to arriving at the truth of some
highly polycentric matters, for example. If so, its democratic character
favours deference – but indirectly, through its relevance for reliability.
Admittedly, there are cases like Nottinghamshire County Council v
Secretary of State for the Environment71, in which the House of Lords was
reluctant to find that a decision was unreasonable, partly because the
decision was subject to a House of Commons affirmative resolution
procedure. Such cases are sometimes thought to favour deference on
democratic grounds. But the better reading, accepted by later courts, is
that deference was warranted in these cases because the matters decided
in them were political or economic, and therefore lay within the
expertise of ministers and Parliament.72

70
R v Lord Saville of Newdigate, ex p A [2000] 1 WLR 1855 (CA), 1877 (Lord Woolf
MR).
71
Nottinghamshire County Council v Secretary of State for the Environment [1986] 1 AC
240 (HL); see also R v Secretary of State for the Environment, ex p Hammersmith &
Fulham LBC [1991] 1 AC 521 (HL).
72
This reading is accepted in R (Javed) v Secretary of State for the Home Department
[2001] EWCA Civ 789, [49] (Lord Phillips MR); R (CENTRO) v Secretary of State
for Transport [2007] EWHC 2729 (Admin), [36] (Beatson J). But cf R (Rotherham

17

Electronic copy available at: https://ssrn.com/abstract=4207429

 Fifth, why does unreasonableness depend on the evidence available to
an authority? Because a decision is unreasonable only if a court ought to
consider it wrong, and a decision is wrong only if the authority’s
evidence gave it most reason not to take the decision. Earlier I gave an
example of a parole board panel releasing a prisoner who had been
smuggling contraband into the prison.73 The decision was reasonable
because the panel had not been informed of the smuggling. That makes
sense, on my analysis, because the prisoner’s smuggling was not part of
the panel’s evidence. It was therefore not a reason for the panel to refuse
his release. Since the reasons that were part of the panel’s evidence
overall favoured the prisoner’s release, the panel’s decision was not
wrong, and was therefore not unreasonable.
Conversely, a fact does not support an authority’s conclusion unless
it was part of its evidence. In Hollings v Bath & North East Somerset
Council74, for example, a care home sought planning permission to extend
its premises to include 52 more beds. The council granted permission,
despite local opposition and the premises’ Grade I listed status. The
council did so because it believed that only larger care homes were
financially viable. The court, ‘having considered all of the information
available to the councillors’, found it ‘impossible to identify the
provenance or foundation’75 of this belief. There was simply ‘no
evidence’76 in support of it. This consideration had to be ‘stripped
away’77. It ‘could not legitimately be placed in what was overall clearly a
relatively fine balance’78. And it was ‘very difficult indeed to see how the
balancing exercise leading to the decision does not inevitably fall as a

Metropolitan Borough Council) v Secretary of State for Business, Innovation and Skills
[2015] UKSC 6, [23] (Lord Sumption).
73
R (Dickins) v Parole Board for England and Wales [2021] EWHC 1166 (Admin).
74
R (Hollings) v Bath and North East Somerset Council [2018] 5 WLUK 375 (QB).
75
R (Hollings) v Bath and North East Somerset Council [2018] 5 WLUK 375, [40] (Cotter
J).
76
R (Hollings) v Bath and North East Somerset Council [2018] 5 WLUK 375, [44] (Cotter
J).
77
R (Hollings) v Bath and North East Somerset Council [2018] 5 WLUK 375, [79] (Cotter
J).
78
R (Hollings) v Bath and North East Somerset Council [2018] 5 WLUK 375, [41] (Cotter
J).

