Epistemic corruption and the research impact agenda

Contemporary epistemologists of education have raised concerns about the distorting effects of some of the processes and structures of contemporary academia on the epistemic practice and character of academic researchers. Such concerns have been articulated using the concept of epistemic corruption. In this article, we lend credibility to these theoretically motivated concerns using the example of the research impact agenda during the period 2012–2014. Interview data from UK and Australian academics confirm that the impact agenda system, at its inception, facilitated the development and exercise of epistemic vices. As well as vindicating theoretically motivated claims about epistemic corruption, inclusion of empirical methods and material can help us put the concept to work in ongoing critical scrutiny of evolving forms of the research impact agenda.


Introduction
About a decade ago, the mere mention of the word 'impact' in many academic circles caused colleagues to react with words of protest, sceptical glares, raised eyebrows, open scorn or bemused resignation -and sometimes, all of these at once. Even if most academics, then as now, would in fact accept the idea that academic research should hope to influence the world for the better (Chubb and Reed, 2017;Trowler, 2001), impact was regularly resisted as an academic mandate indicative of a structural move towards a more marketised higher education system (Holmwood, 2014;Watermeyer, 2016). A common reason for hesitation and resistance was a sense that the obligation to have such influence -or impact -might not be smoothly consistent with academics' wider roles as teachers, researchers and scholars as informed by, for instance, Mertonian norms or their allegiance to entrenched Enlightenment ideals (Chubb and Watermeyer, 2017;Merton, 1942). A pursuit of impact, many argued, should not come at the expense of our fidelity to other academic values and freedoms (Holmwood, 2014;Ladyman, 2009).
By 'the impact agenda', we mean the move to attempt to bureaucratically assess the social, cultural and economic impact of research. When the impact criterion was first introduced into the research funding process in the United Kingdom and when concurrent developments began in Australia, this move was met with significant criticism from the field of higher education studies and from the academic community more broadly. Early critics in both contexts were quick to offer both principled and practical worries about the potentially deleterious effects of institutionalised mechanisms of impact-tracking or the ways that the impact agenda may jeopardise academic authenticity with respect to gaming, autonomy and identity (Battaly, 2017;Chubb, 2017;Battaly, 2013;Henkel, 2000;Martin, 2011;Smith et al., 2020;Watermeyer, 2016). We grant that more recent work notes some positive effects (Weinstein et al., 2021) and that there have been concerted efforts in the policy sphere to shake up the research system, like the government consultations on the efficacy of research evaluation in the Nurse (2015) and the Stern (2016) reviews of research.
Still, the impact agenda remains as fraught now as it was at its inception following the Warry Report (2006). Then, as now, mandating an impact agenda was seen to represent the very thing that universities might militate against -where knowledge becomes commodity and academic freedoms are eroded (Collini, 2012;Docherty, 2011;Tight, 1985). In fact, at this point, we maintain that impact is now more than a mere agenda: it is an implemented global policy, with the United Kingdom and Australia at the 'vanguard' of these developments (Upton et al., 2014). Since evolving forms of impact are here to stay, at least into the immediate future, we suggest that we should think carefully about the many worries raised by critics, particularly in those earlier days since few studies have considered their empirical accuracy. This includes our main preoccupation: that the impact agenda is often operationalised in ways that are epistemically corrupting. Before we explain the reasons behind this concern, laying out some of the history and rationale behind the impact agenda will be useful.

The emergence of an impact agenda in higher education
As outlined, the UK dual support funding system of research includes a need to articulate, demonstrate and perform both ex post and ex ante impact as a formal part of the 'academic contract' (Watermeyer, 2014: 359). The United Kingdom's quality-related research (QR) funding allocation system and the Research Excellence Framework (REF) officially define impact as 'an effect on, change or benefit to the economy, society, culture, public policy or services, health, the environment or quality of life, beyond academia' (Research England, 2020: 1). In 2009, there was the formalised introduction of 'Pathways to Impact' in Research Council bids (UKRI, 2019). The latter mechanism was recently abandoned by UKRI in an attempt to better 'embed' impact implicitly in overall research funding applications, rather than in a specially dedicated section. Under this new arrangement, impact is expected to be implicitly addressed throughout any research proposal, portending that it is here to stay in UK HE (EPSRC, 2020).
A key theme of the continued focus on impact is a shift to 'competitive accountability' (Smith et al., 2020;Watermeyer, 2019). Researchers are increasingly required to promote and (to some extent) 'market' or 'sell' the value of their endeavour to non-academic groups, with the aim being to assure them of the worth of continuing investment in the academic sector. All this is consistent with the shift to an increasingly market-orientated audit culture in UK Higher Education (Chubb and Watermeyer, 2017;Shore and Wright, 2003). Such tendencies, as described, easily spread to other national contexts: the United Kingdom and Australia, for instance, were engaged in policy borrowing concerning research impact from the start, with a focus on impact statements in funding applications (through the Australian Research Council) in the early days and trials in impact assessment (e.g. the Excellence in Innovation Assessment trial, 2012) and more recently the Excellent in Research Australia assessment. For these reasons, focusing on the British and Australian contexts is useful for comparison and discussion (see also Chubb and Reed, 2018: 299, Figure 1).

