2. Are cases of accessing and spreading falsehoods defamation?

As many have been quick to point outFootnote ¹⁸ , the canonical wrong of propagating falsehoods is defamation, i.e. libel or slander. Propagating the false claim that someone is a murderer – as I did in Murderer 3 – seems like a classic case of defamationFootnote ¹⁹ . So we might view that in telling all my friends about John’s supposedly murderous father, I may have thought I committed the wrong of a privacy invasion (given I assumed the information about John’s father was true), but the wrong I effectively committed – somewhat ironically – was slander.Footnote ²⁰

The problem with this view is that defamation does not neatly capture all intrusions that would, if relevant information were true, qualify as privacy invasions. Firstly, legal actions in defamation typically require demonstrable economic or reputational harm. This is different for privacy invasions, which are viewed as completed as soon as relevant personal information is accessed,Footnote ²¹ irrespective of whether any downstream reputational or economic harm ensues. A further distinction is that actions in defamation typically require that the defendant (or, in civil actions, the plaintiff) acted with the knowledge that the information they published was false, or at least that they were recklessness in deciding to publish the statement without investigating its accuracy.Footnote ²² In other words, in a case such as Murderer 3, where I reasonably (and sincerely) took the information about John’s father to be true, would be ruled out as defamation. Privacy, at least in law, protects more tightly.

One might object that these distinctions are merely a matter of positive law, but not of moral merit. That legal actions in defamation claims in systems like the US and the UK happen to have these various prerequisites should not bar us from using the moral concept of defamation to account for cases of propagating falsehoods (i.e. the case I describe as Murderer 3). So, even though legal actions of defamation might fail in the courts, we might view that morally speaking, in the case of Murderer 3, I committed the wrong of defamation.

But plugging the (moral) wrong of defamation in cases where the person accessing relevant information clearly took themselves to be spreading true information seems odd. Defamation and privacy violations may be complimentary wrongs in the sense that they are both potentially concerned with reputational harm. But their respective relationships with truth have opposing normative implications: the kinds of beliefs about another person that are thought to diminish privacy typically require some inquiry work – some digging or searching or hacking – or they involve exploiting that one already stands in an epistemically privileged (i.e. confidential) relationship to the victim. As Kappel argues, what makes privacy violations intuitively wrongful has to do with the ‘epistemic pathway’, the how one comes by this information.Footnote ²³ The wrongness of defamation seems inverted: defamatory falsehoods are exceedingly cheap to come by; they can just be made up, out of thin air. Indeed, any relevant epistemic effort (i.e. digging or searching or hacking) in the context of coming by information used for defamatory purposes does not ground wrongdoing, but instead counts in the perpetrator’s favour insofar it suggests that active steps were taken to verify the claims’ accuracyFootnote ²⁴ . Put crudely, privacy invasions are about checking propositions about others too thoroughly, while defamation is, at best, about not checking propositions thoroughly enough (and, at worst, about intentionally propagating what which one knows to be false). In defamation cases, truth counts as a defence; it is what licenses spreading statements which cause reputational and economic harm. The normative vectors of the wrongs of defamation and privacy invasions clearly point into opposing directions.

There are many ways of defaming someone that do not at all involve sharing information which is private or personal, or which would be private had the information been true. The International Libel and Privacy Handbook lists the following classic examples for statements found to be defamatory under U.S. law: ‘playing a role in an alleged kidnapping and murderFootnote ²⁵ ; incompetence of a professional’Footnote ²⁶ ; or ‘being a communist’Footnote ²⁷ . More recently, the far-right radio show host Alex Jones repeatedly claimed that the 2012 Sandy Hook Elementary School shooting was a hoax and that the families of the victims were lying about the tragedies they suffered (the families of victims faced harassment and death threats as a consequence). Jones was held liable to pay roughly 960 million USD in damages for defamation.Footnote ²⁸ In none of these cases would the communicated falsehoods qualify as protected by privacy, had they been true.

On the side of privacy, we can equally conceive of many cases of spreading private information which would not count as defamation if false, because they would not be ruinous to one’s reputation. Consider the case of a proud mother who violates her adult son’s privacy by sharing his excellent GRE results on her public Facebook page, even though she knows that the son does not want this information to become public. If the son’s GRE results turned out to be false due to some system error, what the mother did would still not count as libel or slander; the sharing of the test results is aimed at promoting, rather than diminishing, her son’s reputation.

For the purposes of my argument here, a final, compelling reason to remain unsatisfied with the view that defamation can mop up cases falsehoods is that the wrong of defamation is incapable of accounting for the intuitive wrongfulness Murderer 2, i.e. the variant where I read (i.e. access) the letter John received from his mother and keep the contents to myself. Defamation is a wrong of an inherently communicative nature, so we can hardly stretch it to cases of ‘mere’ access.

