Formal conditions of hostile design

What, then, are the formal conditions of hostile design? To what does it pertain? The broadest views here talk simply of space (de Fine Licht Reference de Fine Licht2023), while many add that we are dealing with public or urban phenomena (Jensen Reference Jensen2023b, 1; E Nielsen Reference Nielsen2021, 16; Izzo Reference Izzo2022, 527). The narrower definitions focus on objects in public spaces (Rosenberger Reference Rosenberger2020b; Reference Rosenberger2023). We will take hostile design to pertain to the configuration of public space. ‘Configuration’ refers to physical arrangements made by humans, such as surfaces, structures, signs, furniture, and electronic equipment. The ‘public space’ part of the formulation ensures that we discuss areas that are (or should be) open to the community.Footnote ⁶ Our formal condition excludes things that do not count as designs, as that concept is used in discussions of hostile design, such as weather, terrain, plants, or animals due to nature rather than human intervention; social or legal norms themselves, rather than their physical manifestations; and features of private homes.Footnote ⁷ Some of these things may sometimes be unwelcome or, in some sense, hostile, but they are not the kinds of things that can count as hostile design.

Our formal condition is broad enough to include the designed absence of objects by (1) removing them, as when a bench is uninstalled or a public restroom closed, or (2) not providing them in the first place, as when a new public space has no benches or restrooms. This hostile design by removal or absence is broadly recognized in the literature (Rosenberger 2017, 32–33; Reference Rosenberger2020b, 887; Reference Rosenberger2023, 62; Dakin Reference Dakin2021, 44; CH Nielsen Reference Nielsen2024b), and we take it to be an important reason to prefer space over object in the definition. We would not want to rule out that the Metropolitan Transportation Authority (MTA) in New York described a hostile design when they tweeted that benches were removed from stations to prevent homeless people from sleeping on them (Perrett Reference Perrett2021). The removal of a bench to prevent its use by the homeless is intuitively just as much a hostile design as the introduction of anti-homeless benches. To be sure, the absence of public amenities can be due to genuine low demand, and such cases are still configurations of public space, thereby satisfying our undemanding formal condition. It is their non-hostile character, as defined by the substantive condition we later discuss, that means they are not hostile designs.

The formal condition of hostile design distinguishes it from other ways vulnerable groups are targeted. Of particular relevance for our topic are anti-homeless laws, and related prohibitions on panhandling, camping, trespass, and storing property in public. Such laws and prohibitions are quite common (Kettel Reference Kettel2024). These laws do not count as hostile design in our definition, which rather concerns the configuration of public spaces. Anti-homeless laws do, nevertheless, interact in important ways with anti-homeless hostile design (Rosenberger Reference Rosenberger2017, 28; Izzo Reference Izzo2022; Jensen Reference Jensen2020, 328). An anti-camping prohibition is not itself a hostile design, nor is a police sweep enforcing the prohibition. However, other forms of enforcement of the same prohibition, such as anti-camping signs, anti-homeless spikes, sprinklers, or CCTV, may amount to hostile design.Footnote ⁸ The difference is simply that these measures involve changes to the design of the public space.

Clearly, this is a descriptive rather than normative distinction, and much of the argument we later give against anti-homeless hostile design may apply, mutatis mutandis, to other ways of enforcing anti-homeless laws, as well as to the laws themselves. But it is worth emphasizing that anti-homeless hostile design is not reducible to, nor merely a means of enforcing, anti-homeless law. Much anti-homeless hostile design is not a law enforcement measure at all but rather reflects the bare preferences of public officials and private businesses regarding the use of public space. The interest of our focus on hostile design is, then, twofold, considering both hostile design to enforce anti-homeless laws, and hostile design that does not have that legal motivation. It might be added that anti-homeless hostile design may in part be an important topic of discussion because anti-homeless laws have come under increased pressure. If activists succeed in, for example, arguing that bans on panhandling impede free speech, or that broad laws banning activities carried out mostly or exclusively by homeless people are equivalent to banning ‘a status’, which the 8th Amendment of the US Constitution outlaws (Kettel Reference Kettel2024, 121), we can expect those who pursue an anti-homeless agenda to look for different and more subtle means to achieve their ends. The normative status of anti-homeless hostile design then becomes even more important.