Shifting ground: The evolution of an agenda
The evolution of the impact agenda is shaped by all sorts of complex and contingent historical, political, cultural, ideological and institutional events and developments. Not all of these are necessarily problematic and we think it is important for critics not to reject out of hand the general principle that academic work ought to positively impact on wider, non-academic audiences. Nor should critics ignore or dismiss the positive attributes of this agenda, like those involving tangible social improvements and promotion of academic engagement tied into rejuvenated notions of public epistemic responsibilities (Chubb and Reed, 2017;Hill, 2016;Oancea, 2019). At its best, we believe that research produced in higher education institutions from across disciplines can, and often does, serve to enrich our common culture, bolster civic agency and improve the quality of democratic discussions. As a result, we propose that criticisms of the impact agenda must be respectful of its theoretical promise, but cognisant of the risks involved in its practical implementations. We, critics, should also consider how the impact agenda is related to other themes that are familiar in wider critical discourses about the current state of higher education -its 'culture of speed', for instance, or the creeping erosion of scholarly integrity in an arena of marketised competition (see, inter alia, Berg and Seeber, 2016;Chubb and Watermeyer, 2017;Olssen and Peters, 2005;Rhoads and Torres, 2006). All these things are related, but they should not be haphazardly conflated with one another, even if some of them -like the 'culture of speed' -have also been accused of being epistemically corrupting in the same sense at work in this article (Kidd, in press).

Concerns about the impact agenda
It would take a very long article to survey all the different concerns and criticisms directed at the impact agenda and doctoral work and larger scale studies have begun to do so (Chubb, 2017;Cohen, 2021;Grant et al., 2010). We focus on a striking and underexplored criticism, one that shows itself in the commonly employed language of corruption: not in the sense of selling degrees or taking money from Big Tobacco, but in the sense of legally sanctioned pressures eroding or damaging the character or integrity of academics and, sometimes, of the academic enterprise itself. 1 A critical rhetoric of corruption is common in educational discourses and keys into a sense that much that is going wrong in academia causes damage to the character or integrity of academics and students. Collini warns that the philistinism of economically focused academic cultures risks the 'corruption' of scholars' venerable role as 'custodians' of a 'complex intellectual inheritance' (Collini, 2012: 199). At the same time, Martha Nussbaum warns that pedagogical emphasis on rote learning and memorisation drives out the cultivation of humanistic capacities, like empathy and an expansive sensibility, which, if left unchecked, will 'corrupt the mission of humanistic scholarship' (Nussbaum, 2010: 130). For Michael Sandel, a relentless culture of assessment defined in terms of performative goals tends to 'erode, or crowd out, or corrupt' the 'love of reading', learning and education for its own sake (Sandel, 2012: 61). In these cases, being corrupted is a matter of damage done to the motivations, dispositions or character of academics and their students.
A language of corruption is interesting and consistently connected by those employing it to issues of personal moral and intellectual character. A common complaint is that gradual subjection to the pressures of marketisation, philistinism and other social and institutional factors damages or in severe cases destroys personal-level excellences -virtues, autonomy, integrity. This extends to the impact agenda too. Collini, for one, criticises it on the grounds that it erodes academic autonomy: the ability of academics to wisely and judiciously select projects of relevance and worth. If, as he warns, 'politicians have lost the confidence to tell the electorate that universities best perform their distinctive and peculiar role when granted the intellectual autonomy to decide what areas of research will be most fruitful', then academics get stripped of their capacity to exercise those virtues, like discernment and judiciousness, integral to autonomy (Collini, 2011, para 3). An impact mandate will, therefore, start to interfere with our capacity to fully cultivate and properly exercise virtues that are, in Alasdair MacIntyre's sense, 'internal' to our practice as academics.
For Collini, the impact agenda is symptomatic of wider corrupting forces denigrating the 'Haldane Principle' (Haldane, 1918), a well-established principle in science that academics must retain ultimate authority in matters of decision-making with regard to which bodies of knowledge are worth furthering through research and scholarship. As a basic framework and ethos of academic autonomy, it is invoked in arguments against the impact agenda and as a part of appeals for a broader and more adequate understanding of the value of universities. For our purposes, we also want to emphasise that the Haldane Principle, when honoured, creates institutional conditions receptive to the exercise of a variety of virtues, like conscientiousness and judiciousness. If an academic system does not encourage or facilitate the exercise of such virtues, it is understandable why critics naturally reach for a vocabulary of corruption -for one cannot easily cultivate and display the personal qualities that are really integral to proper performance of the activities integral to academic enquiry. This is the point articulated 50 years ago by Michael Oakeshott in his famous lecture on 'The Frustration of Education', where he remarked that to corrupt something is to treat it in ways that tend to deprive it of its character, of its essential, defining, positive features -which, in the case of an individual person, like an academic, are their virtues, understood in the Aristotelian sense of 'excellences of character' (Oakeshott, 1971: 57).