3. Can we commit privacy violations without diminishing privacy?

The intuitive wrongfulness of accessing falsehoods (i.e. Murderer 2) has worried other access theorists. Notably, Klemens Kappel suggests we allow that privacy diminishments and privacy violations (or ‘wrongs’, as he calls them), come apart in instances where someone forms a false belief based on illegitimately accessing sensitive information about another person. In Kappel’s hypothetical (which I will proceed to call Adam), I break into my neighbour Adam’s house and read a letter from which I mistakenly infer (because of my poor German) that Adam is a member of a neo-Nazi party (in fact, the letter contains a rejection of an application for party membership)Footnote ²⁹ . Kappel contends that Adam’s privacy has not been diminished by my reading the letter, given that the belief I form as a consequence is false. Nevertheless, he maintains, our intuitions mandate that this case must still count as a privacy violation.Footnote ³⁰

The distinction between privacy simpliciter and a normatively laden right to privacy is fairly common in the philosophical literatureFootnote ³¹ ; particularly amongst access theorists who view – somewhat controversially, in my view – that privacy simpliciter is lost or ‘diminished’ every time personal information is divulged voluntarily to another personFootnote ³² , but who do not want to normatively label such innocuous cases privacy violations. On this view, not all privacy diminishments are privacy violations, so the right to privacy is drawn more narrowly and thus only covers a strict subset of the things that are part of privacy simpliciter.

However, what Kappel suggests when he argues that it is possible to violate a right to privacy – without actually diminishing privacy – is that the sets of things covered by privacy simpliciter and by the right to privacy merely intersect. In other words, he, in some respect, draws the right to privacy more broadly than privacy simpliciter. This begs the question of what exactly this supposed right to privacy is tracking. Lauritz Munch and Jakob Mainz have described Kappel’s way of distinguishing a right to privacy from privacy simpliciter as incurring a theoretical debt, since ‘on any plausible account of the right to privacy, privacy must be the object of this right’.Footnote ³³

One account that offers a more substantive explanation for how violations of privacy can be drawn more broadly than privacy diminishments is Carissa Véliz’ ‘hybrid account’ of privacy. She distinguishes between privacy, which she defines roughly along the lines of access-theoretical accounts, and broader ‘right to privacy’ or ‘robust privacy’, which tracks other accounts of privacy as control over informationFootnote ³⁴ . Accordingly, cases where ‘A secures a position from which to invade S’s privacy with the intention of invading her privacy […] in some counterfactual circumstance’Footnote ³⁵ already qualify as violations.Footnote ³⁶ In other words, the right to privacy is violated even in cases where privacy simpliciter remains undiminished (‘unaccessed’), such as a case where a friend acquires control of one’s diary ‘just in case she might feel like reading it in the future’Footnote ³⁷ . This is possible because Véliz defines the ‘right to privacy’ by adopting Philip Pettit’s notion of ‘robustly demanding goods’.Footnote ³⁸ She argues that to be said to have ‘robust privacy’Footnote ³⁹ , it is not enough that one’s privacy remains free from violation under the narrow conditions of the here and now, but that privacy remains ‘unaccessed’ in alternative possible words: ‘Robustly demanding goods are ones that require counterfactual assurances’.Footnote ⁴⁰ What does it mean to assure something counterfactually? Pettit himself gives the example of love and explains:

The contention is that if my love for you is built on exceedingly shallow, contingent criteria – that your name happens to be ErnestFootnote ⁴² , for example – then this ‘love’ is so thin that it does not properly count as love at allFootnote ⁴³ . Love, according to Pettit, is a robustly demanding good in that it requires assurances beyond the contingent world we happen to live in. Is this also true of privacy? Véliz certainly thinks so; if the only reason I am not reading my roommate’s diary right now is that we presently get along amicably – but if we were to fall out, I would read it immediately – then I cannot be said to be respecting my roommate’s ‘robust privacy’. In this way, Véliz’ argument richly illuminates what it means to positively be guided by another person’s privacy in genuine, non-instrumental ways.

Before I discuss this account further, I should point out that Véliz does not herself claim that her hybrid account of privacy protects falsehoods from access or dissemination. But we might read her account such that it accommodates the view that accessing falsehoods count as violations of the ‘right to privacy’ (without actually diminishing privacy simpliciter). One way to do this is to sketch the truth of the relevant information as part of the counterfactual world; we might thus contend that if you are disposed to reading my letter in a possible world where the content of the letter was true, you do not respect my robust privacy. Again, the violation here depends on what you would do if things were (slightly) different, so the charge regarding Murderer 2 is not that you did in fact read the letter containing falsehoods, but that you would (also) have read it in a possible world where the information it contained was true.Footnote ⁴⁴ (In Murderer 2 and 3, the perpetrator who read the letter assumed the information was true, so we can grant that had it actually been true, they would have acted just the same.)

For the case of Murderer 2, this solution has some appeal because it captures the detail the perpetrator clearly demonstrated a lack of appreciation for the victim’s privacy (even if we think they did not succeed in diminishing it). This view takes the perpetrator to task for the possible world they presumed themselves to be acting in (rather than blaming them for something they did not take themselves to be doing, i.e. defamation).