Substantive conditions of hostile design

Our second key definitional topic concerns substantive conditions of hostile design. While it is natural to think of hostile design as consisting of specific objects, such as benches or spikes, on reflection it is clear that physical objects, or even the spaces they occupy, are not sufficient for hostile design. Some further condition is required.

The standard view holds that an intent behind the design is also needed. The relevant intentions are those of the decision makers, such as politicians, planners, architects, and owners and managers of businesses and institutions adjoining public spaces. To illustrate, tiny benches might be harmless when they are provided in a schoolyard while amounting to hostile design when they are placed in a park as an anti-homeless measure because the intent is very different in the two cases.

What kind of intent is required for hostile design? Some substantive definitions foreground hostile design’s intended changes to behaviour. These may identify hostile design as deterring people from using public space as their private space (E. Nielsen Reference Nielsen2021, 16), ‘excluding activities and social groups’ (Jensen Reference Jensen2023b, 1), or ‘targeting social control and behavior’ (Izzo Reference Izzo2022, 527). De Fine Licht similarly suggests that hostile design is when some actor A prevents (or tries to prevent) a bad outcome or realizes (or tries to realize) a good outcome by changing a group or a person’s behaviour, with a threat or a realization of harm (de Fine Licht Reference de Fine Licht2023).Footnote ⁹ But some crucial cases seem to be excluded by the various behaviour-based definitions. For instance, closing off a trashcan to prevent food or bottle picking does not seem to be an attempt at deterring the use of a public space as a private space. Moreover, a general problem with definitions focusing on behavioural change is that they rule out a perhaps rare but concerning kind of hostile design, which simply targets the homeless without concern for whether their behaviour changes.

This leads us to a different branch of intent-based definitions, which identifies hostile design with hostility towards the targeted group. One version focuses on an intention to create discomfort and interfere with the use of public space (Savić and Savičić Reference Savić and Savičić2014, 977). Another requires explicit or implicit ill will toward someone, trying to harm them, or not caring if they are harmed (de Fine Licht Reference de Fine Licht2020, 5).Footnote ¹⁰ Considered in the abstract, it may seem obvious that hostile design requires hostile intent. This account also offers a pleasingly tight fit with certain core examples of hostile design, such as anti-homeless spikes. Yet it excludes other cases that are widely considered hostile design. The impact of CCTV on the homeless is not necessarily motivated by a desire to create discomfort or interfere with the use of public space, nor ill will. Even anti-homeless benches, the very paradigm of hostile design, may sometimes be used without a deliberate intent to discomfort or harm. A well-publicized example is a bench with anti-homeless armrests that featured prominently in the launch of the She-Hulk: Attorney at Law television show in Los Angeles (Gale Reference Gale2022). It is improbable that Disney intended to display hostility towards a vulnerable group, undermining their own premiere, and much more likely that they simply did not appreciate the implications of mid-bench armrests. Even so, we submit that there is a case for saying that the bench, and ones installed in similar ignorance but away from the glare of cameras, amounts to hostile design, and should be encompassed by our definition.

The formal conditions alone are not enough for hostile design, but the substantive conditions proposed in the literature on hostile design may exclude some important cases. The solution is to propose less demanding substantive conditions. We propose one simple, relatively easily satisfied condition: that the design disadvantages members of the group in question (here, the homeless) compared to another group. ‘Disadvantage’ is here to be understood in a very broad sense, including disadvantages in particular goods, opportunities, or rights like money, healthcare, education, employment, or political representation; disadvantages in metrics of distribution such as welfare, resources, and capabilities; and even non-distributive, relational disadvantages of respect, dignity, authority, status, or standing.Footnote ¹¹ This condition is substantive in that the homeless would actually have to be made worse off, along some dimension, by a configuration of public space in order for it to count as a hostile design. But it is also wide, not presupposing intent nor any particular philosophical account of disadvantage. We have, then, arrived at the following definition:

This definition may seem too wide or easily satisfied. Does it not need to be more demanding if our treatment of hostile design is to have a critical bite? Is not much of the objection to hostile design precisely the bad intention behind it?

Such concerns suggest combining definitional issues with normative appraisal as a single stage. The intent-based views surveyed earlier lean towards this approach: identification of hostile design with ill will or intent to cause discomfort bestows a certain normative status on it. We rather propose to treat definition and appraisal as two stages. Specifically, we give relatively undemanding conditions for hostile design, capturing all cases we might expect to capture, then normatively assess them from the perspective of philosophical accounts of discrimination. This two-stage approach has three advantages.