We think that the concept of corruption points to an important style of criticism of the impact agenda that is implicit in critical discussions, even if rarely explicitly articulated. We do not think that one needs to be a virtue theorist to worry about corruption of this sort: everyone engaged in academic work recognises that it could be done in better or worse ways and that one main determining factor will be the personal qualities of those academics who are doing it. A careful and diligent scholar or scientist is better than one who is lazy and sloppy, even if possession of those virtues does not, by itself, guarantee them success in their endeavours (Baehr, 2011: 1). Intellectual character changes over time, though, being shaped by all sorts of factors, including the various pressures, incentives and temptations built into one's social and professional environment. It is at this point that the concept of corruption becomes salient.
We take the term epistemic corruption from recent work by Ian James Kidd, who uses it to describe a form of damage done to people's epistemic character by their subjection to conditions or processes that erode epistemic virtues such as curiosity and thoughtfulness and facilitate the epistemic vices like dogmatism or closedmindedness (Kidd, 2015;Kidd, 2019Kidd, , 2020. 2 A main claim of his work is that many features of modern higher educational systems tend to be epistemically corrupting in the sense that they are prone to damage the epistemic characters of academics and students, a strategy of criticism developed by other contemporary philosophers of education (Battaly, 2013;Forstenzer, 2018). 3 We are sympathetic to this style of critical discourse about higher education and want to develop it further by applying it to the impact agenda. However, to do so, we must amend Kidd's theoretically motivated account by adding empirical details that serve to test its latent empirical claims. In what follows, we offer a retrospective case study involving the impact agenda in its relatively recent incarnation in the United Kingdom and Australia. We use historical empirical data in the form of interview data from academics in those countries describing their experiences of the effects that the impact agenda has on their academic practice (Chubb, 2017;Chubb and Watermeyer, 2017). By doing so, we can test claims about the epistemically corrupting tendencies of the impact agenda and gain a richer understanding of the relevant mechanisms of corruption.

Qualitative methods
Interviews were 30-60 minutes long and the participants were sourced via the research offices at two research intensive institutions. Participants were contacted via email and invited to participate in a study concerning the challenges of the impact agenda in the United Kingdom and Australia and were specifically asked about perceptions on its relationship with freedom, value and epistemic responsibility. Participants comprised mid-senior career academics with experience in grant writing, from across the range of disciplines broadly representative of the arts and humanities, social sciences, physical science, maths and engineering and the life and earth sciences. Participants were given the opportunity to decline the invitation, provided an information sheet about the study and an informed consent sheet. Almost all potential participants contacted could take part and those who did not cited time as the main barrier for participation. Interview transcripts were analysed using discourse analysis, inductively coded at keywords related to the prominent vices/virtues referenced in the data. Informed consent forms were signed by participants, which promised anonymity and confidentiality of the data provided. This article, through the lens of this empirical data, clustered vices by approaching the data inductively and recognising and forming patterns through an inductive analysis. This was cross-checked and discussed across the research team, who kept a reflexive account of how these were formed. The data included in this article will be identified by the discipline and the national context in brackets, from the study by Chubb (2017).
By using these data and the concept of epistemic corruption, we offer an analysis of the effects of the impact agenda on the epistemic conduct and character of academics that is both conceptually sophisticated and empirically informed. Our claim is that the evolving forms taken by the impact agenda have created at least three core structural conditions that facilitate the development and exercise of two specific clusters of epistemic vices. A main value of the interview data is that they reveal the awareness of interviewees of what we are calling epistemically corrupting structures and their complex responses to them.

Limitations of the research
We draw our findings from a sample of 50 academics from the United Kingdom and Australia (2012)(2013)(2014). These data represent the common attitudes and feelings of academics working in higher education at the time of the emergence of the impact agenda in both settings. In the case of Australia, talk of an impact agenda was especially naissantas such talk of impact was similar to talking about something in the abstract, despite its appearance in some funding calls. The 'newness' of impact, at that time, could be seen to be reflected in some of the strong and emotional responses found in the data. We suggest that though these data are historical-despite developments and changes in and to the impact agenda-they form a rich and authentic baseline from which to explore experiences of and responses to epistemic corruption. Some of these experiences will evolve in line with new iterations of the impact agenda, though previous experiences will set up expectations that shape future ones. The content of some of the responses may be specific to particular iterations too, although the range of general kinds of responses is likely to be fairly stable.