18

Electronic copy available at: https://ssrn.com/abstract=4207429

result’79. In my terminology, a consideration cannot ‘legitimately be
placed’ in the balance of reasons unless it is part of the authority’s
evidence, that is, unless it is an available reason. Since all the reasons
that were available to the council were reasons not to grant permission
(e.g., listed status), the decision was contrary to the balance of available
reasons.
Sixth, what makes it plausible that unreasonableness review is
justified? We all want authorities to decide rightly. We do not want
them to decide wrongly. Everyone will agree on that. Of course,
authorities do sometimes decide wrongly. But they will do so less often if
courts review decisions for unreasonableness, understood as
demonstrable wrongness. Since courts have reason to do what will
reduce the likelihood of wrong decisions, they have reason to review
decisions for unreasonableness. That is the rough argument. Here it is
more precisely:
P1. If there is a reason for one to X, and one’s Ying brings it about
that one Xs, then one has a reason to Y.80
P2. There is a reason for courts to reduce the likelihood of wrong
decisions by authorities.
P3. Courts reviewing authorities’ decisions for unreasonableness
brings it about that they reduce the likelihood of wrong decisions by
authorities.
C. So, there is a reason for courts to review authorities’ decisions for
unreasonableness.
I take it that P1, or a like rule of instrumental reason, is uncontroversial.
I also assume that P2 is uncontroversial. It is good, surely, if courts can
make it less likely that authorities will decide wrongly. I do, however,
want to say a bit more about the final premise, P3.
Assume that most of the decisions that courts find unreasonable are
in fact unreasonable. Assume, too, that if a court finds an authority’s
decision unreasonable, and it is unreasonable, the authority will make a

79
R (Hollings) v Bath and North East Somerset Council [2018] 5 WLUK 375, [79] (Cotter
J).
80
This is a simplified version of an account of instrumental reasons in N Kolodny,
‘Instrumental Reasons’ in D Star, The Oxford Handbook of Reasons and Normativity
(OUP 2018).

19

Electronic copy available at: https://ssrn.com/abstract=4207429

reasonable decision instead.81 Thus, the House of Lords found Burrows
and Mackays’ appointments unreasonable. They were in fact
unreasonable. The court’s intervention led the Secretary of State to
make suitable appointments to the Parades Commission. Now, an
unreasonable decision is, we are supposing, one that a court ought to
believe is wrong. A decision that a court ought to believe is wrong likely
is wrong, relative to the court’s evidence. So, given these two
assumptions, findings of unreasonableness mostly lead to authorities to
make likely right decisions instead of likely wrong decisions. That
establishes P3. Are these two assumptions justified? I believe so, at least
under ordinary conditions. But I do not need to defend that strong
claim here. These assumptions are plausible, and that is enough to make
it plausible that unreasonableness review is justified.
Finally, why is unreasonableness indicated by oppressiveness,
inconsistency, incoherence, and a lack of stated reasons? Let me take
these indica in order.
1.—Oppressiveness. Recall that a decision is oppressive if the benefits
of a decision are greatly exceeded by the costs it imposes on an
individual. These benefits and costs are reasons for and against the
decision. Since the costs are excessive, the decision is contrary to the
balance of reasons. If all this is demonstrable to the court, then the
decision is unreasonable. In F, for instance, it was plain to the court that
by refusing to give F extra time to submit a psychiatrist’s report, the
adjudicator had effectively doomed F’s asylum application for the sake
of avoiding minor administrative inconvenience. Unsurprisingly, the
decision was unreasonable.
2.—Incoherence. Incoherence – a lack of appropriate connection
between an authority’s chosen means and ends – is the next indicium of
unreasonableness. A good illustration is R (Johnson) v Work and Pensions
Secretary82. The Universal Credit Regulations 2013 awarded benefits
based partly on income earned in a monthly assessment period. Some
claimants were paid by their job on the last day of the month. When
their pay day fell on a bank holiday or weekend, the pay date would be

81
It is enough that they make a less unreasonable – less likely wrong – decision
instead. That would also result in a reduction of the likelihood of error.
82
R (Johnson) v Work and Pensions Secretary [2020] EWCA Civ 778.

