Can the IHRA working definition be implemented?

What are the likely consequences of implementing the IHRA working definition in university codes of conduct?


1.    Introduction    
2.    Internal employer procedures    
2.1  Time    
2.2  Panel competence    
2.3  Selectivity    
2.4  Overcommitment    
2.5  The effect of market reforms    
2.6  Employment tribunals: the tip of the iceberg    
3.    Codes of conduct and the determination of cases    
3.1  Discrimination, harassment and victimisation    
3.2  Unacceptable behaviour and lesser charges    
3.3  Tests for unwanted conduct, bullying and harassment    
3.4  Universality and reciprocity, and the risk of institutional racism    
4.    Could the IHRA working definition be integrated with existing codes of conduct?    
4.1  The core definition    
4.2  The examples    
4.3  The HAC caveats    
5.    In conclusion

Executive Summary

If universities ‘adopt’ the IHRA working definition of antisemitism, can they use it? Can it be employed to identify the bounds of acceptable behaviour or adjudicate on allegations? And if it is introduced into codes of conduct for staff and students, what are the implications for teaching and research, for classroom debate and for freedom of expression and academic freedom?

In this document, we examine the IHRA working definition and explore how it could be implemented in employment codes of conduct. We discuss the real-world processes used by university panels to handle complaints and come to decisions, their processes and their competence. We also carefully examine the IHRA working definition, identifying that at its heart is a focus on mental states: on antisemitism as a system of belief, rather than as sanctionable acts. This approach creates an unsafe platform for deciding cases, and undermines freedom of expression.

Our conclusion is that the IHRA working definition is so worded with caveats that it is of limited practical use in determining guilt in cases where allegations are made, and to determine guilt using it as an instrument or reference point would place the university at risk of litigation. By contrast the Equality Act 2010 is the law, and provides a much safer basis for determinations.

The Equality Act prohibits racist acts against Jews, but it does not prohibit criticism of the state of Israel. The only purpose of introducing the IHRA working definition into internal codes of conduct would be to make criticism of the state of Israel a disciplinary offence (possibly caveated by ‘certain circumstances’). But such a step would breach universities’ legal obligations to freedom of expression, core to chartered academic freedoms to teach and research – and for students to learn.

1.   Introduction

In October 2020, UK universities were enjoined by the Education Secretary, Rt Hon Gavin Williamson MP, to ‘adopt’ the IHRA working definition of antisemitism.[1] This is a complex document, containing a definition, a set of examples and – in the version universities are being asked to adopt – two caveats proposed by the Government Home Affairs Committee.

In 2019, the Council of University College London (UCL) agreed to adopt this definition, but set aside the question of implementation. This decision, which was not agreed by Academic Board, led to the formation of a Working Group on Racism and Prejudice to examine the proposal.

At no time was UCL UCU, the recognised academic trade union, consulted over the prospect of implementation of the working definition. The purpose of this paper is to explore the implications of such a move. What would it mean for universities to ‘implement’ the IHRA working definition of antisemitism into their student and staff codes of conduct?

Note: This paper contains preliminary remarks intended for staff to understand the predictable consequences of such an implementation. It does not substitute for consultation!

However, our carefully considered view, in the light of a deep and practical understanding of how UCL and other universities implement current codes of conduct, is that such an adoption presents identifiable and serious risks to the academic freedom of staff and freedom of expression in education curriculum and wider extra-curricular context.

Staff in the front line of this problem at present include two groups:

  1. staff whose areas of research and teaching require them to provide comment and analysis of the state of Israel; and
  2. staff whose national identity is associated with conflict with the state of Israel (Palestinian, Jordanian, Egyptian, Lebanese, etc.), whose rights to self-expression have been ignored.

Although the current matter primarily concerns attempts to prohibit statements critical of Israel, this is unlikely to be the end of the matter. China has recently uttered warnings to academic staff not to denigrate the Chinese state in published work.[2] Any incorporation of this kind sets a precedent that other state actors are likely to follow.

The UCL Working Group published a Report in December 2020,[3] which argued that the IHRA working definition should not be adopted. Instead, it should be withdrawn in favour of robustly pursuing an educative policy and relying instead on existing legal definitions and policies of the institution prohibiting harassment and discrimination. Failing that, the Group said, if adopted it should be treated as educational only, rather than being treated in an instrumental or procedural manner in evaluating the conduct of staff or students.

The main reason given in that Report for this conclusion is that adoption of the IHRA working definition represents a significant threat to both academic freedom and to freedom of expression, protected freedoms that universities are legally required to defend. This is also our view.

However, the Williamson letter is clear that the Government’s intention is for universities to implement the IHRA working definition internally, i.e. in their codes of conduct. Williamson says that he is

strongly urging you [vice chancellors] to formally adopt the IHRA definition and use it on all appropriate occasions.

This is also the view of most supporters of the IHRA working definition. For them, formal adoption is not enough.

The risk is therefore that were UCL or another institution to decline to implement the IHRA working definition but merely adopt it in principle, then this would invite the citation of the IHRA working definition document (see Section 4) in complaints, and compel Panels tasked with hearing complaints to consider it.

A number of lawyers have examined the compatibility of the IHRA working definition with current UK law.[4] They have pointed out that whereas the Equality Act 2010 defines discrimination, harassment and victimisation according to a coherent set of principles and case law (and this case law strikes a balance between freedom of expression, academic freedom and other obligations such as education), the IHRA working definition does not. The body of Equalities legislation and judgement on these matters extends back 45 years to the creation of the Race Relations Act, and internationally through its incorporation of European law. This means that not only does this corpus of law have a lengthy history of case law guiding its interpretation, but also that it has developed alongside other societal advances and potentially competing principles that require balancing against it.