Epistemic vice and corruption
Contemporary educational theory and practice is seeing renewed interest in the venerable idea that cultivation of moral and intellectual character by educating for the virtues ought to be the primary aim of education. 4 Such interest encompasses many different groups of virtues, comes from many quarters, from policy-experts and education researchers to theorists of education. It has also elicited different sorts of responses, ranging from warm welcomes to staunch criticism -a situation complicated by long-standing debates internal to philosophical virtue theory. An unfortunate feature of these debates, however, is a failure to attend carefully to the possibility that educational experiences and systems can promote vices as well as virtues (Battaly, 2013;Cooper, 2008;Forstenzer, 2018). Whether or not one thinks that education should promote a certain set of virtues, one can still worry about it feeding an array of vices, including epistemic vices like arrogance, closed-mindedness, dogmatism, epistemic inflexibility, un-reflectiveness and other objectionable attitudes, character traits and ways of thinking (Cassam, 2019).
Epistemic vices develop under the influence of all sorts of conditions, which will include structural features of our social environments, in ways that the concept of epistemic corruption aims to capture. For this reason, epistemic corruption should be understood as a dynamic process, unfolding over time as individuals interact with their social environments. When analysing epistemic corruption, one should attend to four things (Kidd, 2019: §2): 1. The epistemic vice(s); 2. The corruptor (the source or agent of corruptor); 3. The corruptees (the person or thing whose character is being corrupted); 4. The corrupting conditions or processes.
Before we proceed to our case study -which covers (2) to (4) -we need to say more about the nature of epistemic vices. Following Heather Battaly (2014), we conceive of epistemic vices as traits of epistemic character that make us bad thinkers. Vices can do this in one or both of two ways. Effects-vices are traits that tend to cause a preponderance of bad epistemic effects: inattentiveness, for instance, means one will systematically fail to detect epistemically relevant features or a situation, in ways that impair one's ability to acquire knowledge and understanding. Motives-vices are traits that are vicious because they express or manifest bad epistemic motives or values, such as an indifference to truth, or a desire to interfere with the epistemic agency of others, a sort of epistemic malevolence (Baehr, 2011;Cassam, 2019).
Since epistemic vices are integral to appraisals and performance of epistemic agency, an important part of vice epistemology involved attending to issues of aetiology and responsibility -that is, to issues of the origins or sources of people's epistemically vicious characters and the extent, if any, to which we can hold them responsible and blameworthy. The typical automatic movement from attribution of vices to accusation and blame is unwise, since the aetiology of epistemic vice is often psychologically and sociologically complex. Compare two dogmatic people: the first grows up with dogmatically aggressive peers in a deeply repressive society of underfunded schools, a culture of sneering at science and learning and a monopolistic media environment; the second grows up with open-minded peers, good teachers and a pluralistic social environment with a genuinely free press, rich sources of information and a culture of public debate. All things being equal, intuitively, we would consider the former person not to be responsible for either becoming or remaining dogmatic, given the conditions of their epistemic socialisation, whereas the second person, afforded every opportunity to become openminded and critically lucid, would be considered to be much more responsible for their dogmatism (Battaly, 2016b;Kidd, 2016). When analysing agential epistemic vices, a wiser strategy involves exploring the aetiology of their epistemic character, attending to the array of conditions, interactions and structures, which can feed or starve the development of virtues and vices. When epistemic vices are the products of our sustained subjection to epistemically corrupting systems, then we can attribute vices without necessarily wanting to also add blame into the mix (Battaly, 2019).
We can now say that epistemic corruption occurs when one or more epistemic vices are acquired, intensified or entrenched as a result of the agent operating within or being subjected to environments that facilitate their development and exercise. Individual agents have more or less malleable epistemic dispositions -that is, dispositions to assess, criticise, evaluate, reflect, theorise and understand in effective ways. The dispositions can become stronger and more stable, until they acquire the status of virtues or vices, depending on whether they are positive or negative. Our epistemic characters may, therefore, consist only partly of virtues and vices, construed in the traditional Aristotelian sense of fully formed and cross-situationally stable virtues and vices -an empirical claim defended by Christian Miller (2017).
Given the plasticity of our epistemic characters, a natural concern of practically minded character epistemologists is to find ways of organising our communities and environments to try to ensure they are maximally conducive to the cultivation of virtues and the nullification of vices. A newly emerging literature has begun to explore the range of options from individual practices of epistemic self-reform to more ambitious proposals for transformation of the wider social-epistemic environment (Battaly, 2016a;Medina, 2012;Wilson and Miller, 2018).
Using the terminology offered by Kidd, we claim that a practice, community or environment that is conducive to the cultivation and exercise of virtues is edifying -in the sense of building or strengthening a structure (edificio, in Romance languages, means 'building', in its nominal and dynamic senses). A person is edified when their character is built up or strengthened through the cultivation of their virtues, the excellences or strengths of character. By contrast, an agent becomes corrupted when their character is damaged or weakened, hence the sense of vices as 'defects'. Such corruption might involve the acquisition of new vices, never previously a feature of one's character, or the entrenchment or the intensification of one's vices -perhaps one's incipient inflexibility slowly becomes a relentless dogmatism.