20

Electronic copy available at: https://ssrn.com/abstract=4207429

pushed back a day, and they would receive two pay cheques in one
month. Their universal credit award would be reduced. In the next
month, they would receive no pay cheque. Their universal credit award
would be increased, but not by as much as it had been reduced the
previous month. So, claimants who were paid on the last day of each
month were left worse off than claimants who were not, other things
being equal. This problem was called the ‘non-banking day salary shift’83.
The Court of Appeal held that it was unreasonable for the Secretary
of State to fail to correct the non-banking day salary shift. The
‘legislative policy behind universal credit in general’ and these
regulations in particular ‘is to encourage work by being responsive to
changes in earned income’84. The non-banking day salary shift made it
harder for claimants to predict how much income they would receive in
any given month. That made budgeting difficult and increased debt.
Many claimants would conclude ‘that it is preferable in many ways not
to work’85. Those claimants who stayed in employment would be
encouraged to choose jobs based on salary pay dates, rather than on the
potential to make ‘use of … [their] skills and educational achievements’
and ‘previous work experience’86. Instead of promoting the aim of the
universal credit regime, the non-banking day salary shift frustrated it.
For this and other reasons, it was unlawful for the Secretary of State to
fail to address the problem.
Here is how my analysis makes sense of cases like Johnson. Grant that
the Secretary of State ought to ‘encourage work’ by introducing a
benefits scheme that does not disadvantage recipients for working and
using their skills. Grant that on the evidence before the Secretary of
State addressing the non-banking day salary shift was a necessary means
of achieving that end. Finally, grant the general principle that if one
ought to X, and one ought to believe that one will X only if one Ys, then
one ought to Y.87 It follows that the Secretary of State ought to have
addressed the non-banking day salary shift and that failing to do so was

83
R (Johnson) v Work and Pensions Secretary [2020] EWCA Civ 778, [2] (Rose LJ).
84
R (Johnson) v Work and Pensions Secretary [2020] EWCA Civ 778, 100]
85
R (Johnson) v Work and Pensions Secretary [2020] EWCA Civ 778, [106].
86
R (Johnson) v Work and Pensions Secretary [2020] EWCA Civ 778, [106].
87
N Kolodny ‘Why be Disposed to be Coherent? (2008) 118 Ethics 437, 452; B
Kiesewetter, The Normativity of Rationality (OUP 2017) 265.

21

Electronic copy available at: https://ssrn.com/abstract=4207429

wrong. All of this was plainly in evidence before the court, so the failure
was unreasonable.88
3.—Inconsistency. Inconsistency with a prior decision suggests that a
present decision is unreasonable; but why? On the one hand, a court
should treat an authority’s present decision as evidence that it is right.
On the other hand, a court should treat an authority’s prior and
inconsistent decision as evidence that it was right and thus as evidence
that the present decision is wrong.89 And evidence that the present
decision is wrong tends to establish its unreasonableness.
In Patel, for example, there were reasons in favour of counting
absences from the UK due to unlawful rule changes against migrants
applying for ILR. There were also reasons against doing so. The Home
Secretary’s choice to place more weight on the former reasons in Patel’s
case deserved deference. However, the Home Secretary’s choice in other
cases to place more weight on the latter reasons also deserved deference.
The deference paid to one decision is offset, to at least some degree, by
the deference paid to another decision. The decision is more likely
unreasonable than if deference were due only to the present decision.
My claim is not that inconsistency is conclusive of wrongness. It is
merely only one piece of evidence among others. Suppose that the total
evidence (including the Home Secretary’s judgments) gives the court
most reason to believe that Patel’s absences should have been counted
against him. The decision in Patel would be reasonable, despite its
inconsistency with the previous decisions. This accords with how courts
think about inconsistency. They, too, stress that inconsistency with a
prior decision does not, on its own, make a decision unlawful.90
4.—Lack of stated reasons. If all the evidence before a court is of
reasons against a decision, and the authority has not stated any reasons
for the decision, then says that the decision is unreasonable. That makes

88
If the Secretary of State ought not to pursue the end he did, then his decision is
likely unlawful under the improper purposes doctrine. If it is neither the case that
the Secretary of State ought or ought not to pursue this end, then matters are
much more complicated. For philosophical discussion, see, e.g., B Kiesewetter, The
Normativity of Rationality (OUP 2017), 267-294.
89
This assumes for simplicity that both decisions cannot be right.
90
O’Brien v Independent Assessor [2007] UKHL 10, [30] (Lord Bingham); R (Gallaher) v
The Competition and Markets Authority [2018] UKSC 25, [24] (Lord Carnwath).