First, and as we have seen, more demanding, one-stage definitions are prone to exclude some cases that intuitively amount to hostile design, forestalling full consideration of them and, potentially, practical resistance to them.

Second, our approach is flexible. Those who give special importance to bad intentions will find their concerns accommodated in the concept of direct discrimination as well as in certain disrespectful accounts of wrongful discrimination. Those who consider alternative aspects of hostile design to be fundamental, such as the harm it inflicts on the homeless, can be accommodated with other accounts of wrongful discrimination.

Finally, one-stage definitions are likely to be dominated by moral intuitions or ‘considered judgements’ about the particular cases of putative hostile design under consideration. According to the method of reflective equilibrium that is dominant in analytic moral and political philosophy, principles are better justified where such case-specific judgements are viewed in the light of theoretical considerations, allowing for mutual adjustment of judgements and principles (Rawls Reference Rawls1971, sec. 9; Knight Reference Knight2017b; Reference Knight2023). Theories of discrimination are a rich source of such considerations, guiding us away from overreliance on pretheoretical case-specific judgements. These theories are, furthermore, independent of judgements about specific cases of hostile design, a key desideratum within reflective equilibrium (Daniels Reference Daniels1996, 22–24). To see the importance of this, consider the situation where a theory (or definition) is not independent of case-specific judgements about hostile design, being motivated by these exact judgements. Here the theory offers no support for the judgements that are not already there in the judgements – the theory is too dependent on the judgements to offer non-circular justification. By contrast, theories of discrimination are certainly not motivated by hostile design, which has not featured in their development, nor are they an important basis for existing judgements about hostile design. They offer a genuinely independent basis for moral assessment of hostile design, in line with the method of reflective equilibrium.

A socially salient group?

As both direct and indirect discrimination focus on social groups, or ‘socially salient groups’, we should clarify whether the homeless classify as such. Lippert-Rasmussen states that ‘[a] group is socially salient if perceived membership of it is important to the structure of social interactions across a wide range of social contexts’ (Lippert-Rasmussen Reference Lippert-Rasmussen2013, 30). The social salience of a group is a matter of degree, and group membership may be important in different ways. It may ‘be anything from slightly important to wholly dominant in the structure of social interactions, and it may be important in more or fewer social contexts’ (Lippert-Rasmussen Reference Lippert-Rasmussen2013, 31).

Is perceived membership of the group of the homeless important across a range of social contexts? We argue that it is. The lack of a permanent place to live restricts job opportunities, makes it difficult to establish and maintain social relations, and creates social stigma (Moran and Atherton Reference Moran and Atherton2018, 12; Grohmann Reference Grohmann2020, 222; Rea Reference Rea2023). It makes it more difficult to receive public benefits and can effectively mean that one loses the opportunity to vote. The drastic effects of perceiving someone to be homeless are, furthermore, shown by empirical research. Fiske has demonstrated that the homeless are considered less competent and trustworthy than almost any group in society (Fiske 2013; 2012; Reference Fiske2018). In fact, the part of our brain that lights up when we meet other people does not light up when confronted with a homeless person (Fiske Reference Fiske2012, 20–21; Harris and Fiske Reference Harris, Lasana and Fiske2009). In short, it is clear that being perceived as homeless is socially salient. It profoundly affects how others view a person, and thus matters to many important interactions.

As mentioned in the introduction, homelessness is a broad notion, which includes a wide range of life situations. That, of course, complicates the argument just made, as perceptions of couch surfers might be different from perceptions of the street homeless. But taking this into account does not greatly alter the picture. One approach says that the street homeless are the relevant socially salient group. An alternative approach is to consider the broader group of the homeless as socially salient. Whether one lives on the streets or can secure a temporary bed at a shelter or with a friend affects the extent to which the person experiences some of the disadvantages associated with homelessness, but does not change the fact that they are a member of a group that is negatively perceived. Indeed, being perceived as homeless could, for many, mean being perceived as living on the streets, even where this perception is false. As either of these approaches seems appropriate, we will remain neutral on whether the relevant socially salient group is the street homeless or the homeless in general.