However, counting what one might have done in a counter-factual world as a violation of a right seems overzealous. Is ‘failing to respect’ another person’s robust privacy akin to violating her right to privacy?Footnote ⁴⁵ Véliz commits herself to the view that I am actively violating my roommate’s right to privacy in the here and now – despite the fact that I am not even reading her diary – just because I would read it, if circumstances were slightly different (i.e. if we got into a fight)Footnote ⁴⁶ . That seems inflationary. It is questionable that the terminology of ‘right’ serves Véliz’ argument well. Pettit himself characterises robustly demanding goods like love and friendship as morally valuable objectives which we ought to, prima facie, pursue. But he comes up short of suggesting that these goods coagulate into definable ‘rights’. In other words, when I fail to love others in robust ways, I am doing just that – failing to love; failing to maximise some ‘rich good’Footnote ⁴⁷ , acting in morally suboptimal ways – but I am not quite violating anyone’s right to be lovedFootnote ⁴⁸ . Rights, on a more orthodox philosophical understanding, are more modest deontological concepts than virtues or consequentialist ideals; they leave room for supererogationFootnote ⁴⁹ . Even if I lack the disposition to respect your privacy counterfactually, across various other possible worlds, it seems intuitive that as long if I am not reading your diary in this one – however reluctantly or contingently – then I am not quite in violation of your right to privacy (at least not yet Footnote ⁵⁰ ).

Accounts which draw the right to privacy more broadly than privacy itself are potentially able to count Murderer 2 and 3 as privacy violations whilst holding that privacy (or privacy simpliciter) remains undiminished. But accounts which have been offered in the literature so far either fail to explain what, aside from privacy simpliciter, the (broad) right to privacy protects, or they draw what this right protects so broadly that they end up with an indefensibly broad view of what it is to violate the right to privacy.

4. Can control views of privacy account for the wrongfulness of accessing falsehoods?

‘Control views’ of privacy typically confirm violations of the right to privacy when another person establishes control over a person’s information without their permission; even when that information remains ‘unaccessed’Footnote ⁵¹ (i.e. before beliefs with any relevant propositional content were formed).Footnote ⁵² Control accounts thus view that a privacy violation has occurred at an earlier logistical step than access accounts. Some consider this a feature and some a bug: those who seek to describe widespread data collection practices of, e.g. the NSA as constituting a loss of privacy (rather than a mere privacy risk) can use control accounts to do so. Others take the view that control accounts are overinclusive in picking out privacy diminishments, especially in cases where relevant information technically comes under another person’s control, but never accessed. For example, against control accounts, Kevin Macnish has defended the view if I leave my diary on a table in a coffee shop and when I come back to retrieve it after 30 minutes, a stranger finds it and returns it to me – without ever having opened it or having intended to open it – then I have lost no privacy.Footnote ⁵³

The merits and demerits of control views are beyond the scope of this article. But it is noteworthy that control views draw distinctions when it comes to false beliefs that access accounts of privacy pass over. Note that the case of Kappel’s Adam differs slightly from the Murderer cases: in the case of Adam, the letter from the Neo-Nazi party does contain sensitive, true information; the intruder just did not end up forming true beliefs about Adam’s Neo-Nazi party membership because of some epistemic failing on their part (i.e. lack of linguistic competence). By contrast, in the case of Murderer 2, the letter John’s mother wrote to John contained false information. For access accounts of privacy, it is irrelevant whether the relevant false beliefs are due to some epistemic failure, or whether they are due to information being false. Any failure to form true beliefs about another person will mean the other person’s privacy remains, to that extent, undiminished. Why the belief ends up being false – i.e. because the information is encrypted; because is written in another language; because the information is false; because the information is simply misunderstood – does not matter.

By contrast, control accounts would confirm that a diminishment of privacy has occurred in cases where control over true information was established, even if it did not lead to true beliefs (i.e. in cases where information turns out to be written in a foreign language or encrypted; in cases where relevant information ends up being misunderstood). But control accounts would deny – at least to the extent that they view privacy as factive – that a diminishment of privacy has occurred in cases where relevant information was itself false (i.e. Murderer 2 and 3).

Is there any normative justification for treating these cases differently? We might grant that in cases where control views confirm privacy losses, true information was at stake. This would matter to those who define privacy as factive (i.e. as being concerned with truths). Then again, this view is begging the question when it comes to the wrongfulness of accessing falsehoods.

On the other hand, taking up a more pragmatic perspective, we might not think it matters much at which step in the belief formation we locate the epistemic shortcoming. After all, the line between encrypted information and false information can seem arbitrary. Imagine that, as part of an encryption mechanism, I insert ‘it is not true that’ before every sentence in my private correspondence, such as before the sentence ‘I have an intimate relationship with Anna’. What would the control view say about a person who acquires my letters and reads them, but who does not succeed to decrypt my writing and thus forms the false belief of ‘it is not true that she has an intimate relationship with Anna’? We might view that the (post-encrypted) information printed in the letter was false, in which case control accounts would tell us no control over true information was ever established (ergo no privacy loss). Or did the person who read my letters acquire control over true information (plus a layer of encryption)? In that case, control accounts would view that a privacy loss did occur.