22

Electronic copy available at: https://ssrn.com/abstract=4207429

sense, on my analysis, if we assume that there would be evidence before
the court of reasons for a decision were such reasons to exist.
Consider R v Penwith District Council, ex p May91. The Campaign for
Nuclear Disarmament had for many years sold literature, badges, and
other material along a street in Penzance. In 1984, the District Council
decided to require consent for trading activities on that street. The
CND duly applied for consent. On the evidence before the court, every
reason counted in favour of granting its application: there were no
complaints about the CND’s activities, they were not disruptive, they
did not compete with those of other businesses, etc. Indeed, the
Council’s own Environment Committee had recommended granting the
application. And yet the Council had refused the application without
explanation.
Assuming that the court would have had evidence of reasons to
refuse the CND’s application were there such reasons, the court was
entitled to infer that there were no reasons for the decision. Since there
were many reasons against the decision, the court was entitled to
conclude that the decision was wrong. The decision was demonstrably
wrong, in other words. According to my analysis, that means it was
unreasonable – which is the result the court reached, based on the dicta
from Padfield.
We said that an analysis of unreasonableness should account for
everything truistic of it. My analysis explains why unreasonableness is
concerned with the balance of reasons and while it is a difficult standard
to meet. It accounts for the difference between unreasonableness review
and merits review. It explains why unreasonableness depends on
evidence and context, why judges might believe that unreasonableness
review is justified, and why seemingly heterogenous factors all indicate
unreasonableness. In short, my analysis does all we said a good analysis
would do.

91
R v Penwith District Council, ex p May, unreported, 22 November 1985.

23

Electronic copy available at: https://ssrn.com/abstract=4207429

 V. OTHER ANALYSES

In this section I compare my analysis with the three main existing

analyses of unreasonableness. All have their virtues; none, however, is
adequate.

1.—I start with John Gardner’s analysis because it is simplest. According

to Gardner, the ‘word “reasonable”, in legal contexts’ means ‘no more
and no less than “justified”’92. That, he said, is what it means in tort law
and contract law; it is also what it means in administrative law.93 What
makes a decision ‘unjustified’? If a reason is weightier than a conflicting
reason, we say that the less weighty reason is defeated. If a reason is not
defeated by any competing reason, then we say the reason is undefeated.94
A decision is ‘taken for a reason’ when, roughly, one believes that there
is some reason for some decision, makes that decision, and makes that
decision because one believes there is that reason. Putting these points
together, a decision is justified just when it is taken for an undefeated
reason.95 A decision that is not justified is unjustified and a decision that
is not reasonable is unreasonable. So, the claim is that a decision is
unreasonable just if it is unjustified.
Gardner’s analysis suffers from at least three problems. First, if a
decision is wrong, then the reasons against it are weightier than the
reasons for it. Hence there is not an undefeated reason for it. Hence it is
unjustified. So, every wrong decision is unreasonable. That flatly
contradicts one of the things everyone knows about unreasonableness,
namely, that it is a high standard.96

92
J Gardner, ‘The Mysterious Case of the Reasonable Person’ (2001) 51 University of
Toronto Law Journal 273, 273.
93
J Gardner, Torts and Other Wrongs (OUP 2019) 276-7.
94
J Gardner, Torts and Other Wrongs (OUP 2019) 274. If there are exclusionary
reasons, then exclusion is an additional way in which a reason may be defeated.
See J Raz, Practical Reason and Norms (rev edn, Princeton 1991), ch 1.
95
J Gardner, ‘Justifications and Reasons’ in A Simester and ATH Smith (eds), Harm
and Culpability (OUP 1996); J Gardner, Torts and Other Wrongs (OUP 2019) 276-7.
96
This criticism is made effectively and in detail by H Dindjer, ‘What Makes an
Administrative Decision Unreasonable?’ (2021) 84 MLR 265, 275-6.

24

Electronic copy available at: https://ssrn.com/abstract=4207429

 Second, Gardner’s emphasis on lack of justification (as opposed to
wrongness) leads to unacceptable results. Suppose that a government
department could change the font it uses in its letters. It uses
Baskerville, say, and it could change to Times New Roman.97 There is no
reason to change the font. There is no reason not to. Since there is no
reason for either decision, no decision on this issue can be taken for a
reason. No decision can be taken for an undefeated reason. So, no
decision can be reasonable or lawful – says Gardner. Surely that is
mistaken. Authorities do not break the law every time they make an
inconsequential decision.
The third problem concerns evidence-dependence. Since it arises
more squarely with respect to Stephen Perry’s analysis, I raise it below.