Direct discrimination

Having established that the homeless can reasonably be considered a socially salient group, the next question is whether putative cases of anti-homeless hostile design should be considered direct or indirect discrimination. This section considers direct discrimination, while indirect discrimination is considered in the next section.

In deciding whether hostile design is best understood as direct discrimination, the crucial aspect to clarify, as per the definition above, is whether the homeless are treated worse than others because they are homeless. The pull of the idea that it is not direct discrimination is evident. A bench modified to prevent sleeping or spikes on the ground under a bridge prevent everyone from sleeping in these places, not the homeless alone. While on the surface, hostile design treats everyone equally but has differential effects, we cannot easily conclude that it is not direct discrimination. We would also have to show that those who implement hostile design do not treat the homeless worse because they are homeless. Therefore, the intentions of the broad set of actors who implement hostile design are crucial.

A useful test considers whether the actors would have still acted as they did had they not believed the affected people to be members of the specified group (here, the homeless) (Berndt Rasmussen Reference Rasmussen2020, 732). But it seems at least usually wrong to suggest that anti-homeless hostile design would still have been implemented were the affected people not perceived as homeless. There would be no cause for anti-homeless benches or spikes if none of the affected people were homeless – if, for instance, they were commuters or shoppers. Many hostile designs are specifically targeted at the homeless and therefore qualify as direct discrimination.Footnote ¹³

This still might seem too quick. One could say that if the purpose of hostile design is to deter specific activities in certain public spaces, by making sleeping uncomfortable or impossible, then this applies equally to all groups. This counterargument says hostile design does not treat the homeless worse than others. Instead, it makes it uncomfortable for anyone to sleep in a public space, homeless or not. It is people sleeping in public rather than the homeless that are targeted by hostile design, which, therefore, does not amount to direct discrimination against the homeless.

Yet it typically seems rather implausible that it is sleeping as such that is targeted, rather than the homeless sleeping. This can be evoked through the following analogy. Suppose someone wanted to remove all the ramps from public buildings, and claimed that this was not direct discrimination against people with disabilities – they just wanted to stop people using wheels to go into buildings, whether they are disabled or not. Even if the occasional person would otherwise go into the building on a bicycle or skateboard, we surely would not find it very plausible that this is not direct discrimination against people with disabilities. For the same reason, it is not usually plausible that barriers to public sleeping are only that, rather than direct discrimination against the homeless.

Indirect discrimination

Suppose, however, we had a case where sleeping people really were the intended target of hostile design rather than homeless people. Perhaps many young people who are not homeless begin to sleep on benches due to the lack of nighttime public transportation, and benches in train stations are modified to prevent this. Assuming this is the real purpose of the hostile design, this would not directly discriminate against homeless people. The key point for our purposes, however, is that it would still indirectly discriminate against them. This hostile design does not disadvantage the homeless because they are homeless, but it still disadvantages them.

Likewise, while CCTV cameras sometimes target the homeless, they can also be installed for other reasons – notably to deter drug dealing, prostitution, violent crime, and property crime. In some such cases, they may be, all things considered, beneficial to the homeless (for example, where the homeless themselves are frequently victims of crime). In such cases, they are clearly neither hostile design nor discriminatory against the homeless. However, where the CCTV is genuinely installed to deter crime but has overall negative consequences for the homeless, it seems the conditions for indirect discrimination are satisfied: the CCTV would have been installed even if the homeless were not homeless, but nevertheless impacts the homeless far more than other people. A broad account of discriminatory hostile design follows from this. For instance, CCTV located in a place never used by the homeless might nevertheless indirectly discriminate against them if they would use it, were it not for the CCTV. Likewise, the defensive architecture proposed by advocates of defensible space theory (Newman Reference Newman1977) and Crime Prevention Through Environmental Design (CPTED) (Crowe and Fennelly Reference Crowe and Fennelly2013; Cozens Reference Cozens2013) may sometimes amount to indirect discrimination against the homeless, where they are negatively impacted by its territoriality, surveillance, and image maintenance.Footnote ¹⁴ To reiterate, it is a separate question whether such discrimination is wrongful.