Epistemically corrupting environments are ones whose norms, practices and structures tend to facilitate the acquisition, entrenchment or exacerbation of epistemic vices and failings by agents embedded within them. The crucial diagnostic task, for those with ameliorative concerns, is to identify the relevant corrupting conditions in that environment, so that they can be removed or weakened. Since our epistemic environments are hugely diverse in their scale, structure, goals and so on, we cannot provide anything like a comprehensive listing of the corrupting conditions. An additional set of complications is that (1) there may be generic corrupting conditions that feed a variety of vices as well as those specific to certain vices, (2) individual agents differ in their susceptibilities to epistemic corruption due to a variety of idiosyncratic and situational factors, (3) conditions can be stronger or weaker and therefore sometimes difficult to detect and (4) all but the most highly epistemically corrupting environments also contain at least some edifying conditions as well. All of this suggests that we ought to conceive epistemic corruption as a diachronic and dynamic process, rather than a static inevitability within a given environment -a process agents can repond to in all sorts of ways (Monypenny, 2021;Tessman, 2005).
We suggest, however, that there are certain generally corrupting conditions, ones that act such as to facilitate the acquisition, entrenchment or exacerbation of many epistemic vices and failings. 'Facilitate', as it pertains to epistemic corruption, can mean 'make possible', 'enable', 'encourage', 'legitimate', 'reward' or 'conceal', among other things. Consider, for instance, these generically corrupting conditions, each of which, in different ways, facilitates epistemic viciousness: • • The absence of systems to monitor and correct for epistemically vicious actions and patterns of behaviour: Since an agent can perform vicious acts without being considered at that time as having a vice, it is crucial that as many of those acts as possible should be reliably detected and effectively counteracted. Monitoring and corrective systems can, therefore, act as a guard against corruption, such that their absence tends to be a corrupting condition (for instance, studies of business institutions show that honesty is encouraged by a combination of regular audits and positive peer-established social norms (Cardinaels and Jia, 2016). • • The absence or derogation of exemplars of virtue: Some people are inspirational and are practical models for personal epistemic excellence act as a spur to the development of virtue in others. Absence of exemplars is corrupting since, without them, the community lacks one of the most effective means for identifying, understanding and cultivating virtues. Derogation of exemplars -through acts of mockery, sneering, ridicule -corrupts by undermining our ability or willingness to recognise exemplars or even to admit their existence. Without them, a drift into vice becomes more likely, since the counterbalance of virtuous exemplars is lacking (Croce and Vaccarezza, 2017; Gray and Jordan, 2012). • • Increasing the exercise costs for virtue: One can corrupt someone by making it harder for them to exercise their virtues which, over time, weakens the rootedness of virtue in their psychology and contributes to its dishabituation. Being truthful, for instance, often incurs real social costs since it can involve embarrassing or 'betraying' people by 'telling uncomfortable truths' and reaping the wrath of those who consider themselves aggrieved. By reacting negatively to the exercise of the virtue of truthfulness, people generate a corrupting environment, since their judgements and actions increase the price to be paid for truthfulness -a price not all of us can always afford to pay. • • The rebranding of vices as virtues: Since many agents would be alarmed at the idea of their being or becoming vicious, one can circumvent their resistance by rebranding their vices as virtues. Once a vice is perceived as a virtue, many agents will cease practical measures aimed at its removal or nullification. Dogmatism becomes 'firmness', arrogance becomes 'confidence', rudeness becomes 'directness'. As a strategy of corruption, rebranding is especially insidious, because it reinforces our natural desire to ignore or deny our vices. • • The establishment of conditions that necessitate the exercise of vice: Our vices are always expressed in certain forms of moral, social or epistemic practice. Whether by accident or design, certain practices can develop such that performing them requires one to exercise certain vices. In some circumstances, the preservation of an advantageous but unfair system, requires exercising the vice of lying or occluding the truth from others. Think, for example, of Major League Baseball in the 1990s and early 2000s, the culture of doping had become so pervasive and the popularity of the sport so high as to make speaking out about the widespread use of steroids very costly for individual players. Although some did eventually speak out -notoriously Canseco (2005Canseco ( , 2008) -most players kept quiet precisely because their careers depended on it. • • The valorisation of vicious conduct and exemplars: Assuming that our conceptions of good and bad epistemic character and conduct are informed, to some significant degree, by the examples available from within our communities, a person can become corrupted if those held up as exemplars are actually vicious. By emulating those people, one becomes vicious and, therefore, corrupted (Zagzebski, 2017: ch. 2, sects 3-5). Consider situations where those 'big names' in one's department or discipline -held up as heroes or as professional models -are aggressively adversarial, egoistic and prone to dominative behaviours, like deriding their 'rivals' or sneering at 'lesser' work (Moulton, 1980;Rooney, 2010). Unless their status as admirable exemplars is challenged, they will almost certainly be emulated and this leads to the corruption of their admirers.