Among the case law corpus are judgements, such as Fraser vs. UCU 2012,[5] which found that the civil obligations of trade unions towards freedom of association and freedom of speech of their members to permit debates at union conferences (such as in this case, on a potential academic boycott of Israeli institutions) outweighed any possible interpretation by Jewish members of the union that the act of hosting such a debate was contrary to the union’s obligations towards inclusion and preventing harassment under race relations and equality law.

This case is particularly relevant to the current discussion regarding potential adoption of the IHRA working definition. Among the questions determined by the court were

  • Can the right to debate by a body be denied due to the risk that someone might make antisemitic remarks during the debate? Answer: no, because (a) antisemitic statements may be made in any context, and (b) the legal obligations towards freedom of expression and association outweigh any additional risks for the particular debate.
  • Second, can the protected characteristics of ‘race’, or ‘religion or belief’ be extended to include support for a political project such as a nation state, and thereby argue that expression of opposing views to that political belief are acts of ‘harassment’? Answer: no, the definition of ‘belief’ in the Equality Act is limited to philosophical stances equivalent to religions, but which do not refer to themselves as such, such as humanism.

Indeed, religions or belief systems are not protected from criticism by the Equality Act, and disputing them, however vehemently and unfairly, is not classed as harassment. Objections to religion are not offences, whereas discrimination or harassment of individuals or groups because of their adherence to a religion is outlawed.[6]

In addition to case law, the Equality Act is supported by the Equality and Human Rights Commission (EHRC) which publishes Codes and Technical Guidance documents which have statutory weight, i.e. may be cited in cases.[7]

In this document we focus on the practical reality of internal employer procedures in a university context. This is an area that trade unions, and in particular UCU, the academic trade union, are well placed to comment on.

UCL UCU represents staff as both complainants and respondents to complaints within the university, and can make general observations about how staff codes of conduct are interpreted in practice. We do not have the same rich experience of the student code, so we will limit our remarks in that respect.

We recognise this is an area of university life that few staff or students are likely aware. Staff are usually introduced to these processes if they make a complaint, receive a complaint or are asked to investigate or adjudicate one. The processes are also strictly confidential, to protect all parties, including witnesses, investigators, panel members and members of Human Resources.

2.   Internal employer procedures

To-date, few analyses of the IHRA working definition seem to have considered whether internal employer procedures are sufficiently robust and competently conducted to avoid the types of error that defenders of the IHRA working definition say would not occur.

However, UCL UCU reps regularly observe that even basic considerations, including balance-of-probability determinations and balancing considerations of academic freedom with alleged offences, are not always properly and fairly applied.

The main problems of these procedures can be summarised as time, panel competence, selectivity, and overcommitment.

2.1 Time

The first source of unfairness is time. As the UCL Working Group noted about two allegations of antisemitism concerning academic activities (an exhibition and a book launch), the length of time involved in investigating these complaints is a major factor that damages the target of these complaints, even though the complaints were not accepted.

At UCL, formal investigations (for disciplinary matters against staff or grievance matters raised by staff) nearly always take in excess of 3 months, and a typical duration ranges from 6 months to a year. The risk is simply that a minor or unmerited complaint can have a serious punitive effect simply due to the time it takes to discharge. A staff member may face allegations, be sworn to confidentiality, and have the allegation hanging over them for more than a year.

The Courts expect employers to discharge investigations in 3 months or less. The additional complexity involved in the IHRA working definition is liable to cause a further prolongation of the process.

2.2 Panel competence

In the case of employee disciplinary panels, unless a matter is sufficiently serious to warrant dismissal as a sanction (termed ‘gross misconduct’ cases), in UCL the panel will consist of a single individual manager, advised by a member of Human Resources. In gross misconduct cases the panel has a membership of three people.

Internal procedures rely on non-experts making civil judgements on the ‘balance of probabilities’ rather than a criminal level of proof. Legally this simply means that a panel should ‘be more than 50% confident’ in their decision, but this criterion is often wrongly interpreted with a high degree of subjectivity.

Common errors of this kind include a tendency to rely on testimony rather than material evidence, and a failure to properly examine exculpatory evidence.[8] In harassment cases there can be a confusion between Prevention of Harassment tests (objective and subjective tests under criminal law) and the Equality Act tests (see Section 3).

A second, related error confuses this decision-making threshold with the procedural obligation on a complainant to establish the facts (the ‘burden of evidence’). Allegations are entertained even when a complainant cannot place the event in time or find any contemporaneous evidence or witnesses.

The fact that civil internal panels may not be legally expert may lead to errors in determining categories of offence (such as ‘bringing the university into disrepute’, or identifying cases of harassment, see section 3.3 below), and interpreting the right balance between competing legislative principles (such as between obligations towards academic freedom and education, and obligations to prevent harassment).

Staff can even find themselves threatened with disciplinary sanction because they allowed themselves to be accused of an offence (thereby bringing disrepute), not because the allegation is true.

2.3 Selectivity

One of the ways in which a selection bias may operate is simply that those who complain the loudest get heard.

Secondly, organisations are often selective in practice as to which complaints they choose to investigate and which they do not. This can cause substantial inequity, irrespective of the policies that might exist on paper.

Thirdly, some individual’s complaints are given higher priority than others. Normally management complaints against staff (disciplinary complaints) generally take precedence to staff complaints about managers (grievance complaints). A staff member who believes that they were singled out by a manager for disciplinary action whereas others are not are not well provided for – they have to first defend the disciplinary case before raising questions of inconsistency and inequity.