It should finally be noted that some instances of hostile design are directly discriminatory against some homeless people and indirectly discriminatory against other homeless people. A regulation proposed in Hamburg for the 2006 World Cup would ‘prohibit at least professional bands of beggars from Eastern Europe’, but ‘would not be directed against the old established beggars of Hamburg’ (S. Nagel Reference Nagel2007, 11). While this regulation was ultimately defeated, with no actual hostile design following from it, it and similar cases elsewhere in Europe (for example, the removal of sleepable benches or introduction of anti-homeless benches in Italian cities to exclude immigrants) suggest that the motivation for some hostile design may be anti-immigrant rather than anti-homeless per se (Bergamaschi, Castrignanò, and Rubertis Reference Bergamaschi, Castrignanò and Rubertis2014; Tosi Reference Tosi2007). Such hostile design exhibits indirect discrimination towards the non-immigrant homeless, as they are not the intended target of the measures, but are nevertheless among its victims. As regards homeless immigrants, who are the express targets of this hostile design, the discrimination is direct.

In sum, it seems reasonable to consider most anti-homeless hostile designs as direct discrimination against the homeless, with the remainder being indirect discrimination or a combination of direct and indirect discrimination.

The comparative account

What, then, does the comparative account make of hostile design discrimination? We should first consider whether the homeless face inequality of opportunity. This question can be considered at two levels.

First, there is a clear and vast inequality of opportunity between a homeless person and an average member of society. As noted above, being homeless makes it much more difficult to find and keep a job, start and maintain a family life, and access social insurance and voting rights. It also has severe impacts on mental and physical health, access to medical care, and mortality. A German study found that at least 75 per cent of the homeless currently have a mental illness requiring treatment (Kaduszkiewicz et al. Reference Kaduszkiewicz, Bochon, van den Bussche, Hansmann-Wiest and van der Leeden2017). The physical health of the homeless is often affected by respiratory problems, cardiovascular disorders, injuries, foot problems, intoxication, and infectious and parasitic diseases (Kaduszkiewicz et al. Reference Kaduszkiewicz, Bochon, van den Bussche, Hansmann-Wiest and van der Leeden2017; Liu, Chai, and Watt Reference Liu, Chai and Watt2020; To, Brothers, and Van Zoost Reference To, Brothers and Van Zoost2016). In addition, the homeless are more likely to be undertreated (if at all) in the medical system (Kaduszkiewicz et al. Reference Kaduszkiewicz, Bochon, van den Bussche, Hansmann-Wiest and van der Leeden2017) and, when hospitalized, more likely to experience readmission (Lewer et al. Reference Lewer, Menezes, Cornes, Blackburn, Byng, Clark, Denaxas, Evans, Fuller, Hewett, Kilmister, Luchenski, Manthorpe and McKee2021). For all these reasons and more, mortality rates among the homeless are significantly higher than the general population (O’Connell Reference O’Connell2005; Funk et al. Reference Funk, Greene, Dill and Valvassori2022). All this suggests that the homeless face unequal opportunities.

Yet it might be countered that these current or ‘point in time’ inequalities of opportunity are consistent with lifetime equality of opportunity and that it is lifetime equality of opportunity that matters. After all, most leading theorists of equality hold that it is lifetime prospects that must be equal (Rawls Reference Rawls1971, 78, 178; Dworkin Reference Dworkin1981, 304–5; Arneson Reference Arneson1989, 85–86). It has been suggested that anti-homeless measures ‘are designed with the assumption that homelessness is caused by the poor life choices of the individual’ (Southworth and Brallier Reference Southworth and Brallier2023, 41), perhaps informed by media portrayals of the homeless as ‘elderly male alcoholic rough sleepers who invariably beg and are work-shy’ (Devereux Reference Devereux2015, 264). The British Home Secretary recently complained of the homeless ‘living on the streets as a lifestyle choice’ (Otte Reference Otte2023). This takes us to the second, deeper level of analysis.