Similarly, consider cases of linguistic ambiguity. Suppose that I, a German speaker who has lived abroad for many years, write my diary using a random mix of German and English. On one page, I write:

(This translates to ‘This morning, I bought the gift for my husband. I will give it to him tonight’.) When I write the word gift, I use it with its English meaning; i.e. ‘present’. But in German, ‘Gift’ means ‘poison’. Assume now that a German speaker acquires control of this page of my diary without my permission. Not realizing that I sometimes use random English words in my writing, they form the false belief that I am about to poison my husband. Is this false belief due to the information on the page being false, or due to some hermeneutic shortcoming on the intruder’s part? What I wrote is subjectively true (i.e. true if interpreted as pragmatically enriched by the idiosyncratic way I use languagesFootnote ⁵⁴ ; I did buy a present for him after all.), so control accounts would view that privacy was lost. But what I wrote is also literally false, ergo no privacy loss. Which view to take? The point is that there are various cases where seems arbitrary to view that a privacy loss turns on the whether the information was encrypted or false.

The bottom line is that control views may be more accommodating of the view that privacy losses have occurred in some cases where false beliefs are concerned. But their generosity is arbitrary. Therefore, they are not obviously preferable when it comes to explaining whether or not falsehoods diminish privacy. In any case, my objective here is to explain the intuitive wrongfulness of Murderer 2 and 3; i.e. of cases where false information was accessed. On control views, no privacy loss/ diminishment occurs in these cases, so they do not explain what, if anything, was wrongful in these cases.

5. Do falsehoods diminish privacy indirectly by pressuring the disclosure of truths?

In the academic literature on falsehoods, one frequently discussed insight is that propagating falsehoods may lead to privacy diminishments down the line. Ruth Gavison explains that falsehoods, if sufficiently spectacularFootnote ⁵⁵ , lead to a loss of anonymity and increased attention:

Gavison suggests that the circumstances of such enhanced scrutiny constitute a loss of ‘attentional privacy’Footnote ⁵⁷ , ‘or at least the threat of such a loss’Footnote ⁵⁸ . Similarly, Véliz points out that false information may create pressure on those it concerns ‘to reveal more than they would’ve otherwise wanted to’.Footnote ⁵⁹ In other words, private facts may end up being divulged in defence of false rumours.Footnote ⁶⁰ In an example Véliz gives, a person who is the victim of rumours about her sex life feels compelled to defend herself by disclosing that she is ‘single and celibate’.Footnote ⁶¹

Applied to the Murderer 2 and 3 cases, this view suggests that privacy is ultimately lost only when John feels pressure to fight these false claims about his father by divulging true pieces of personal information about himself. We can imagine that John might react to other people’s false beliefs about his father by telling the story of how his mother lied to him in her letters. A false rumour leading to some dark family secrets being revealed; this seems much like the characteristic ‘downstream’Footnote ⁶² privacy loss Gavison and Véliz have in mind.

But it seems far from given that any divulsion of true facts will follow from accessing and spreading falsehoods in the Murderer cases. First, John might never learn that the claims about his father’s criminal status were false, ergo he will feel no pressure to divulge sensitive truths. Second, in the case of Murderer 2, John might never find out that another person read his letter and now holds the relevant false beliefs about his father’s criminal status. Again, there is no reason he would feel tempted to divulge truths. Third, even if John did find out the claims about his father’s criminal status were false and that others (falsely) believe them, he might just choose not to correct these false beliefs. Alternatively, John might choose to undermine others’ false beliefs by spreading yet another falsehood about his father (i.e. a falsehood he is more comfortable with, such as ‘My father works for the CIA’). In all these cases, accessing and spreading falsehoods does not have the downstream consequence of leading to true information being divulged, hence no privacy losses.

Finally, if John did feel compelled to defend himself against false beliefs about his father’s criminal status by divulging truths, it seems he might just do so by parsimoniously stating ‘the claim that my father is a convicted murderer is false’. Does such a limited correction diminish his privacy? One might take the view that facts about one’s parent’s criminal status is ipso facto personal information, as a matter of genre. So the fact that John’s father is not a convicted murderer is equally a matter of privacy. But this view misses that it is often a proposition’s particular content – rather than its mere subject matter – which grounds our interest in keeping relevant information unknown.