2.—Hasan Dindjer recognises, as Gardner did not, that unreasonableness

is not simply lack of justification. He proposes a sophisticated
alternative. According to Dindjer, a decision is unreasonable if and only
if it is not justified on any ‘eligible view’98 of the balance of reasons. An
eligible view of the balance of reasons is determined by the ‘eligible
upper and lower bounds [of weight] for each individual reason’99
considered by the authority. A ‘weighting for a reason is eligible if the
court cannot, consistent with its epistemic and constitutional position,
treat it as mistaken’100.
How does this analysis work in practice? Suppose that in Duffy the
reason to appoint Burrows and Mackay is to gain unionist support for
the commission. The reason not to appoint them was that it might cause
conflict. Given the court’s ‘epistemic and constitutional position’ it
should defer to the Secretary of State as to the weights of each reasons –
up to a point. That point sets the upper and lower bounds of the
weightings for each reason. The decision is reasonable if and only if the

97
I owe this example to Hasan Dindjer (who uses it for a somewhat different
purpose).
98
H Dindjer, ‘What Makes an Administrative Decision Unreasonable?’ (2021) 84
MLR 265, 282.
99
H Dindjer, ‘What Makes an Administrative Decision Unreasonable?’ (2021) 84
MLR 265, 281.
100
H Dindjer, ‘What Makes an Administrative Decision Unreasonable?’ (2021) 84
MLR 265, 281.

25

Electronic copy available at: https://ssrn.com/abstract=4207429

weight of the reason for the decision, taken at its upper bound, is at least
as great as the weight of the reason against the decision, taken at its
lowest bound.
We can make this more concrete by attaching numbers to the
weights. Suppose that these are the weights of the reason for and against
the decision:

Table 1: Duffy

Lower bound Upper bound

Unionist support 1 2
Conflict avoidance 3 4

The weight of the reason for the decision, at its upper bound, is 2. The
weight of the reason against the decision, at its lower bound, is 3. Thus,
the decision is unreasonable.
Although there is a lot to like about Dindjer’s analysis, it suffers from
at least two problems. One problem is that it is not clear from Dindjer’s
discussion how to fix the upper and lower bounds of a reason’s
weighting. They are supposed to be determined by what a court can,
consistent with its epistemic and constitutional position, treat as
mistaken. But I do not know exactly how we are supposed to work out
what weightings a court is entitled to treat as mistaken. Although there
may be ways to fill this gap, the only way I can think of leads to the
second problem.
Suppose that a court, based on its independent assessment of the
merits, is somewhat confident that the weight of the reason to appoint
Burrows and Mackay is 1. It is highly confident, meanwhile, that its
weight is between 1 and 2. These beliefs are, let us suppose, rational. The
deference due to the Secretary of State ought to be enough to convince
the court to accept a view that it is somewhat confident is wrong, but
not highly confident is wrong. So, it should be enough to convince the
court that the reason’s weight is between 1 and 2 – no lower, no higher.
Anything outside this range the court is entitled to treat as mistaken.
Something similar is true of the other reason. The court is somewhat
certain that the reason has a weight of 3. It is highly confident it has a
weight between 3 and 4. So, it is entitled to treat a weight less than 3 or
greater than 4 as mistaken. In essence, a court is entitled to treat a

26

Electronic copy available at: https://ssrn.com/abstract=4207429

weighting of some reason as mistaken if, based on its own assessment, it
is entitled to be highly confident that it is mistaken. (To be clear, I am
not attributing this way of thinking to Dindjer. He does not use the
language of confidence, probability, etc.)
While this way of filling the gap may seem plausible, it creates
anomalous results. Suppose – purely for the sake of argument – that
these are the House of Lords’ confidence levels for the associated weights
of the reasons:

Table 2: Reason to appoint

Weights Confidence
1 0.4
2 0.4
3 0.05
4 0.05
5 0.1

Table 3: Reason not to appoint

Weights Confidence
0 0.05
1 0.1
2 0.05
3 0.7
4 0.1

Assume that “highly confident” is confidence equal to or greater than

0.8. The court should be highly confident that the reason to appoint’s
weight is 2 or less (0.4 + 0.4 = 0.8). It should be highly confident that the
reason not to appoint’s weight is 3 or more (0.7 + 0.1 = 0.8). It follows
that the balance of reasons does not favour the appointments unless at
least one reason is given a weighting that the court is entitled to treat as
mistaken. There is no eligible view of the balance of reasons that favours
the appointments. Dindjer’s analysis would – on my reading – say that
the appointments were unreasonable.
The trouble is that the court should have a confidence level of only
0.67 that the reason to appoint is outweighed by the reason not to