Student complaints are increasingly prioritised over staff ones, a fact that may appear surprising to students! They may be taken up as disciplinary matters by managers rather than investigated first, heard and only if upheld, then treated as grounds for disciplinary action.

2.4 Overcommitment

We have already noted that investigations are lengthy and time-consuming processes. In the case of disciplinary investigations into employee conduct, the focus of the investigation (commissioned by management) is narrow: it is on the individual’s actions and their personal culpability. Largely excluded from the investigation are the actions of others and the surrounding circumstances. In fact, from the commencement of disciplinary procedures to the hearing itself, the entire exercise is premised on determining ‘how guilty is X?’ rather than ‘what happened?’.

Due to the time and personnel costs expended, there is inevitably a tendency for panels to find some fault and to issue a Formal Warning to the subject. The question considered at this stage is not ‘whether X is responsible?’ for some error, but merely ‘how guilty is X?’ and therefore what level of sanction is applicable.

Union representation often results in a lower sanction for the staff member, but sadly, rarely does it lead to no sanction at all. There is clearly considerable social pressure on the Panel to find some level of responsibility for fault that would justify the cost of the entire process.

Yet the union does see low-level, first-offence cases that should never have been brought, where the resulting Oral Warning seems to be a face-saving measure for managers. Naturally this leads to inequity, with colleagues who commit the same ‘offence’ in other cases receiving a telling-off at most. The six month-plus wait for the hearing and an adjudication is also not taken into account, even though this can have the biggest penalty.

2.5 The effect of market reforms

An additional complicating factor since 2010 has been the rise of students as ‘customers’, with complaints potentially referred to the Office of the Independent Adjudicator. This has placed greater weight on student complaints than hitherto, and encouraged panels to find in favour of student complaints, whether against individual staff or departments. It has also encouraged the practice of taking forward student complaints promptly to disciplinary investigation, even if the complaint itself is misconceived.

Students, particularly ex-students who cannot be sanctioned, face no penalty for making allegations that they know to be untrue (termed ‘malicious allegations’). A staff member who makes an allegation of this kind is expressly warned they may be disciplined. Meanwhile staff on the receiving ends of complaints can be dismissed and be subject to public harassment and humiliation.

Institutional reputation management is another factor encouraging panels to support student complaints. Whereas staff may be disciplined for speaking out about allegations made against them, students are under no real sanction, especially if they have left. The rise of social media has placed universities in a difficult position, and in some cases, universities have simply failed to defend the academic freedom of individual staff.

2.6 Employment tribunals: the tip of the iceberg

Trade unions can cite hundreds of examples where employer judgements err in evidence and law; cases referred to the Employment Tribunals (ET) service are a tip of a large iceberg. It is only at this point of referral are legal matters fully tested.

However, the cases referred to ET tend to be a very small subset of cases heard internally. There are several reasons for this.

First, the ET service is slow and expensive, and the damage is usually done to the accused. A wait of over a year is typical; cases involving discrimination and harassment are costly for the complainant (a week’s court time per case being the expected minimum); courts rarely wish to overturn the verdict of an internal civil panel, even when the internal panel errs on matters of law. (This may change were the IHRA working definition introduced, as there are major areas of inconsistency with the Equality Act and case law.)

Second, with a few exceptions, the outcome is usually only financial compensation. ETs are not libel courts and do not consider reputational damage to the individual as a factor to be compensated for. ETs occasionally, but rarely, make reinstatement orders (instructing employers to reinstate dismissed employees).

Third, the majority of internal cases do not lead to dismissal. Instead, they lead to internal sanction, in the form of a Formal Warning from the employer not to repeat the alleged conduct. If a staff member is given such a warning for an allegation of antisemitism, even if the allegation is unmerited, the most likely outcome is that the staff member changes their conduct, lowers their profile, and does not take their employer to court. Although taking one’s employer to an ET is a legally protected right, fear of further sanctions from their employer is a strong deterrent.

3 Codes of conduct and the determination of cases

Employers are advised by ACAS to have employee codes of conduct that prohibit acts of discrimination, harassment and victimisation, defined under the Equality Act.[9]

Employee codes of conduct are the principal mechanism (alongside training and specific instructions) used by employers to address their vicarious liability for the conduct of their employees. They may be developed independently from student codes of conduct, although the Equality Act has tended to lead to a higher level of standardisation of personal conduct codes.

3.1 Discrimination, harassment and victimisation

Under the existing Equality Act, both discrimination and harassment must be ‘related to’ a protected characteristic – in the case of antisemitism, this is of race or religion, i.e. the characteristic of being Jewish in ancestry; or the characteristic of practising the Jewish religion in one of its many forms.

Most cases of discrimination, including race discrimination, requires a ‘comparator’ – someone who is not discriminated against, in contrast to the person who is. Harassment does not require a comparator, rather the key test is whether the act is unwanted, and sufficiently serious as to justify sanction.

In cases of direct discrimination and harassment, perception and association also apply: someone may be discriminated or harassed because the perpetrator believes them to be Jewish, likewise someone may be discriminated or harassed because they are associated with Jewish people. The Equality Act also recognises indirect discrimination, where procedures are implemented that as a by-product cause Jews a disadvantage.

Victimisation is a second-degree claim, intended to protect complainants or would-be complainants of harassment or discrimination, or indeed victimisation itself, under the Equality Act.

3.2 Unacceptable behaviour and lesser charges

Although not defined in law, other categories of offence, such as ‘unacceptable behaviour’ and ‘bullying’ may also be referenced in these codes. These can be seen as representing step changes, with ‘unacceptable behaviour’ being at the lowest level worthy of sanction:

  1. unacceptable behaviour is usually understood as unwanted conduct directed toward an individual;
  2. bullying may be defined as unacceptable behaviour which has the purpose or effect of violating a person’s dignity or creating an intimidating, hostile, degrading or humiliating environment;[10] and
  3. harassment may be defined as bullying related to a protected characteristic.