Might the homeless have had equal lifetime opportunities, which they squandered? This is highly unlikely given the significant unchosen disadvantages faced by the homeless prior to their becoming homeless. The U.S. Department of Housing and Urban Development reports that 16.8 per cent of people experiencing homelessness are under 18 (de Sousa et al. Reference de Sousa, Andrichik, Cuellar, Marson, Prestera and Rush2022, 10). Furthermore, many homeless adults transitioned into homelessness as children: one study found that 35 per cent of the homeless came from a youth-to-adult pathway, being 21 or older but having first been homeless when 18 or younger (Chamberlain and Johnson Reference Chamberlain and Johnson2013). Disadvantaged racial minorities are often massively overrepresented among the homeless – for instance, Black and African Americans make up 37.3 per cent of the homeless (de Sousa et al. Reference de Sousa, Andrichik, Cuellar, Marson, Prestera and Rush2022, 10) but only 13.6 per cent of the US population (U.S. Census Bureau 2023). It might be added that a large majority of homeless people are from lower class backgrounds, with homelessness ‘not represent[ing] much of a disjuncture from their lifeworlds, which have already been shaped by poverty, disadvantage and marginalization’ (Hodgetts et al. Reference Hodgetts, Stolte, Nikora and Groot2012, 1214). When housing opportunities are inadequate, housing prices are high, and social benefits are insufficient, people may be driven into homelessness (Crisis 2023). Homelessness may also be due to poor mental or physical health, unemployment, major life events such as leaving an abusive partner, or intellectual disability (Crane et al. Reference Crane, Byrne, Fu, Lipmann, Mirabelli, Rota-Bartelink, Ryan, Shea, Watt and Warnes2005; Mercier and Picard Reference Mercier and Picard2011). Given this catalogue of unchosen disadvantages, one could hardly find a group more gravely denied lifetime equality of opportunity.

The homeless do, then, face inequality of opportunity. This takes us to our second question: is the inequality of opportunity faced by the homeless worsened by hostile design? The answer is surely yes. Hostile design makes it harder for the homeless to find places to rest, sleep, shelter, and maintain personal hygiene, directly lowering their opportunities for welfare as well as their prospects for work and an escape from homelessness. It may also interfere with their deliberative freedom (Moreau 2010; Reference Moreau2017; cf. Waldron Reference Waldron2006), thereby making it more difficult for them to navigate their lives without being confronted by their homelessness.Footnote ¹⁶ Hostile design worsens the options available for those who are already much worse off. If discrimination is wrongful where it is contrary to equality of opportunity, as the comparative account maintains, it is clear that hostile design amounts to wrongful discrimination, as it deepens a large existing inequality of opportunity.

The non-comparative account

We move on to apply the non-comparative account of the wrongness of discrimination to hostile design. This account considers three elements: (1) the overall amount of well-being, (2) how badly off the affected people are, and (3) their desert levels. The application of (3) seems straightforward. While it may be that ‘the public often views the homeless as undeserving of aid’ (Southworth and Brallier Reference Southworth and Brallier2023, 41), it is plain from the preceding discussion that the disadvantages of the homeless are not deserved but the result of circumstances beyond their control.

More complex issues are, however, raised by (1), which expresses the view that well-being is inherently valuable, and (2), which expresses the view that the well-being of the worse off has a higher priority (Parfit, Reference Parfit1998). In principle, this might offer support for hostile design, as harm to the homeless might be outweighed by larger benefits to others. But this is most unlikely to be true of typical cases of hostile design. Consider, for instance, the interests involved in the cases of anti-homeless benches or spikes: on the one hand, we have the need for somewhere to sleep, and on the other, a presumed desire to not see people sleeping in public. The well-being gain from having somewhere to sleep seems to vastly outweigh the supposed aesthetic costs. Footnote ¹⁷ And following from (2), the benefits to the homeless should have additional weight. In other cases, such as a planned absence of public amenities, things are clearer still, as hostile design is bad for the homeless and general public alike.

Yet there may be circumstances where the scales tip in the other direction. Here two factors should be noted. First, where benefits to the general population increase, hostile design may be less objectionable. Suppose, for instance, that an area suffers from serious violent or sexual crimes and that CCTV would deter this. It may not be unreasonable to install CCTV, even though, as discussed earlier, this might amount to indirect discrimination against the homeless. The discrimination would not be wrongful, as it results in overall benefit as per (1), and benefits for at least some of the worse off (victims of serious crime), as per (2). But it must be emphasized that the need for CCTV must be real and not a pretext for wrongful direct discrimination against the homeless. Furthermore, even where the need is real, it must still be balanced against the interests of the homeless, which, as per (2), have additional weight. Current CCTV schemes are very unlikely to be the upshot of such considerations as the potential for wrongful discrimination is not considered.