27

Electronic copy available at: https://ssrn.com/abstract=4207429

appoint.101 Since 0.67 < 0.8, the court should not be highly confident that
the decision was unjustified. We said a court should adopt the
authority’s view unless it is highly confident, based on its own
assessment, that it is mistaken. Thus, the court ought to accept that the
Secretary of State’s decision is justified. So, this is a decision that (i)
Dindjer’s analysis says is unreasonable but (ii) a court should believe is
justified. Now, no good analysis would tell courts to interfere in
decisions that they ought to regard as justified. For one thing, it would
be a form of review even more intrusive than merits review. On merits
review, courts impose the decisions they think are right. On Dindjer’s
analysis, courts should impose decisions they and the authority agree are
wrong. For another, it is not plausible that a court can legitimately set
aside decisions that it ought to regard as justified.
Why does Dindjer’s analysis produce this odd result? The problem is
his focus on the weighting of ‘each individual reason’. His analysis starts
by determining the permissible weightings of each reason. Based on
those weightings, we construct the eligible views of the balance of
reasons. But uncertainty compounds. If one is uncertain that event A
occurred, and uncertain that B occurred, one should be even less certain
that both A and B occurred. A court that is not certain that the reasons
for a decision are below a certain weight and not certain that the
reasons against that decision are above a certain weight should be even
less certain that both things are true. It should be even less certain as to
the balance of these reasons, in other words. My analysis does not suffer
from this problem because it is concerned with the overall balance of
reasons.

3.—Stephen Perry proposed his analysis of unreasonableness almost

thirty-five years ago.102 It has been entirely neglected by administrative
law scholars since. That is a shame – it is both the best existing analysis
of unreasonableness and nearly correct.

101
0.67 = 1-[2(0.4*0.1) + 3(0.4*0.05) + 4(0.05*0.05) + 5 (0.05*0.1) + 2(0.05*0.7) + (0.1*0.1)]
102
S Perry, ‘Second-Order Reasons, Uncertainty and Legal Theory’ (1989) 62 S Cal. L.
Rev. 913.

28

Electronic copy available at: https://ssrn.com/abstract=4207429

 Like Dindjer, and unlike Gardner, Perry knew that unreasonableness
is not mere wrongness or lack of justification. There was, he thought,
only one other serious possibility:
[W]hat else could it mean that a person regards a decision as wrong
but reasonable than that she believes the decision is mistaken but
recognizes that there is at least a nonnegligible possibility that she
might be wrong and the tribunal right?103
Thus, ‘an unreasonable decision … is one which she is convinced to some
relatively strong degree of certainty could not be right’104. A decision is
‘clearly’ mistaken or wrong from the court’s perspective just when the
court is convinced to a high degree or certainty that it is wrong.
For Perry, a decision is mistaken or wrong when it is at odds with the
‘objective balance of reasons’105 or ‘right reason’106. The objective balance
of reasons is the ‘balance of reasons understood as facts’107. It ‘consists of
all practical inferences and weighing processes that would be carried out
by an agent who, when deciding what ought to be done in a particular
situation, possessed true information about all relevant facts’108 and who
‘reasoned validly’109. So, a decision is Wednesbury unreasonable – for
Perry – just when it is clearly contrary to the objective balance of
reasons.
Taken literally, Perry’s analysis has some odd consequences. If a
decision has not yet been litigated, no court could be convinced – to any
degree – that it is contrary to the objective balance of reasons. Any
decision that has not yet been litigated would necessarily be Wednesbury

103
S Perry, ‘Second-Order Reasons, Uncertainty and Legal Theory’ (1989) 62 S Cal. L.
Rev. 913, 938-9.
104
S Perry, ‘Second-Order Reasons, Uncertainty and Legal Theory’ (1989) 62 S Cal. L.
Rev. 913, 939.
105
S Perry, ‘Second-Order Reasons, Uncertainty and Legal Theory’ (1989) 62 S Cal. L.
Rev. 913, 922, 939.
106
S Perry, ‘Second-Order Reasons, Uncertainty and Legal Theory’ (1989) 62 S Cal. L.
Rev. 913, 941.
107
S Perry, ‘Second-Order Reasons, Uncertainty and Legal Theory’ (1989) 62 S Cal. L.
Rev. 913, 922.
108
S Perry, ‘Second-Order Reasons, Uncertainty and Legal Theory’ (1989) 62 S Cal. L.
Rev. 913, 922.
109
S Perry, ‘Second-Order Reasons, Uncertainty and Legal Theory’ (1989) 62 S Cal. L.
Rev. 913, 922.