3.3 Tests for unwanted conduct, bullying and harassment

In this way one can evaluate an act at three stages:

  1. Is it unwanted? Then it meets the test for unacceptable behaviour;
  2. Does it have the purpose or effect of violating someone’s dignity or creating an intimidating (etc.) environment for them? Then it is bullying;
  3. Is the act related to a protected characteristic? It is harassment.

The first stage, namely determining whether an act is ‘unwanted’, is further elaborated in case law as consisting of three tests:

  1. Whether the act was previously challenged by the recipient, so that the person who carried it out knew in advance that it was unwanted;
  2. Whether the act was intrinsically so serious that a reasonable person would conclude that it would be unwanted by the recipient; or
  3. Whether the act in combination with its context was so serious that would cause a reasonable person to conclude that it would be unwanted by the recipient.

This concept of ‘unwanted conduct’ must be interpreted against a threshold, to avoid a situation where all life stops for fear of causing offence. This is called the reasonable person test.[11] Understanding such a threshold requires care, and it is where case law is relevant. An act may be expressly objected to by a recipient – such as when a manager insists to an employee they must attend a meeting in order to discipline or dismiss them – without the act being characterised as ‘unwanted’ according to these criteria. This is because in many circumstances it may be unreasonable for a recipient to object, for example, because they signed on to a course in which they were warned they may be presented with upsetting material, or because they accepted the terms of a contract of employment.

Similarly, the threshold of what would constitute a ‘serious unwanted act’ under point 2 or a ‘serious unwanted act in the context’ under point 3 is also a matter of case law.

Antisemitic acts, matters which the IHRA working definition is solely concerned with, will primarily apply to a subset of acts prohibited by the Equality Act, namely harassment.[12] It is conceivable that an attempt might be made to demand that universities stopped particular courses or modules on the grounds they were “offensive to Jews” citing discrimination rather than harassment, but such an argument would face the same legal hurdles as Fraser.

This means that the IHRA working definition will primarily be interpreted at the interface between the act and the characteristic, in other words, how the act and the characteristic are ‘related to’ each other.

The question we will return to is whether it serves any useful protective purpose in doing so, or does it undermine the structure of decision-making outlined above that the law expects?

At this point it is worth briefly noting that the Equality Act is far broader than the IHRA working definition, by including within its scope, and prohibiting

  1. acts of harassment where the state of mind of the actor (intent) is not established
  2. acts of direct discrimination
  3. acts of indirect discrimination

3.4 Universality and reciprocity, and the risk of institutional racism

At certain points the IHRA working definition (see Section 4 below) would, if implemented, effectively rewrite legal definitions of harassment in the Equality Act 2010 by extending protections afforded to Jews to a nation state. It creates a presumption against voices critical of the state of Israel (termed ‘anti-Zionism’) and labels them as potentially racially prejudiced against Jewish people (antisemitism). This is despite the fact that organisations (Israel itself) and beliefs in them (‘pro-Zionism’) are not protected under the Equality Act, and that criticism of Israel is a legally protected act of political freedom of expression.[13]

The document is clearly prejudicial to Palestinian staff and students (and those from other countries that have been in conflict with Israel over the decades) because it creates unclear barriers to their freedom of expression to testify to their own lived experience.

This could be seen as institutionally racist, first by applying a higher standard to those staff and students in policing their speech, but second, by prioritising the rights of another nationality over theirs, particularly likely to be exacerbated in circumstances of unresolved national conflict. This is entirely foreseeable discrimination related to the protected characteristic of Palestinian nationality, if not acts ‘violating their dignity or creating a hostile environment for them’, i.e. harassment.

The universality of the Equality Act and the particularity of the IHRA working definition are not easily reconciled.

4.   Could the IHRA working definition be integrated with existing codes of conduct?

4.1 The core definition

The IHRA working definition does not have the same methodology of determination as harassment under the Equality Act. The core definition reads:

Antisemitism is a certain perception of Jews, which may be expressed as hatred toward Jews. Rhetorical and physical manifestations of antisemitism are directed toward Jewish or non-Jewish individuals and/or their property, toward Jewish community institutions and religious facilities

The first notable aspect of the core IHRA definition is that it defines antisemitism as a mental state, ‘a certain perception of Jews’. However, mental states are almost impossible to determine, except in the procedurally trivial cases where explicit hate speech is uttered, or hatred admitted.

In practice this is not a diagnostic tool that is readily employed. It invites investigation managers and panel members to ascribe beliefs to the accused, apparently on an all-or-nothing basis. Such allegations are liable to be libellous if unsupported by considerable evidence.

The question becomes simply ‘Is person X antisemitic’, requiring a high standard of proof, rather than ‘Does X understand that their actions may be interpreted as antisemitic, and what do they say in response?’

By contrast, the Equality Act allows for the fact that potentially antisemitic views may not be completely formed. The question is about purpose and effect. One may ask how much the individual ought to have known that their actions could be seen as antisemitic by the recipient, tested through a process of repudiation of unwanted conduct.

Since it is ultimately impossible to know what an individual thinks, focus will likely be directed towards the second sentence in the core definition, which – like the Equality Act – concerns acts: ‘rhetorical and physical manifestations’. There is some overlap between the Equality Act’s definition of harassment ‘related to’ a protected characteristic and this second sentence. However, there is a crucial difference.

In the Equality Act, antisemitism is defined in part by the establishment of a relationship between the act and ‘Jewishness’. It is in establishing that relationship that is crucial to identifying the antisemitic nature of the act.