These are just some of the generic processes of corruption, applicable to most if not all the vices, using examples from various institutional and professional domains (business, philosophy, sport and so on). All of them facilitate the development and exercise of vices, although in different ways -the valorisation of exemplars of vice mainly encourages viciousness, for instance, whereas rebranding arguably conceals the presence of actual or emerging vices.
With these remarks on epistemic vices and corruption in place, we will now focus on the claim that the impact agenda is creating epistemically corrupting conditions in the higher education systems of Australia and the United Kingdom. Our claim is that the sociological data about the effects of the impact agenda on the conduct and character of Australian and British academics clearly indicates that conditions are being created that encourage the development of two main sets of epistemic vices, the appetitive and the alethic.

Vice-clusters
There are many different epistemic vices, organisable in different ways, and an epistemically corrupting system can promote some or all of them. Based on the interview data, we think it is clear that two main clusters of epistemic vices were most consistently reported as being in some way encouraged, promoted or otherwise facilitated by the institutionalised forms of impact agenda in Australia and Britain at that time. A vicecluster, here, is a specific set of epistemic vices that share some characteristic in common, and the theoretical practice of identifying these clusters can be called epistemic vice taxonomy (Kidd, 2019: §2B). The vice taxonomies are best conceived as pragmatic artefacts that help us think efficiently about such things as patterns in the specific vices of the mind within a specific community, rather than as delineations of any objective ordering.
The two vice-clusters most commonly reported in the interviews were what we shall label 'appetitive vices' and 'alethic vices'. The appetitive vices are those that involve excesses or deficiencies of epistemic appetites, that is, our appetites for epistemic goods, like beliefs, knowledge and understanding. Incuriosity is a vice marking a deficient epistemic appetite, a lack of appropriate appetites for relevant, salient epistemic goods and, therefore, a vice and another is epistemic insensibility (Battaly, 2010;Battaly, 2013;Bloomfield, 2019). By contrast, the most famous appetitive epistemic virtue is curiosity (Watson, 2018). Alethic vices involve failures to respond appropriately to truth such as dishonesty and untruthfulness and the alethic virtues include honesty, sincerity and truthfulness (Williams, 2002). Expressions of the alethic vices include lying, deceit, dishonesty, 'bending the truth', failing to properly contextualise claims and what Frankfurt (2005) memorably described as 'bullshitting'.
Our interview data contain many references to these two clusters of epistemic vices, alongside a more diffuse set of other failings. In what follows, we focus on alethic vices, not least since truth is one of the most fundamental epistemic ideals of modernity and a central value of academic enquiry (Williams, 2002: ch.1). Our claim is that the impact agenda has established and is currently entrenching norms, structures and cultures that are facilitating the acquisition and exercise of alethic vices. The interviewees very helpfully identified and explained the relevant sorts of corrupting conditions. Specifically, the data indicate three main sorts of general structural conditions that facilitate the exercise and development of alethic vices: incentive structures, structural constraints and increased exercise costs for virtuous truthfulness.

Incentive structures
To start with, the interviewees consistently argued that the impact agenda was creating and entrenching incentive structures that encourage the giving of false, inflated or exaggerated estimates of the actual or likely impact and benefits of research. By providing those sorts of professional incentives, those structures corrupted alethic vices like inaccuracy, insincerity and hyperbole.
Consider one Australian academic, working in the Arts and Humanities, who describes how the impact agenda militates against truthfulness about the methods and outcomes of research in two ways: I think that one would be hard pressed to write a successful grant application that's fully truthful, even if it's for no other reason than to comment on one's own research environment, but also when one has to go out of one's way to make up a story about the kinds of significance and social impact that grants have to have. (Arts, Australia; Chubb, 2017).
How do grant application systems militate against properly truthful description of one's research environment? First, when asked privately, many academics willingly report the absence from their institutions of vital research resources -from access to essential journals to research assistance support to funding for research trips. But truthful reporting of the inadequacies of one's research environment would weaken an application, since one is admitting that one would be attempting research under suboptimal conditions. Second, statements of the actual or likely impact of research often require exaggeration, whether of the scale or significance or certainty of 'social impact'. If some research has a limited scale or modest significance, then it may be regarded as less impressive, creating an incentive to inflate. But such inflation is often a failure of truthfulness, especially of accuracy and sincerity.