Second, where the harm to the homeless decreases, hostile design may be less objectionable. Suppose, as a hypothetical, that the effect of the homeless not sleeping on benches or next to buildings is that they use shelters, which an enlightened city makes freely available, along with effective outreach and support services. While anti-homeless benches and spikes might still be, to an extent, disadvantaging in such circumstances, as they are still somewhat exclusionary, their harmfulness is reduced as the outcomes for the homeless are better. Now, this may seem a paradoxical case, as there would presumably be greatly reduced demand for rough sleeping in such a city, so there would be reduced ‘need’ for hostile design. This is correct but does not undermine the point of principle that hostile design is less objectionable, the better the alternatives for the homeless. Note as well the corollary, which has greater practical importance given the unjust circumstances of actual modern cities, that hostile design is more objectionable the worse the alternatives for the homeless.

In sum, we have found that the two harm accounts identify typical hostile design as wrongful discrimination. While there may be some permissible cases of indirect discrimination, the wrongfulness of hostile design is exacerbated by the more broadly disadvantaging circumstances faced by the homeless.

The mental-state account

On the mental-state account, disrespect concerns what or how the discriminator thinks about the discriminatee. Important versions of this account have been proposed by Arneson and Alexander. Arneson’s version holds that wrongful discrimination occurs when the mental state of the discriminator reflects ‘unwarranted animus or prejudice’ against the discriminatee, where ‘[a]nimus is hostility or, more broadly, a negative attitude, an aversion’ (Arneson Reference Arneson2006, 779, 787). In Alexander’s version, wrongful discrimination is about beliefs regarding moral inferiority (Alexander Reference Alexander1992). This could take various forms – for example, Alexander highlights that discrimination based on inaccurate stereotypes or prejudices will often be wrongful.

Is it reasonable to suggest that hostile design would be disrespectful discrimination on such accounts of discrimination? The first question is, whose mental state? We noted already that hostile design is brought about through a wide range of actors, such as public officials, private businesses, and designers. It is their mental states that are of interest.

We do not, of course, have direct access to the mental states of such actors. But they are, to some extent, revealed by their actions and, in some cases, by whistleblowers (Ong Reference Ong2020). The very purpose of directly discriminatory hostile design seems revealing. Regarding Arneson’s view, it is easy to interpret the design of cities with the purpose of targeting a particular group of citizens as something that reflects at least a negative attitude towards the homeless, and perhaps outright hostility. It is arguably more difficult to apply Alexander’s version of the mental state account, but it is not outlandish to suggest that there is a belief about inferiority involved when the already rough conditions of homelessness are purposely worsened.

These assessments can be corroborated by empirical evidence about attitudes toward the homeless and policies targeting the homeless. While broad attitudes of sympathy and an understanding of the social or structural causes of homelessness are identified (Tsai et al. Reference Tsai, Lee, Byrne, Pietrzak and Southwick2017), this is not the full picture. The already mentioned studies, where people categorize homeless people among the least warm and least competent, suggest a different conclusion (Fiske Reference Fiske2012; Harris and Fiske Reference Harris, Lasana and Fiske2009; Fiske 2013; Reference Fiske2018), even if the implications for policy preferences are not always clear (Cikara and Fiske Reference Cikara and Fiske2011). Another study found that people’s preferences for exclusionary policies against the homeless are rooted in feelings of disgust (Clifford and Piston Reference Clifford and Piston2017), and being homeless is stigmatized to a similar high degree as being mentally ill (Phelan et al. Reference Phelan, Link, Moore and Stueve1997).

However, the mental-state account does not seem to oppose indirectly discriminatory hostile design. For example, where CCTV impacts the homeless but does not do so because they are homeless (for example, it is purely motivated as an anti-crime measure), there is no animus or judgement of inferiority regarding homeless people, and therefore no wrongful discrimination on the mental-state account. Likewise, where public amenities are removed purely for cost reasons, without a mental state of hostility or superiority regarding the homeless, there is no wrongful discrimination on this account. It stands repeating that it is the actual, rather than merely publicly stated, motivations that are relevant here. CCTV installed expressly for anti-crime purposes, but with collateral effects on the homeless that would not be tolerated were they to fall on others, would cross over into direct discrimination and suggest that the homeless were treated with animus or as inferiors. Still, at the level of principle, while hostile design that directly discriminates against the homeless is wrongful on the mental-state account, indirectly discriminatory forms are not.