By contrast, in the IHRA working definition, antisemitism is pre-given, and the relationship is purely in relation to the individuals who are targets of acts. However, now the relationship is drawn so widely that it is non-definitional. Indeed, this approach is contrary to conventional definitions of racism. It is not the fact that a Jew (or non Jew, and/or their property etc.) is the recipient of an offensive act that causes the act to be antisemitic, but the nature of the act itself. Even deliberately causing offence, rudeness, etc., is not the same as racism.

The core IHRA working definition therefore does not add anything useful and operational to the Equality Act definition of harassment itself, and engages in a psychologism which requires a very high standard of proof.[14]

In addition, it does not address the other manifestations of antisemitism that are covered by the Equality Act, including discrimination.

For example:

  • A focus on mental states specifically rules out any ruling on consequential (indirect) discrimination where there is no evidence of conscious animus towards Jews. These may be omissions, rather than direct acts. In order to bring these under the working definition and its focus on mental states it would be necessary to extend it to ‘unconscious bias’, which risks being purely speculative. By contrast the Equality Act has ample case law on indirect discrimination.
  • By failing to define antisemitism in terms of the act itself, antisemitism in the core definition can only be established on the basis of the target being Jewish. Although it says that non-Jews may be the target of antisemitic acts, this is not a complete definition: the core working definition has no mechanism for determining that these acts are indeed antisemitic.

By contrast, the Equality Act will additionally categorise an act as antisemitic where

  • someone is targeted for associating with Jewish people,
  • someone is targeted because they are perceived to be Jewish, or
  • the act itself is connected to Jews, or historical and cultural associations (such as swastika graffiti or a tattoo) irrespective of the person targeted by the act.

Although all of the above are common sense definitions of antisemitic acts, they are not found in the core of the IHRA working definition. Instead, the definition only extends to some of these cases through additional examples.

What would happen were the core definition alone introduced into Codes of Conduct without examples or caveats? Let us assume that the definition is introduced in the form of an addition to, rather than as a replacement for, the Equality Act. In this way some of the weaknesses noted above are avoided.

What problems remain?

The main risk is that it requires Panels to speculate on the mindset of the accused. As we have seen this is very different from the Equality Act, which prohibits acts of harassment that have the purpose or effect of violating the recipient’s dignity or creating a hostile environment for them (etc).

Naturally, once speculating about beliefs, a panel may inappropriately attribute beliefs to them that they do not hold, and proceed to find them guilty of holding these beliefs. Given what we have already noted about the competence of panels (which may be a single individual), this represents a serious risk. By contrast, civil courts tend to avoid interpreting the mindset of the accused, as it is usually a very unsafe basis for a determination.

Secondly, it is also liable to prolong investigations, as investigating managers will feel obliged to evaluate this question, essentially by seeking evidence to support beliefs about the individual. Yet the ACAS Guide to Disciplinary Investigations[15] require that employers conduct investigations focused on finding facts (and examining both evidence for and against allegations), and not relying on the opinion of investigators.

Finally, the fact that the IHRA working definition with examples exists beyond the walls of the employer, but the employer has only adopted the definition without examples, also presents the risk that in practice those examples will be relied upon, even if they are expressly disavowed.

This risk increases with the existence of continuing social pressure to adopt these examples, which – having conceded the core definition – will likely be increased. Yet it is the methodology of the IHRA working definition, centring on a psychological definition of antisemitism, that is fundamentally flawed.

4.2 The examples

Others have commented on the rather eclectic and arbitrary nature of the examples included. They are expressly not exhaustive (they ‘may serve as illustrations’), but they offer the only available mechanism to establish that an act is antisemitic according to the definition.

Many of the examples assert a relationship between antisemitism and the State of Israel. Yet political criticism of Israel, including criticism that many Jews may find offensive, is a protected right of freedom of expression. In some examples the introduction of a reference to the state of Israel adds no probitive value, but is done anyway.

First, we read

To guide IHRA in its work, the following examples may serve as illustrations: [16]

Manifestations might include the targeting of the state of Israel, conceived as a Jewish collectivity. However, criticism of Israel similar to that leveled against any other country cannot be regarded as antisemitic. Antisemitism frequently charges Jews with conspiring to harm humanity, and it is often used to blame Jews for ‘why things go wrong.’ It is expressed in speech, writing, visual forms and action, and employs sinister stereotypes and negative character traits.

If the reader was in doubt, this passage makes clear that the intention of the examples is to place criticism of the state of Israel at the forefront of definitions of antisemitism.

Again, we see an upfront reliance on mental states: ‘conceived as a Jewish collectivity.’ Apart from being highly ambiguous, who, in this definition, is supposed to perform this act of conception? Is it the accused, the complainant or the panel? (We have discussed above the problem of relying on unstated beliefs of the accused to find against them.)

The caveat that follows – that ‘criticism of Israel similar to that leveled against any other country’ is not antisemitic – is worthless from a diagnostic perspective. Aside from the vagueness of ‘similar to’, there is no international yardstick of acceptable criticism of nation states against which to determine this question! Were one to attempt to construct one, such an exercise could restrict acceptable criticism in such a way as would inevitably constrain freedom of expression and academic freedom. It would be a weapon in the hands of every dictatorship.[17]

The remainder of the passage is an attempt to characterise a type of specific ‘classical’ antisemitism developed from feudal antisemitism through escalation under fascist persecution in Nazi Germany into coded conspiracy theories expressed by present-day fascist organisations. Whereas this provides an enumeration of some instances of antisemitic tropes, they are not definitive of antisemitism and could not be exhaustive.