We tend to hold a range of defeasible presumptions – quietist default beliefs about others – that we take to be ‘normal’ or ‘neutral’. These presumptions often lack statistical justification; they are not necessarily based on fact. For example, as a matter of default, I tend to presume that any one of my colleagues does not suffer from cancer, that they are not presently depressed, that they are not victims of domestic violence, and that their parents are not convicted murderers. Given that, statistically, a relevant number of my colleagues are bound to have cancer, be depressed, etc., my quietist default beliefs are epistemically unjustified.Footnote ⁶³ We tend to make these presumptions about others as a matter of social convention. It would seem impolite or intrusive to presume that some of my female colleagues are victims of domestic violence, even though I am more likely to hold correct beliefs by guessing so than by ignoring the possibilityFootnote ⁶⁴ . There is, in other words, a socially mandated benefit-of-the-doubt bias in favour of ‘neutrality’ that governs what we believe about others. These conventional ‘neutral’ presumptions are part of our public personas: I hold these default beliefs about others and assume that others hold them about me. When information is revealed which merely validates these quietist default beliefs – information that confirms I really am, in this sense, normal, such as the fact that my father is not a convicted murderer – then it is hard to see how privacy has been lost. (Or, at the very least, it is hard to see that privacy has been lost to a comparable degree as if it became public that my father was indeed a convicted murderer. In other words, facts about by father’s criminal status weigh on my privacy in radically different degrees, depending on their content.) If it was publicly revealed that my father is not a convicted murderer, I would not take my privacy to be diminished, as this information would only confirm what others already tacitly took to be true about me. (I expect people would respond by saying something like: ‘Oh, I did not think he was a murderer anyways’.) So, in correcting other people’s false beliefs, it does not seem clear that John would diminish his privacy in a relevant sense.

The privacy losses pointed to by Gavison and Véliz are, at most, indirect,Footnote ⁶⁵ so the insight that accessing and spreading falsehoods may lead to privacy losses down the line does not explain the intuitive wrongfulness of the actions in Murderer 2 and 3 in hypothetical variants where no sensitive truths end up being divulged.

6. Do falsehoods diminish ‘propositional’ privacy or ‘attentional’ privacy?

Pierre Le Morvan is very motivated to hold that falsehoods can diminish privacy (he gives us eight arguments that suggest excluding falsehoods from information protected by privacy is counter-intuitive). In order to offer an account of privacy that is properly viewed as ‘diminished’ when falsehoods are accessed or disseminated, Le Morvan proposes to distinguish between propositional privacy (i.e. knowing, or ability to know, of a proposition) and factive privacy (knowing, or ability to know, that a proposition is true)Footnote ⁶⁶ . He illustrates the distinction in the following way:

Falsehoods only diminish propositional privacy, since they do not enable those who believe them to know that the relevant proposition is true.Footnote ⁶⁸ Applied to the Murderer cases, my reading the letter and disseminating that John’s father was a convicted murderer would leave John’s factive privacy intact, but it would diminish his propositional privacy because it would entail that I and others form propositional attitudes about John’s father’s criminal status. Does Le Morvan’s account help us explain in virtue of what my actions in Murderer 2 and 3 were wrongful?

The point of propositional privacy seems to be that people are forming thoughts about an issue at all. This echoes Gavison’s account of ‘attentional privacy’ on which ‘[a]n individual always loses privacy when he becomes the subject of attention.’Footnote ⁶⁹ To be sure, Gavison suggests – sensibly, in my view – that turning our attention to others by concentrating our thoughts on them (i.e. ‘discussing, imagining or thinking about another person’Footnote ⁷⁰ ) is ‘related to privacy in a more indirect way, if at all… For the most part, however, thinking about another person, even in the most intense way, will involve no loss of privacy to the subject of this mental activity’.Footnote ⁷¹

By contrast, according to Le Morvan, propositional privacy is diminished by ‘any propositional attitudes (including speculation)’Footnote ⁷² . Do we have a right that people not entertain any propositions about sensitive facts about us? This seems like an exceedingly broad definition of privacy; on its view, I already diminished my colleagues’ privacy a few paragraphs ago when I wondered whether some of them might be suffering from cancer, depression, or other ills.

In defining propositional privacy such that the mere entertaining of a proposition constitutes a privacy diminishment, Le Morvan effectively equivocates my reading of John’s mother’s letter and then telling others about it (which seems intuitively wrong) and my mere pondering whether or not my colleagues are healthy (which seems intuitively fair game). In other words, yes, propositional privacy counts accessing and spreading of falsehoods as diminishments of privacy. But the stakes of Le Morvan’s account are so deflated that he is unable to explain that diminishing privacy is wrong at all. In fairness, Le Morvan’s objective is to provide a normatively neutral account of privacyFootnote ⁷³ ; he explicitly states that he does not wish to ‘engage in arguments concerning what is, or ought to be, moral and/or legal’.Footnote ⁷⁴

Propositional privacy is too broad and overinclusive to work as a normative account of privacy/an account of a right to privacy. It therefore is unable to explain what, if anything, was wrongfully committed in the cases of Murderer 2 and 3.

In the following two sections, I offer two alternatives for explaining the wrongfulness of accessing and disseminating falsehoods. These are choate offences (section 7) and weakening the truth requirement (section 8).

