On Anonymity

After the England men’s football team lost to Italy on penalties on the 11^th of July, the three England players who missed penalties—Marcus Rashford, Bukayo Saka, and Jadon Sancho—were targeted with racist abuse on their social media accounts. Both events had an air of inevitability about them. The pattern of abuse reminds us that social media is a proper part of the real world: the monkey emojis posted to Bukayo Saka’s Instagram accounts are continuous with monkey chants and bananas thrown by spectators at matches, and were complemented by defaced murals in Darlington and Manchester. The reactions to these events have been a mixed bag: the mural in Manchester were covered with supportive messages and England flags, while the Greater Manchester Police said that they believed that the slogan scrawled over the mural in Manchester was “not of a racial nature”.

The failure of social media companies’ content moderation systems to deal with racist abuse leaves a space for newspaper columnists to resurrect the idea that the best way to deal with abuse on social media is get rid of online anonymity. Policies in this area have been floating around for a while, but seem to be building momentum. A petition to the UK parliament to require an ID check before opening a social media account started by the writer and businesswoman Katie Price currently has 687,000 signatures. The government’s response to this petition promises that upcoming Online Safety legislation “will address anonymous harmful activity.”

Debates about the benefits and costs of anonymity are complicated, and involve balancing the harms of anonymous abuse with the benefits of anonymity for minority groups. And, as Francesca Sobande points out in a quote on these proposals in the New Statesman, that it is a profound distortion to present racism as a social media problem while murals are being defaced, the police are denying that racist abuse is ‘of a racial nature’, and the Prime Minister has a history of racist newspaper columns.

Before we take on these questions, we need to take a step back and think about a more basic question: what exactly is anonymity?

Etymologically, anonymity is connected to the idea of namelessness, and with pseudonymity, which we might think of as the use of a name other than one’s ‘real’ name. At first pass, we might think that a person is anonymous in some context if they use no name, or use a name other than their real name. Ordinary language doesn’t appear to have a strong mandate about whether pseudonymity is a species of anonymity. This might make us think of anonymity as primarily a linguistic phenomenon about what name one uses.

But, as with England penalties, things quickly get messy.

First up, what is a ‘real’ name? I use lots of names in different contexts. Sometimes I use my first name, other times various shortenings of my first name, one or both of my surnames, and sometimes my full name with my middle name. I’m not cool enough to have a nickname, but lots of people are primarily known by a nickname, or are called different names by family, friends, or different communities (is a drag name a pseudonym?). Naming practices are, to put it bluntly, pretty chaotic. All of these names are directly referential devices which behave semantically like proper names, but is one of them my real name? The best candidate for a ‘real’ name is the name printed on government documents (and the names it contains), but in lots of cases this will not be the name that someone uses in their day-to-day life.

Secondly, it isn’t clear that the use of non-standard names is the only important sense of anonymity. The pseudonym ‘Elena Ferrante’ is rather different to ‘James S. A. Corey’. Ferrante’s identity is a closely guarded secret, whereas ‘Corey’ is an open collective name for the writers Daniel Abraham and Ty Franck. Since both authors are using non-standard names, in the linguistic sense both are anonymous, but we would want to say that there is another sense in which Ferrante is anonymous and Corey is not. These sense of anonymity that distinguishes Ferrante and Corey is epistemic rather than linguistic. There are a couple of ways we can think about this epistemic sense. Kathleen Wallace argues that we should think about anonymity as a gradable phenomenon concerning how many of someone’s traits can be coordinated. I think that this proposal is on the right lines, but we can give it a clearer gloss by thinking about knowing who someone is. If you know who someone is, they are not anonymous to you. (It’s not clear to me whether the absence of knowledge-who entails anonymity, or if there are extra conditions required).