The next paragraph then introduces a series of bullet points, examples which are also expressly non-exhaustive. They have been numbered below for reference.

Contemporary examples of antisemitism in public life, the media, schools, the workplace, and in the religious sphere could, taking into account the overall context, include, but are not limited to:

1 Calling for, aiding, or justifying the killing or harming of Jews in the name of a radical ideology or an extremist view of religion.
2 Making mendacious, dehumanizing, demonizing, or stereotypical allegations about Jews as such or the power of Jews as collective — such as, especially but not exclusively, the myth about a world Jewish conspiracy or of Jews controlling the media, economy, government or other societal institutions.
3 Accusing Jews as a people of being responsible for real or imagined wrongdoing committed by a single Jewish person or group, or even for acts committed by non-Jews.
4 Denying the fact, scope, mechanisms (e.g. gas chambers) or intentionality of the genocide of the Jewish people at the hands of National Socialist Germany and its supporters and accomplices during World War II (the Holocaust).
5 Accusing the Jews as a people, or Israel as a state, of inventing or exaggerating the Holocaust.
6 Accusing Jewish citizens of being more loyal to Israel, or to the alleged priorities of Jews worldwide, than to the interests of their own nations.
7 Denying the Jewish people their right to self-determination, e.g., by claiming that the existence of a State of Israel is a racist endeavor.
8 Applying double standards by requiring of it a behavior not expected or demanded of any other democratic nation.
9 Using the symbols and images associated with classic antisemitism (e.g., claims of Jews killing Jesus or blood libel) to characterize Israel or Israelis.
10 Drawing comparisons of contemporary Israeli policy to that of the Nazis.
11 Holding Jews collectively responsible for actions of the state of Israel.

The first observation we might make is that all of these examples are heavily caveated by ‘taking into account the overall context’. In the most borderline cases it is clear that context is everything.

But the document does not provide examples of which factors should be present in the context that might lead one to decide whether or not a particular instance was antisemitic. Nor does it provide a mechanism for appropriately weighing up this context.

Examples (1), (2) and (3) are not particularly controversial, although again the wording is poorly defined. Thus a ‘mendacious’ statement (2) is simply a lie, whereas a ‘dehumanizing’ one is intended to create the impression that the person is less than human. Some examples are caveated to the point that even the most egregious examples are insufficient to be used to determine a case! Thus (1) ‘calling for… or justifying the killing of Jews’ is caveated by ‘in the name of a racist ideology or an extremist view of religion’. This is strange: ‘calling for the killing of Jews’ seems self-evidently racist, liable to be criminal incitement, and the naming of a racist ideology or not would be irrelevant! Perhaps what is meant is that it is not necessarily antisemitic to comment on the Israel-Palestine conflict and for onlookers to justify the right of Palestinians to ‘kill or harm’ Israelis. But the entire statement is imprecise.

Examples (4) and (5) referring to Holocaust denial are not particularly controversial. Indeed (5) is too narrow. Under the Equality Act, it is unnecessary to establish a connection between Jews as a people (or the state of Israel) to establish that Holocaust denial is ‘related to’ Jews, simply because of the historical fact that Jews were one of the targets of (and comprised by far the largest numbers of victims of) the Nazi Holocaust.

Example (6) is more complicated. A Jewish citizen may in fact be ‘more loyal’ to the state of Israel than to another state without this being an issue. The allegation itself is not automatically antisemitic, another case where ‘context’ is everything. Thus in 1987 the spy Jonathan Jay Pollard plead guilty to spying for Israel against the United States: was this case antisemitic?

There is international support for Israel among many Jews living in other countries. Such support is entirely permissible, but since this is a fact, drawing attention to such identification may also be legitimate comment and not ipso facto antisemitic. Similarly, legitimate debate in the classroom may cause a student to express favouritism towards the interests of their own or another state over others. It is entirely possible that non-Israeli Jewish students may express themselves freely in support of Israel such that this question may naturally arise. Should a lecturer shut down a classroom discussion on the risk that they be accused of permitting antisemitic statements to be made?

Where Example (6) would be antisemitic is where it concerns the ascription of a generic negative belief to all Jews: a prima facie example of racism. However, this is covered by Example (11). Racism would also be inherent in a situation where Jews were singled out to prove their loyalty to the British state (a demand that Jews faced in the first half of the 20th Century, and that British Muslims have faced for decades – including by Government ministers).

Example (7) is misconceived. The ‘right of self determination’ in the sense of the right to set up and create a nation state on a parcel of land, is by definition one that necessarily impinges on the rights of others. As such, the right is always a limited one, and criticism of the exercising of that right is in the realm of political criticism: a protected right of freedom of expression! The claim that Israel is, or has become, a racist endeavour is by definition a political claim.

Example (7)’s basis is in the assertion that Jews are denied ‘self determination’ unless Jews are permitted to form a national state, Israel. However, to argue that Jews are uniquely denied this right requires first that self determination is a universal right. Consequently, it must involve reciprocity. It must address the question of Palestine. It would be racist to refuse rights to Palestinians offered to Israelis.

Therefore, the direct consequence of accepting Example (7) is that the right to self determination must be offered to Palestinians. Expressions of support for Palestine would, like expressions of support for Israel, then also become protected rights! Rendering criticism of Israel impermissible would also mean that criticism of Palestinian Authority becomes impermissible as well (and, potentially by extension, state and non-state actors on either side). Since self determination does not exclusively concern an actually-existing state but an ambition towards a national state (cf. Kurdistan), it would also mean that criticism of the idea of regaining an historic Palestine on the same land would also be impermissible!

Freedom of speech, political criticism and analysis are thereby silenced.