7. Letting attempts account for wrongfulness?

I suspect we are so motivated to find wrongdoing in the Murderer cases and in Kappel’s Adam in part because the relevant offenders clearly act with malicious intent.Footnote ⁷⁵ But this does not necessarily mean privacy was diminished with view to the propositions that Adam is a member of the Neo-Nazi Party and that John’s father is a convicted murderer. We could bite the bullet and view that in the cases of Adam and Murderer 2 and 3, no privacy diminishment occurred (and that thus, no privacy rights were violated), but still hold the actions of reading the letters (and telling others about it) constitute wrongs because they are attempted privacy violations. We could, in other words, let inchoate offences do the work of capturing the intuitive wrongness of the cases of accessing and spreading falsehoods.

Inchoate liability is a primarily a tool within (the philosophy of) criminal law, so one might wonder whether it is broadly applicable to moral wrongsFootnote ⁷⁶ . However, arguments that morally justify imposing liability for inchoate offences – i.e. the view that inchoate offenders express especially morally blameworthy mental statesFootnote ⁷⁷ , and/or because that they create unreasonable risksFootnote ⁷⁸ – clearly apply to wrongdoing in general. In the Murderer and the Adam cases, the relevant intruders clearly fit the paradigm for inchoate offences; they clearly intended to violate the victim’s privacy and furthermore did all they could to bring about the relevant harm, so the fact that their actions did not result in choate privacy invasions was simply a matter of luck. In the words of Thomas Nagel, it seems intuitive that people should not be morally assessed – positively or negatively– for ‘what is due to factors beyond their control’Footnote ⁷⁹ . The concept of inchoate offences explains moral blameworthiness with a view to situations where such facts which ought not be counted in the defendant’s favour. The chance of the information in the letter being false, or written in a foreign language, should not extinguish moral culpabilityFootnote ⁸⁰ .

One benefit of using the concept of inchoate offences is that we can derivatively explain the wrongfulness of Murderer 2 and 3 in reference to characteristics of the full offence of privacy violation. This is intuitively apt; the wrongs committed in the Murderer cases (accessing and disseminating falsehoods) seem to have more to do with privacy violations than with some other offence, like defamation. It seems fitting to view those who intrude into falsehoods blameworthy in ways that depend on the wrong of privacy violations. ‘Attempted privacy violations’ conveniently allow us to take the view that some ‘privacy wrong’Footnote ⁸¹ has occurred, without viewing that privacy was diminished. So far so good: the reading of the letter in Murderer 2 and the dissemination of information about John’s father in Murderer 3 – as well as other intrusions resulting in false beliefs – are wrongful because they constitute attempted privacy invasions.

However, there are reasons to remain dissatisfied with the ‘attempts’ solution. In particular, the more a case of accessing falsehoods resembles cases of accessing true information, the more it seems arbitrary to let sheer falsehood be the distinguishing factor between ‘merely attempted’ and ‘choate’ privacy violations. Consider a letter from my healthcare provider which is addressed to me and contains a positive Covid_19 diagnosis. My employer opens and reads it without my permission. Based on having read the content of this letter, my employer now forms the belief that I have Covid_19. This is clearly a paradigmatic privacy violation. Still, medical diagnostics remains imperfectFootnote ⁸² . Let us change the hypothetical such that the diagnosis was a false positiveFootnote ⁸³ . On this version, I do not, in fact, suffer from Covid_19. Those who hold that privacy cannot be diminished by false beliefs could thus only conclude that my employer attempted to invade my privacy by reading my letter. (Since the information about my diagnosis was incorrect, reading it was bound to remain an ‘impossible attempt’Footnote ⁸⁴ .) Does this mean that if someone hacks into a medical database and reads the health information on hundreds of patients, they only violate the privacy of those whose diagnoses happen to be medically accurate (and cases where medical information was incorrect must count as mere attempts)? It seems odd that whether or not my employer committed a choate violation should turn on diagnostic accuracy. This is especially true for cases where the ‘true’ information is never revealed, and everyone, including the ‘victim’, lives on assuming that the diagnosis was correct. In such cases, the falsehood of the diagnosis fails to acquire any sociological or psychological relevance. The point here is not that holding those who access falsehoods for attempted privacy violations is misguided in principle, but rather that it does not go far enough: some cases of accessing falsehoods seem to warrant treatment as fully-fledged privacy violations, rather than as inchoate ones.

In consolation to this objection, we might grant that some privacy violations are surely choate in the cases of the false diagnosis, because, after all, my employer ended up with some true beliefs. Specifically, my employer now knows that (a) I underwent a Covid_19 test and I received a diagnosis from my medical provider (whatever the result of the diagnosis may have been), and (b) that the letter informs me the result was positive (whether or not this is medically accurate). Beliefs with this propositional content remain true. The mistake merely lies at the level of the inference, based on the contents of the letter, that my actual health status was Covid_19-positive.