Another way that structures can entrench alethic vices is by introducing incentives for academics to consciously succumb to the temptation to strategically abandon the virtue of truthfulness for the sake of participation in certain academic structures. Truthfulness is, after all, one of the virtues that acts to help us resist the temptations to distort and exaggerate our claims, consistent with the virtuous desire for what Williams calls a 'sense of reality' (Williams, 2002: ch. 6). For example, one interviewee states, The problem is, it's ex ante! It's a guess! And people make it up -they make it sound huge! (Life and Natural Science, Australia; Chubb, 2017) Another interviewee, a UK-based researcher in the Physical Sciences and Engineering says, 'there is no possible way they could make [claims about future impact] concrete, because there needs to be a lot of additional future research before one can have any concrete idea about what the impact will be' (Physical Science, United Kingdom). When making claims about future outcomes, truthfulness would typically require the exercise of epistemic virtues like cautiousness, tentativeness and reticence -a cluster of dispositions that can correct for our tendencies to overstatement, exaggeration or pretension to a greater degree of confidence than one could epistemically warrant. Truthfulness, in these cases, requires that one makes carefully qualified, conditional claims of the sort Michel de Montaigne (1991Montaigne ( : 1165 praised when stating his preference for words and phrases that 'soften and moderate' the typical 'rashness' of speech -'probably', 'possibly', 'as far as I know', 'based on available information' and so on. The worry is that incentive structures are discouraging the giving of cautious, tentative accounts of the social impact of research. If so, they corrupt by encouraging the alethic vices.

Structural constraints
The impact agenda will also promote the exercise of vices of truth when it establishes or strengthens structural constraints that tend to discourage or prevent giving properly truthful descriptions of one's research practices. The untruthfulness in these cases is selfdirected, insofar as one is being inaccurate, distorting or dishonest about one's own research practices. A physical scientist, based in the United Kingdom, describes the selfdescriptions of research practices as 'disingenuous', in the sense that they involve providing accounts of those practices that one knows to be inconsistent with the actual realities of those practices.
[T]he trouble is, it's disingenuous. No scientist really begins the true process of scientific discovery with the belief it is going to follow this very smooth path to impact, because he or she knows full well that just doesn't occur, and so there's a real problem with the impact agendaand that it's not true, it's wrong, it flies in the face of scientific practice. (Physical Science, United Kingdom;Chubb, 2017) In this case, the interviewee is pointing to the fact that no scientist could genuinely endorse the smooth, tidy accounts of the process of scientific research required by the impact system -ones that present an untenably neat account of research as a process that inevitably delivers guaranteed outcomes. Although an impression of inevitable progression from research to outputs to impact makes an impact pathway look attractive to evaluators, no-one with both an informed understanding of the realities of scientific practice -in the laboratory, at the workbench or in the field -and a commitment to truthfulness would endorse it. Giving distorted accounts of the process of scientific research is an old problem, which predates the impact agenda. 5 An interviewee notes: '[Y]ou can't get a proposal through if it tells untruths, [viz.] scientific fallacies, but you can get a proposal through if it tells impact fallacies' (United Kingdom, Physical Science). Moreover, an Australian social sciences researcher offers an amusing example of the ways in which the format of funding and impact documentation can require untruthfulness to the point of absurdity: [A]ctually it literally said 'write down what you're going to discover'. Well, if I can write down what I'm going to discover, I've discovered it. It was a dumb question, but it was put in there by someone who doesn't know what research is. I don't know what you're supposed to saysomething like, 'I'm Columbus, I'm going to discover the West Indies. (Social Science, Australia;Chubb, 2017) Two salient points are worth noting here, the first is that the research document in question was either seen to be written by someone with no informed understanding of the realities of research practice or written by someone willing to tolerate the giving of false information with regard to the ultimate object of any one research project. Either way, the document is seen to invite at best a false account of scientific research and at worst outright exaggeration for the sake of gamesmanship.

Increased exercise costs
An epistemically corrupting system both promotes the development and exercise of various epistemic vices and makes it more difficult for people to cultivate and exercise the epistemic virtues. One way to make virtues more difficult is to increase their exercise costs. Exercising truthfulness, for instance, can be made harder by removing or weakening social norms of truthfulness or by introducing formal and informal penalties or punishments for those who are truthful, or by changing incentive structures in ways that militate against acts of truthfulness. The interview data indicate that these sorts of antitruthfulness conditions are part of the institutionalisation of the impact agenda: If you can find me a single academic who hasn't had to bullshit or bluff or lie or embellish in order to get grants, then I will find you an academic who's in trouble with his HoD (Head of Department). If you don't play the game, you won't do well by your university. So, anyone that's so ethical that they won't bend the rules in order to play the game is going to be in trouble, which is deplorable. (Arts and Humanities, Australia; Chubb, 2017) This academic testifies to their experience of social and professional pressures that make it more difficult to exercise truthfulness -the disapproval of their Head of Department, potential threats to promotion prospects, removal of the goodwill of their superiors and so on. Another interviewee reports similar pressures, this time concerning inducements to disguise the true nature of one's work to make it fit into the impact agenda: When somebody says, 'I want to do this', the kind of advice that they have to seek is how to disguise their actual agenda in order to create the agenda set up by funding agencies. (Arts and Humanities, Australia; Chubb, 2017) A certain degree of pragmatism about the presentation of projects is not, in itself, a failure of truthfulness. But this is something different, since it involves disguising in substantial ways the essential character of one's research projects and also feigning