Example (8) referring to ‘double standards’ is a repetition (in negation) of the statement regarding ‘criticism of Israel similar to that levelled against any other country’, and fails for the same reason. It would be left to panels to try to decide whether or not a criticism reflected a ‘double standard’. As written, this statement places a high burden on a defendant to show that their criticism of the Israeli state is not more damning than they would apply to another country. Again, this becomes a focus on the systems of belief of the accused, rather than on the act itself.

Even were a double standard applied, does this actually substantiate antisemitism? Are individuals not permitted to take sides? Thus it might represent personal investment in the plight of the Palestinian people, or political support for their self-determination. This example is clearly intended to brand criticism of Israel as antisemitic, but in fact this is an example where context is everything, and the legal obligation to permit freedom of expression must take priority.

Example (9) is another reference to classic antisemitic tropes. It simply states that antisemitism includes classical antisemitism.

Example (10) is much more problematic. On the one hand, abusing any nationality or racial group, whether contemporary Israelis or Germans, by crudely comparing them with the Third Reich would be racist because it would be an act of abuse connected to their nationality or race. To determine whether this example has any validity is to consider whether the criticism of an act or policy of Israel may not a priori involve a comparison with a comparable act or policy of Nazi Germany.

Outlawing any comparison at all between any nation and Nazi Germany is problematic for other reasons. Banning such comparisons would undermine the modern study of historical fascism, in establishing the unique and non-unique features of that society. Again, were this example to be salvaged it demonstrates the reliance on ‘context’. One might imagine a context when such an utterance would be antisemitic, but the example wording is not definitive.

Finally, Example (11) is not controversial, as holding ‘the Jews’ collectively accountable for anything is the definition of antisemitism!

In summary, the examples add to the core definition by explicitly placing criticism of Israel (a political act), in the space of abuse of Jews (a racist act). Were they added to the core definition in an internal employment code of conduct, they would largely be undecidable as written, leaving a considerable space for interpretation (and prejudice).

They are vulnerable to interpretation by Panels and lobbying by parties on either side of an allegation due to the ambiguity of the examples and the requirement to interpret the context.

We noted that the core definition focused on the mental state of the accused, and did not actually define antisemitism. It was left to the examples to provide a definition. However, the examples are dependent on context. Under their literal reading, some of the examples ((6), (7), (8) and (10)) are incompatible with the legal obligations of universities towards their educative mission, or towards actively protecting freedom of expression and academic freedom.

4.3 The HAC caveats

In 2016 the Home Affairs Committee of the House of Commons proposed the following caveats to the adoption of the IHRA definition:

1. It is not antisemitic to criticise the Government of Israel, without additional evidence to suggest antisemitic intent.

2. It is not antisemitic to hold the Israeli Government to the same standards as other liberal democracies, or to take a particular interest in the Israeli Government’s policies or actions, without additional evidence to suggest antisemitic intent.

Do these additional caveats actually change the meaning of the IHRA working definition and its examples?

They have no effect on the core definition, which still rests on adjudication over the mental states of the accused. What they do is state that ‘additional evidence to suggest antisemitic intent’ is required when criticism of, holding to the same standards as other liberal democracies or taking a particular interest in the Israeli Government.

The scope of these expressions (‘criticism’, ‘holding to the same standards’, or ‘taking a particular interest in’) is unclear. A non-restrictive interpretation of these caveats is that they would reverse most of the effect of the specific prohibitions in the Examples, potentially including ‘criticism’ or ‘particular interest’ of the most impassioned and negative kind.

However, a careful reading shows the wording actually refers to the Government of Israel only, and says nothing about Israeli citizens, non-governmental forces, etc. Nor does the wording refer to the Israeli state, i.e. the political entity, including its foundation. Thus it is unclear whether it would permit (for example) ‘claiming the state of Israel was a racist endeavour’ (Example (7)), which is not limited to a criticism of a particular Government.

Much has been made about the necessity to identify antisemitic intent in the caveats. Intent is a crucial factor in criminal cases and in some civil cases.

But intention, like mental states, is difficult to decide, and internal panels frequently demonstrate that they are not competent to fairly determine it. The Equality Act definition of harassment, by relying on the act having the ‘purpose or effect of’ violating dignity or creating a hostile environment (etc.), avoids this reliance on intent.

What would be required would be for a Panel to either (a) speculate on the mental state of the accused (a central problem in the IHRA core definition) or (b) identify a relationship of any such statement to other, self-evidently antisemitic acts. The former has high risk for the institution and natural justice; the latter renders the first act effectively irrelevant to the determination of a case.

Fundamentally, the problem is that the IHRA working definition is engaged in discussing a psychological internal ‘mental state’ model of antisemitic belief, whereas the Equality Act rests primarily on external evidence of antisemitic acts. Equality Act adjudications of harassment or discrimination simply do not require a decision as to whether the individual had antisemitic beliefs or intent, even if, in grievous cases, courts are prepared to comment on the beliefs of the accused.

This difference renders the two approaches difficult to reconcile in codes of conduct. It may make cases undecidable and expose the university to liability. But it also brings internal processes concerning conduct into a field of allegations about beliefs. Such a repositioning of codes risks creating a ‘thought-police’ culture in universities that is inimical to the university’s educative mission, or to freedom of expression and academic freedom.

5.   In conclusion

Taken as a whole, can the IHRA working definition, core, examples and caveats, be introduced into internal codes of conduct in higher education?

For good reasons, most staff are unaware of the significance of problems with internal complaints and disciplinary procedures in dealing with complaints.