Of course, these bits of information (a and b) are not quite as sensational as the proposition that I actually have a relevant illness. But (a) and (b) are still true beliefs about me that involve sensitive information, so they suffice for a privacy diminishment on my account. Similarly, we might hold that in case of Murderer 2 and 3, I succeeded in diminishing (or violating) John’s privacy in virtue of my now holding the (true) belief that John’s mother wrote to him that his father was a convicted murderer (whether or not this is true). We could piece together the intuitive wrongfulness of my employer reading my medical information – as well as the Murderer 2 and 3 cases and various other cases of false beliefs – with some patchwork of inchoate privacy violations and choate privacy violations regarding of the less sensational (but true) beliefs. For those who are steadfastly committed to the view that privacy is wholly factive, this mix of inchoate privacy violations and (lesser) choate privacy violations may yield a satisfactory solution.

8. Weakening the truth requirement?

I, however, think the view that accessing falsehoods is merely an attempted privacy violation is unsatisfactory. Intuitively, the choate violations I picked out above do not quite capture the propositional content that is really the subject of privacy-related concerns (i.e. that I have Covid_19, that John’s father is a convicted murderer, etc.). So they seem like a band-aid solution to my normative woes. True, attempts manage to explain the wrongfulness of accessing and publishing falsehoods while tracking the presumption that privacy is factive. But they do nothing to justify this presumption: why should the falseness of a piece of information turn a choate privacy invasion into a mere attempt? My example of the false medical diagnosis in the previous section has put pressure on this point. As I argued above, to presume that the truth of the relevant information must make some normative difference (even the mere difference between inchoate and choate versions of the same offence) is to beg the question.Footnote ⁸⁵

To be sure, it would seem odd to view that any blatant fabulations or rumours would violate privacy,Footnote ⁸⁶ especially if these are made up out of thin air, without any relevant epistemic work (the digging or searching I mentioned above). I suggested earlier that such blatant falsehoods fit better into the wrong of defamation. Intuitively, when it comes to privacy, there has to be some concern with accessing or propagating things that are true about another personFootnote ⁸⁷ .

But it may be sufficient to require that beliefs that are capable of diminishing privacy are pragmatically or conventionally true; or perhaps just that they are warranted Footnote ⁸⁸ or rational. The requirement that beliefs can only diminish privacy if they are metaphysically true seems unnecessarily demanding. After all, when it comes to privacy invasions, it is not merely the content of the belief but also how Footnote ⁸⁹ one came to it which makes the normative difference.Footnote ⁹⁰ Beliefs that come about as a result of random speculation and fabulation typically lack reasons for belief.Footnote ⁹¹ Merely requiring that relevant false beliefs are warranted or rational – or, perhaps, that they are based on some authentic source – would not lead to an overly broad account of privacy diminishments, because this view only picks out cases where people had some relevant justification for forming their false belief. In these cases, people were at least somewhat concerned with, or effort put towards, getting at truth.

As Kappel and Le Morvan have argued, false information – at least in cases where believing it is justified – can make us equally socially vulnerable as true beliefs.Footnote ⁹² The moral prohibition against discrimination is a useful analogy: it extends to actual and perceived group membershipFootnote ⁹³ , so if a person is treated disadvantageously on the assumption that they belong to a protected religious minority, we count this as discrimination (rather than as merely attempted discrimination), even when the assumption about their relevant group membership turns out to be wrongFootnote ⁹⁴ . Including cases of perceived group membership within our understanding of discrimination tracks how discriminatory treatment victimises people.

Weakening the truth requirement has benefits with view to privacy diminishments that cannot neatly be represented in terms of propositional content, like sensory dataFootnote ⁹⁵ , accessing evaluative statements like ‘X is dumb and irresponsible’,Footnote ⁹⁶ or changes in doxastic states, such as suspending a belief. Consider the case of my romantic partner who believes that I love him. After reading my diary without permission, my partner becomes aware of the ambiguous feelings I harbour towards him. He thus tells me: ‘I don’t even know if you love me anymore’. He has here suspended his belief that I love him. Such a suspension of belief is not really propositionally true or false; at most, we can criticise someone who suspends their belief for ignoring the epistemic reasons they have for keeping a relevant belief. A wholly factive notion of privacy runs into difficulties in these domains (for this reason, scholars sometimes distinguish between different modes of losing privacy.Footnote ⁹⁷ Such distinctions complicate philosophical accounts of privacy. Weakening the truth condition may simplify things.)

Given all these virtues, why do some philosophers like Blaauw take it to be so incontestable that privacy is factive?Footnote ⁹⁸ When we look at parallel facticity debates – such as debates over whether expressions like ‘information’ is factiveFootnote ⁹⁹ , or whether verbs like ‘to ‘remember’, ‘to learn’, ‘to reveal’, and even ‘to know’Footnote ¹⁰⁰ are factive – we run into a similar ‘intuition stalemate’Footnote ¹⁰¹ . Allan Hazlett has suggested that divergences in intuitions on whether it is possible ‘to know’ a falsehood might be down to methodological differences within conceptual analysis. He argues that so-called ‘factive’ verbs, including ‘to know’, are not semantically factive (i.e. they can be used in non-factive ways in ordinary talk)Footnote ¹⁰² . But this, on Hazlett’s view, ought not worry those working in the ‘post-Gettier’ epistemology tradition; for they ‘are concerned with identifying necessary and sufficient conditions for knowledge attribution’Footnote ¹⁰³ , whereas those who employ a ‘linguistic method’ – like Hazlett himself – elicit intuitions ‘concerning whether or not a character in the story said something acceptable’Footnote ¹⁰⁴ . These approaches – the ‘traditional epistemological’ one; the ‘linguistic/ semantic’ one; and others, like value-oriented inquiryFootnote ¹⁰⁵ – represent distinct projects within conceptual analysis.