29

Electronic copy available at: https://ssrn.com/abstract=4207429

reasonable. That cannot be right. Moreover, a court might be convinced
to a very high degree of certainty that a decision is contrary to the
objective balance of reasons on entirely irrational grounds. Such a
decision would be unreasonable. Again, that cannot be right. Since these
mistakes are easy to see, charity suggests that we do not attribute them
to Perry. I shall take him to claim that a clearly wrong decision is one
that a court should be highly confident is wrong, were it to review the
decision. And I shall take him to claim that an unreasonable decision is
one that is clearly wrong in the sense just defined.
I like Perry’s analysis very much and I have learned a lot from it. But
I think Perry makes a serious error when he endorses an objective,
perspective-independent account of wrongness. It makes it impossible
for him to account for the evidence-dependence of unreasonableness.
And it leads to two further problems. First, it makes his analysis
overinclusive. Remember Dickins, where a prisoner was granted parole
shortly before his smuggling activities came to light. Had the parole
board panel ‘possessed true information about all relevant facts’,
including the smuggling, and ‘reasoned validly’, it would likely have
decided differently. Perry’s analysis would thus that the panel’s decision
was unreasonable. And yet the panel’s decision was reasonable.
Second, Perry’s analysis is underinclusive. In Hollings, there was no
evidence in support of the considerations that motivated the council
(e.g., that only large care homes were viable). On the other hand, there
was no evidence that the concerns were false (e.g., that some small care
homes were viable). There was no evidence either way. As a result, the
court was in no position to draw any conclusion as to how an authority
‘possessed [of] true information about all relevant facts’ would have
decided. The court could not be highly confident that the decision was
objectively wrong. Perry’s analysis would say that the decision was
therefore reasonable. In fact, it was unreasonable.
Like Perry, Gardner’s understanding of reasons is objectivist.110 As a
result, Gardner’s analysis suffers from the same problems as Perry’s. For
instance, Gardner would have to say that in Dickins the prisoner’s
smuggling gave the parole board panel a reason not to release the

110
J Gardner and T Macklem, ‘Reasons’ in J Coleman, KE Himma, and S Shapiro
(eds), The Oxford Handbook of Jurisprudence and Philosophy of Law (OUP 2004).

30

Electronic copy available at: https://ssrn.com/abstract=4207429

prisoner. That reason is plausibly stronger than the reasons to release
him. So, on Gardner’s analysis, the decision to release the prisoner was
unjustified and unreasonable. As I said, that is not how the law sees
things. Dindjer, meanwhile, tries to remain neutral between
perspectivism and objectivism.111 For the reasons I have set out,
neutrality on this issue is not tenable.

VI. SUMMARY

Unreasonableness is a notoriously obscure standard. For decades, courts

have struggled to clearly explain it. Progress is possible, however.
Drawing on courts’ decisions, we can assemble various self-evident
truths about unreasonableness. Unreasonableness is (i) concerned with
the balance of reasons; (ii) a high standard; (iii) not a form of merits
review; (iv) context-dependent; (v) evidence-dependent; (vi) prima facie
justified; and (vii) indicated by various factors including inconsistency,
oppressiveness, incoherence, and the absence of stated reasons. A good
analysis of unreasonableness will account for all these truths, but
existing analyses fail this test. Gardner’s analysis does not explain why
unreasonableness is a high standard. Dindjer’s does not explain why it is
a less demanding form of review that merits review, or why it is prima
facie justified. Perry’s does not explain why it is evidence-dependent. By
contrast, my analysis – that unreasonableness is just demonstrable
wrongness – accounts for all we know about the standard.

111
See, e.g., H Dindjer, ‘What Makes an Administrative Decision Unreasonable?’
(2021) 84 MLR 265, 291-2.

31

Electronic copy available at: https://ssrn.com/abstract=4207429