As we have noted, complaints procedures are not courts. Whereas courts may penalise malicious complainants and vexatious complainants (whose complaints have no reasonable prospects of success), internal procedures may hear complaints from individuals who cannot be penalised, and are therefore not accountable for them.

Complaints procedures are lengthy, and are conducted by staff who are rarely legally qualified. If the complaint gets to an employee disciplinary panel, the fact that it has got so far creates an incentive to find fault and issue a Formal Warning.

Panels are also not qualified to take a view on the mental state of an accused person, so unless someone admitted antisemitic prejudice, findings of antisemitism would depend on an analysis of the accused’s speech and writings in detail. That is, they depend on acts, not beliefs.

Since a finding of antisemitism against an individual staff member can be career-ending, the stakes are extremely high for the employee and for the institution.[18] This means that cases that find against individuals are liable to end in litigation.

Despite its length and complexity, the IHRA working definition (and all of its examples and caveats) adds barely a useful word to the Equality Act 2010 definition of harassment. What is present is contradictory and confusing, has no relationship with case law and legal obligations, and is liable to create serious problems for universities were it used to adjudicate cases. The EHRC Code of Practice for Employment and Technical Guidance for Further and Higher Education must also be considered in interpreting any such lawful incorporation.

On the other hand, the likely consequence of adoption into internal procedures is to encourage mistaken and sometimes malevolent allegations of antisemitism, and create an environment where it is perceived to be reasonable to conflate criticism of the state of Israel with acts of racism. This is where it will have the ‘chilling effect’ noted by the UCL Working Group.

These are very serious problems in a university context in ensuring that the freedom of expression of all staff and students, as well as freedom of academic research and teaching, are protected.

As well as targeting Palestinian staff and students and their supporters, the IHRA working definition is a direct attack on staff who teach and research in any area where the foundation, history and politics of the state of Israel is discussed.

The irony of a document advocated in the name of opposing prejudice against Jews but which will likely have the effect of silencing Jewish scholars in disciplines concerning Israel, Judaism and history, as well as Middle East politics, cannot be ignored.

Implementation is a recipe for inviting ill-considered allegations and arriving at unjust conclusions. It also dangerously risks discrediting the fight against real antisemitism.

[4] These include Hugh Tomlinson QC, Geoffrey Robertson QC and others. See e.g. https://freespeechonisrael.org.uk/ihra-opinion

[6] Clearly there is a line between criticism on the one hand and racist ridicule, such as deliberately circulating cartoons of the prophet Muhammad in order to offend Muslims as Muslims. Where that line is drawn is a matter of concrete context – and much debate! A parallel argument applies to where opposition to the state of Israel crosses over into antisemitism.

[8] Those who defend the IHRA working definition because of an additional burden of demonstrating intent should recognise that internal employer panels are not equipped to properly determine the intention of an individual, save, of course when that individual confesses their intent. Worse, if this standard is used, even prima facie Holocaust denial could be deemed not antisemitic provided that the denier claims that they did not intend to be antisemitic! Under the Equality Act no such requirement to prove intent exists. See Section 4.

[9] ‘Harassment’ has another criminal, legal definition, unrelated to the Equality Act, where it refers to targeting an individual and placing them in fear of injury or death. In this case the Police would be involved. The Equality Act also prohibits sexual harassment or less favourable treatment because they submit to or reject sexual harassment or harassment related to sex or gender reassignment. We are concerned here only with the relevant general definition.

[10] UCL’s most recent Prevention of Bullying, Harassment and Sexual Misconduct policy https://www.ucl.ac.uk/equality-diversity-inclusion/dignity-ucl/prevention-bullying-harassment-and-sexual-misconduct-policy says ‘According to ACAS guidance, bullying is intimidating, hostile, degrading, humiliating or offensive behaviour, through means which have the purpose or effect of violating a person’s dignity or creating an intimidating, hostile, degrading, or humiliating environment. Bullying usually involves a repeated course of conduct.’ Apart from the additional sentence, this definition is in fact the Equality Act 2020 s26 definition of harassment shorn of the connection to a protected characteristic.

[12] Indeed, as others have pointed out, the IHRA working definition is far narrower than the Equality Act definition of harassment, referring to statements rather than other acts. Nonetheless the locus of the matter is here.

[13] It may be that a future Government attempts to rewrite the Equality Act in order to perform such an introduction. However, as we have noted, there are a number of problems that inevitably arise, including that either Jewish self-determination is special, or it is a universal trait shared by Palestinians; and the Equality Act will present a real impediment to freedom of expression and academic freedom.

[14] It is worth noting the Noah Carl case in this context. Cambridge needed a major investigation, and in the end the University focused on his scholarship standards rather than alleging that he held racist and antisemitic views. https://www.theguardian.com/education/2019/may/01/cambridge-university-college-dismisses-researcher-far-right-links-noah-carl

[16] A pedant (or a lawyer) might point out that this sentence expressly repudiates any connection between the core definition and the examples that follow. On this reading the remainder is pure commentary and not analysis. It is intended only to ‘guide the IHRA in its work’ and not to bear on the IHRA working definition! This is yet one of many apparent weaknesses which in fact show how the document was never intended for codes of conduct, but to guide a particular organisation in its work in classifying potentially antisemitic incidents.

[17] The HAC caveats attempt to avoid this objection by referring to ‘liberal democracies’ as if they self-evidently have high standards. (A little more attention to British history, the Empire and the slave trade might be in order. A similar point would apply to other European colonial powers.)

[18] Allegations of antisemitism against scholars should be seen as requiring the kind of evidential basis as Irving v Penguin Books Ltd and Lipstadt 2000, in preparation of which Penguin commissioned a research project into David Irving. See http://www.bailii.org/ew/cases/EWHC/QB/2000/115.html