A similar point can be made regarding the philosophical debate over whether ‘information’ is factive. Proponents of the veridicality thesis (VT) maintain that information must be truthful (‘p counts as information only if p is true’Footnote ¹⁰⁶ ). Note that Floridi’s influential veridical conception of semantic information was formulated, at least in part, to address a particular problem that arises in relation to the objective of quantifying the informational content of sentences in formal languages (the Bar-Hillel Carnap paradoxFootnote ¹⁰⁷ ). VT is confined to the sub-category of declarative semantic information. It does not extend to other kinds of semantic content like stipulations, instructions, or invitations. These – as defenders of VT grant – ‘may be correctly qualified as kinds of information’Footnote ¹⁰⁸ , even if they lack truth/ truthfulness. Developing a concept of ‘information’ for these purposes is clearly distinct from the linguistic project of making sense of the term ‘information’ in ordinary language.

Linguistically, of course, term ‘information’ can clearly be used in a more generic, synecdochicFootnote ¹⁰⁹ sense. In ordinary language, we habitually allow that information can be ‘false’ without insisting that this undermines its ontological status of being ‘information’: anyone who has lived through the Covid_19 pandemic will agree that the expression ‘spreading false information’ is coherent and meaningful.Footnote ¹¹⁰ ‘False information’ is not the mere absence of information (i.e. non-information), but it contains relevant semantic content (which, as proponents of VT grant, might even be somewhat informativeFootnote ¹¹¹ ). As pointed out by Lundgren, ‘our judgment of whether something is information comes before, or at least is separate from, the judgment whether that something is true (truthful) or false’Footnote ¹¹² . At at least in ordinary language, it is permissible to conceive of ‘information’ as ‘alethically neutral’Footnote ¹¹³ . This illustrates that (some) disagreement over the factivity/veridicality of information may be a matter of conceptual analysis method.

To be sure, Hazlett’s claims about non-factive uses of verbs like ‘to know’ has faced challenge even within the ‘linguistic method’ he takes himself to be working inFootnote ¹¹⁴ . But this does not undermine his more fundamental point that differences in intuitions regarding the question of whether a concept is factive may ultimately depend on what exactly it is that analytical philosophers are doing when they engage in analysis. When I endorse the argument, as I did in the present section, that we should adopt a non-factive conception of privacy because we need privacy to protect us from the kinds of false beliefs that make us just as socially vulnerable as true beliefsFootnote ¹¹⁵ , I am pursuing what Hazlett calls a value-oriented inquiry and what others have termed ‘conceptual engineering’Footnote ¹¹⁶ or ‘conceptual ethics’Footnote ¹¹⁷ . I am, in other words, asking: What do we have the concept of privacy for? My implicit answer is that privacy is supposed to provide a normative bulwark against the unwelcome intrusions of othersFootnote ¹¹⁸ . This protective force is more capacious if it covers (intrusive) falsehoods. So non-factive privacy is better at tracking how intrusions victimize people. To be sure, in proposing this, I take myself to be doing something more modest and less creative than philosophers like Sally Halslanger, whose approach to conceptual analysis – also dubbed ‘ameliorative inquiry’Footnote ¹¹⁹ – proposes to focus our efforts not on capturing what we do mean, but on ‘how we might usefully revise what we mean for certain theoretical and political purposes’Footnote ¹²⁰ . Arguably, privacy is already ambiguous and non-factive uses find at least some support in the literatureFootnote ¹²¹ , so in making the case that the non-factive version is normatively preferable, I do not quite propose revisions to the concept of privacy. Still, I am suggesting that ‘moral considerations are amongst those that can be ought to bear’Footnote ¹²² on which of the available concepts of privacy we should pick.

Going back to Hazlett, I grant that those engaged in ‘traditional epistemology’ might not care for my normative arguments in favour of non-factive privacy. For those (perhaps Blaauw is amongst them) who take it to be analytically true that falsehoods cannot diminish privacy, arguments about the protective force of non-factive privacy will not move the needle. Those ‘traditional epistemologists’ might be content with the solution I offered in the previous section, i.e. the view that intrusive ways of accessing falsehoods can count as attempted privacy violations. But some (including this author) might lack antecedent commitments to the notion that privacy is factive. For us, weakening the truth requirement for privacy diminishments is an attractive alternative.