The social meaning account

The second respect-based account of the wrongness of discrimination is the social meaning account (Hellman 2008; Reference Hellman and Lippert-Rasmussen2017). On the social meaning account, it is not a requirement that anybody believes the discriminatee to be inferior. On this account, the mental state is not what is important. This is interesting for the discussion at hand because those introducing hostile design often deny that they are motivated by such feelings or thoughts as those required by the mental state account. Instead, this account is concerned with how hostile design expresses social meaning. The meaning here is understood objectively, which is to be derived from a particular context. According to Hellman, the meaning of a discriminatory action or policy is problematic when it is demeaning. Specifically, this happens when two conditions are satisfied: the action or policy expresses ‘that a person or a group is of lower status’ and the ‘actor or institution expressing this meaning must have sufficient social power for this expression to have force’ (Hellman Reference Hellman and Lippert-Rasmussen2017, 102).

The social power-based condition is clearly fulfilled in most cases of hostile design. When, for example, a municipality introduces a hostile design to the benches on the local train station or makes the pavement under a bridge so uneven that it cannot be used for sleeping, we are clearly dealing with an actor or institution with significant social power.

The remaining question is whether hostile design expresses that the homeless have a lower status. The explicit measures of hostile design, such as anti-homeless spikes or signs, are very visible and, therefore, probably also demeaning, as their express purpose is to exclude the homeless. To see how such signs are demeaning, consider a parallel case, where signs attempt to deter people with disabilities from entering a particular area. This would presumably be taken as demeaning. And that would be true even if the signs show, say, a crossed-out wheelchair, even though many people with disabilities do not use wheelchairs and people without disabilities can use them. This suggests that, even where anti-homeless signs show opposition to homeless-associated activities, such as camping or panhandling, rather than being expressly anti-homeless, they can be seen as conveying that the homeless have lower status.

More implicit kinds of hostile design may seem to have little expressive force. CCTV cameras, altered benches that go unnoticed by most, and the deliberate unavailability of public restrooms are not demonstrable and visible decisions that reinforce existing negative attitudes toward the homeless. The ‘hostility’ of such measures may only be detected by the homeless themselves. But that may in itself be enough for them to be demeaning: equivalently, insulting letters from the government might be considered demeaning even if no one but the recipients knew of their content. In addition to this, Hellman allows that something can be demeaning as long as it has the capacity to be demeaning, even if it is not viewed as demeaning. It is enough that it ‘if known, would have this meaning’ (Hellman Reference Hellman and Lippert-Rasmussen2017, 105). For these reasons, Hellman’s account can consider even implicit measures demeaning.

The deliberative account

The third group of disrespect-based accounts concerns deliberative disrespect. According to Eidelson, ‘discrimination is intrinsically objectionable when it is basically disrespectful of the personhood of those who are discriminated against’ (Eidelson Reference Eidelson2015, 95). Moral personhood consists of moral worth and autonomy. The first pertains to the equal respect that all of us are owed: ‘To respect a person’s equal value relative to other persons one must value her interests equally with those of other persons, absent good reasons for discounting them’ (Eidelson Reference Eidelson2015, 97). The second aspect of personhood is autonomy. It is disrespected by failures to treat a person as an autonomous individual capable of making autonomous choices (Eidelson Reference Eidelson2015 128, 144).

Anti-homeless hostile design often seems to question the choice-making capacities of the homeless, as when anti-homeless benches or spikes signal that the homeless are making such poor choices about where to sleep that they must be overridden. It can, therefore, be seen as sometimes disrespectful of autonomy, the second aspect of moral personhood. The first aspect of personhood, which requires that people’s interests be respected equally, also seems very relevant to the discussion of hostile design. As we pointed out in our discussion of harm, the implementation of hostile design can be seen as a weighing of interests under the assumption that the general population have an interest in not encountering the homeless at certain places in the city. Priority for such ‘interests’ seems clearly to imply lesser regard for the weightier interests of the homeless. Interestingly, this verdict applies to both implicit and explicit, and directly and indirectly, discriminatory cases of hostile design.

Overall, the disrespect accounts tend to oppose anti-homeless hostile design. However, there are some important differences in emphasis between the mental state, social meaning, and deliberative accounts. In particular, while the mental-state account objects only to directly discriminatory hostile design, the social meaning and deliberative accounts also object to indirectly discriminatory hostile design.