One benefit of thinking about anonymity in terms of knowing-who is that we can get a grip on the messiness of claims about anonymity. Our talk about knowing who is notoriously context-sensitive. Knowing-who ascriptions often come with what Maria Aloni calls a method of identification. If we’re in the pub and you ask me whether I know who Justin Webb is, I’m likely to assent (he’s a well-known BBC radio 4 news host). If we were looking at a stack of photographs and you asked me if I knew who Webb was, I would say that I didn’t. Aloni proposes that relative to an auditory or descriptive mode of identification I know who Webb is, but relative to a visual mode of identification, I do not. This means that in lots of contexts it will be possible to know who people are, without knowing what their name is. In a context in which a descriptive mode of identification is salient, I can know who the man walking down the street is if I know that he is the only male presenter on the Today programs. Claims about anonymity also seem to be relative to modes of identification: walking down the street, Webb is anonymous in a way that a television presenter would not be, but talking on the radio he is not anonymous.

The distinction between the epistemic and linguistic senses of anonymity has two upshots for thinking about online accountability.

The first is that anonymity is not a single, or particularly stable phenomenon. There are at least two different senses of anonymity (the linguistic and epistemic senses). And neither of these is especially stable: the linguistic sense is unclear because we don’t have a good distinction between ‘real’ and ‘fake’ names, and the epistemic sense is extremely context-sensitive. Helen Nissenbaum argues that the morally important phenomenon is whether people are identifiable and points out that it is readily possible to identify people without using their names. This might point us toward the epistemic sense as being the core notion. With that said, names remain an important way of acquiring and communicating knowledge about other people, so we will probably want to hold on to the linguistic sense of anonymity.

The second point is that there isn’t a strong connection between anonymity and accountability. Using a ‘fake’ name is a way to control who knows what about you, not a magic ring that makes you completely unknowable. A long-term member of an online community who uses a stable screenname might be anonymous (in the linguistic sense, and relative to some modes of identification), but they will still have a reputation to uphold that incentivises them not to make false or harmful claims. Their screenname doesn’t make them unaccountable, it allows them to curate the community they are accountable to.

In Seeing Like a State, James C. Scott points out that stable names and surnames are an important piece of epistemic technology for the modern states. Without stable names, it would be impossible for a state to keep track of taxes, inheritance, criminal records, or demographic information. Bureaucratic needs have led to some remarkable naming exercises. Before colonisation by the Spanish, people who lived in the Philippines did not use surnames, and those who took surnames when baptised chose from a very small pool, making them unhelpful for discriminating between individuals. In 1849, the colonial governor Narciso Clavería y Zaldúa decreed that all subjects would have surnames, to be drawn from the Catálogo alfabético de apellidos. These names were divvied up by location, starting by allocating ‘A’ names to people from the provincial capital, and finishing the alphabet in the Island of Cataduanes.

Scott’s anarchist take on the social epistemology of names reminds us that the salient effect of instituting some kind of real name policy would be that our online lives would become even more easily surveyable by government institutions (much as they are already easily surveyable by advertising companies). At present, social media sites retain a modicum of the chaos of non-governmental naming practices, and we might worry that ‘real name’ policies replace chaotic naming practices which can accommodate a wide variety of needs by a regimented system which primarily serves the needs of bureaucracy.

I worry that between them, opponents and defenders of anonymity have constructed a false choice that doesn’t do justice to the complexity of the phenomenon. On the one hand, opponents of anonymity worry about Ring of Gyges-style chaos, and on the other defenders of anonymity raise concerns about the complete surveyability of our online lives by government. Part of the point of distinguishing the variety of phenomena which go under the heading of anonymity was to demonstrate that there are lots of different ways in which we can modulate who knows what about us—both by the medium of names, and by other means—and although it is tempting to fall into the authoritarian imaginary, there are lots of legitimate purposes for governments making citizens surveyable (for example to collect population health data). Things might be clearer if we disambiguated our debate into two questions, asking first what we should be able to know about one another, and secondly what the government should be able to know about us.