certain motivations. The inducements to be untruthful about the quality of one's institutional environment also recurs, this time in the context of structural pressures to secure grant income: [Y]ou're also asked to describe the research environment at your own institution. Now, nobody is going to draw attention to the fact, to things that they don't like about the research environment at their own institution, so you're made to lie in these kinds of ways. (Arts and Humanities, Australia; Chubb, 2017) We have considered just three structural conditions that, according to the interviewees, are experienced as epistemically corrupting. The institutional enactments of the impact agenda at that time in those countries were creating structures and constraints that militate against the exercise of alethic virtues, like truthfulness and, in turn, encouraged and rewarded the exercise of alethic vices, like dishonesty and disingenuousness. The identification of specific structural corruptors advances our understanding of the social mechanisms of epistemic corruption in a way that empirically confirms Kidd's initial analysis. Moreover, we can gain other things from the interview data -for instance, the different ways that those who are alert to those corrupting influences can try to resist them, whether through attempting to reform the structures or by developing coping responses at the personal level. Those would all be useful directions for future research on the impact agenda in higher education, which we think is very much needed, since further research might show that there are differences across national context, career stage and, of course, discipline. Furthermore, there have been recent developments in the implementation of the impact agenda in the United Kingdom and Australia and these may well have led to subtle changes in and a sense of accommodation to an agenda now very much into its teenage years as opposed to its fragile infancy (which is the period covered by empirical evidence in this article). Ultimately, we hope that this article shows there can be very useful roles for empirically informed virtue and vice talk in education and especially for the concept of epistemic corruption in critical educational discourses. After all, the main practical motivation for 'the project of vice reduction' is that, as recent experiences have intimated, a world of widespread epistemically vicious conduct is 'too ghastly to contemplate' (Cassam, 2019: 186, 187).

Impact without corruption?
The data suggest that many academics experienced the earlier incarnations of the impact agenda as epistemically corrupting, at least for vices in the alethic and appetitive clusters. It is clear that some of the corrupting conditions are still in place, but also likely that others have been removed or weakened and that other new ones may have emerged -for instance, game-playing and competition (Weinstein et al., 2021). If so, then analyses of epistemic corruption must be active and continuous, not conducted periodically or ex post facto. Some kinds of corrupting conditions can be anticipated in advance, even if others only show themselves once they are in place.
As with many institutional phenomena, what is really needed is constant scrutiny and careful vigilance, but not just to try to minimise the range of corrupting features built into our academic cultures. Some deeper issues are at stake. If knowledge production is really to have a practical benefit to wider society, then we must think seriously about the responsibility that the academic community has to wider social and political culture. We already know that there are complicated moral and political issues about the ways we aim to impact society, not to mention genuine uncertainties about the sorts of impact academic research can have over long periods of time. Many academic developments take a great deal of time to come to fruition and there is no intrinsic connection between research that aims to be of immediate use to practical features of human life and the depth of influence any one body of research has actually ended up having on our communal lives. 'Blue skies research' often yields goods that lead to profound structural changes across many spheres of human activity. But this will be jeopardised by epistemically corrupting systems.
We can also be concerned about epistemic corruption within academia in the context of even larger issues. The challenge of envisioning an impact agenda that is less prone to the kind of corrupting tendencies we have described is more than a mere policy design problem. It invites fundamental questions about, inter alia, the nature of knowledge, the status of truth as an ideal, our confidence in the reliability of the accountability mechanisms essential to academic systems and the moral commitments of academics and higher education institutions. It will never be enough to celebrate the pursuit of truth as an ideal unless we also populate our communities with inquirers who are truthseekers -that is, with people who are disposed to care about and work hard to acquire rich epistemic goods. To do that, we need to ask big questions about how to arrange academic, social and political communities to promote the virtues of truth-telling and truth-seeking and all the other epistemic virtues. And we must also think in similarly energetic ways about those epistemic vices whose baleful effects on life are documented by contemporary vice epistemologists (Kidd et al., 2020).

Concluding thoughts and further research
To thoroughly investigate the effects of epistemic corruption, we must deal with practical, empirical and theoretical concerns. This requires academia to forego antiquated dichotomies between purely intellectual and narrowly practical work and an unhelpful contrast between philosophical and empirical considerations. Impact has implications that cut across each of these domains and so must be dealt with in an accordingly broadminded way. We propose to take seriously concerns about the epistemically corrupting potential of impact and maintain that it should be a central part of further research. Indeed, the application of epistemic corruption as a framework on existing datasets, which shed light on later iterations of the impact agenda might be a useful starting point. If changing forms of the research impact agenda retain their epistemically corrupting tendencies, then there are downstream implications for the place of academic knowledge as it relates to democratic deliberation, civic and social life and human inquiry. If Ralph Waldo Emerson is right that 'character is higher than intellect', then corrupting systems that lead to spoilage of epistemic character entails wastage of our communal collective intellectual resources. Further policies designed to address the impact agenda ought to consider to what extent they are epistemically corrupting. To do this, we suggest that a multi-disciplinary conversation involving voices from across different communities and disciplines combining theory with empirics is a promising way forward.