April 7, 2021

Is “cancel culture” real? Is Britain “institutionally racist”? These arguments, and many like them — about privilege, about “wokeness”, about whether trans women are women — never seem to end; they occupy some huge part of our national conversation.

And all the while, it feels like we’re arguing about something real, something that we could eventually resolve. But they’re not. All of these “culture war” debates are set up to fail: they turn on taking some phrase that sounds like it means something concrete, then change its definition so that we can never pin down an actual point of disagreement. 

It’s as though we’ve decided to argue at length over whether, say, tables are real — only for one participant in the debate to define a “table” as “a glowing green icosahedron, 30 metres on each side, levitating mysteriously above the Bay of Biscay”, and then explain why there is “zero evidence” tables exist. 

The recent row about the Sewell report into institutional racism in Britain is a case in point. A lot has been written wisely about it already, but I want to talk about words.

When used in normal conversation, the word “racism” usually means something like “prejudice against other races”. The first definition of racism in the Collins English Dictionary, for example, is: “The belief that races have distinct cultural characteristics determined by hereditary factors and that this endows some races with intrinsic superiority over others.”

Dictionaries are meant to reflect how language is actually used, and I think this definition broadly does: a person gets called racist if they believe that other races are inferior to their own. It seems to me there’s a bit more to it — you might not have to think in terms of “superiority” and “inferiority” in order to, say, stereotype Chinese people as inscrutable or Jewish people as greedy. I suspect a lot of English-speakers would include that behaviour in the definition of “racism”. Still, though, in common usage, the word “racism” is mainly about attitudes, beliefs and behaviours. 

But in its formal use, it’s not so clear. Following the Stephen Lawrence inquiry, three definitions of “institutional racism” were collated. The third one, from the Commission for Racial Equality, holds: “If racist consequences accrue to institutional laws, customs or practices, that institution is racist whether or not the individuals maintaining those practices have racial intentions.” 

Notice that this is completely different from Collins’s definition. There need be no intent whatsoever. If a law, custom or practice has racist consequences, then the institution which employs that law, custom or practice is institutionally racist.

In practice, it is completely incompatible with its normal use; you can’t use it about people at all. You couldn’t say “John is racist” — it would be meaningless, like saying “John has consequences which are negative for ethnic minorities”. In fact, it means that actions with any differences in outcome between ethnic groups become axiomatically racist. If it happens to be the case that tobacconists are disproportionately run by a particular ethnic minority, and if a law banning cigarette advertising hurts the sales from those tobacconists, then the legal system, by this definition, would be institutionally racist.

The Sewell report uses the first definition of the three I mentioned above, taken from the Macpherson inquiry into the death of Stephen Lawrence. The Macpherson definition is not as straightforwardly about consequences as the CRE definition, but it refers to a failure to “provide an appropriate and professional service” to people because of their ethnic origin, and says that this can be due to “unwitting prejudice” and “thoughtlessness” as well as active racial discrimination. 

The problem is that the word “racism” has too many meanings; it is a “suitcase word”, to borrow AI theorist Marvin Minsky’s term. You can pack a lot of things into it. That isn’t to say that the formal “institutional racism” definition is invalid. It’s simply that it’s so different from our usual definition of the word “racism” that using the same word is actively confusing.

That means that when we argue over it, we’re not arguing over a stable, mutually agreed thing. Is Britain racist? Well, obviously, there are disproportionate negative outcomes for several ethnic minorities, and by the CRE definition, that is axiomatically racist. But it’s not obvious (although it’s a hard thing to measure) that a majority of British people, or even British white people, hold racist views under a more colloquial definition. (It’s also worth noting that racist views have been becoming less common, which is a good thing.)

As Ian Leslie says, one of the key requirements for a productive argument is that the participants agree about what it is they’re arguing over. In this case, people who think it is can point at one definition; people who think it isn’t can use another. Even if they agreed to use the colloquial one, it’s still a movable feast: how many racist people does it take to make a racist country? One? Half of them? All of them?

The “racism” debate is, of course, far from the only place where this is a problem. We’re constantly arguing over whether “cancel culture” is real. But what is “cancel culture”? Is it the “phenomenon of promoting the ‘cancelling’ of people, brands and even shows and movies due to what some consider to be offensive or problematic remarks or ideologies,” as according to the New York Post. Or is it a “a mob mentality, a series of mass movements seeking to end the careers of public figures whose thoughts or opinions deviate from a new set of left-wing norms,” as offered by the New Statesman?

These are totally different things. The first describes individual incidents: if someone “promotes the ‘cancelling’” of someone, then that is an instance of cancel culture, and thus cancel culture is real. The second suggests that if one person is cancelled, that won’t meet the definition, but some unspecified larger number might. 

It means that you can switch effortlessly between the two, or — rather — argue about totally different things and feel like you’re arguing something real. Obviously, some people are “cancelled”, insofar as they lose their jobs as a result of social media outrage. The “phenomenon” is real. But is that enough to meet the “mob mentality/mass movement” definition? In the end, there’s nothing solid there to argue about: there’s no stable or universally agreed definition for any of these things. You can define the phrase however you like, and then argue about the definitions, and carry on doing that forever. No one will ever win. 

If I say that cancel culture is real, that tells you that I’m probably on one side of some big debate or other; if I say it’s not, that tells you I’m probably on the other. But it doesn’t tell you whether I think that, say, a given percentage of academics really have to censor their political views for fear of losing their jobs. All these concepts are so murky that it doesn’t tell you very much about my beliefs. It’s all mood affiliation and tribalism. 

How can we get around this? One way is to talk without using these highly charged, badly defined terms. “Taboo your words”: rather than ask “Is Britain a racist country?”, ask “Do ethnic minorities have worse outcomes than white people?” (Yes.) Or “Are black people less likely to be hired than equivalently qualified white people?” (Yes.) Rather than saying “Is cancel culture real?”, ask “How many people lose their jobs over social media outrage?” (Some, although I have no idea how widespread a problem it is.) Then you avoid the slippery definitions and vague mood-affiliation and can talk about real things.

Alternatively, you can simply define things however you want, and argue about them forever, in the great unending online row. But it won’t get us anywhere. The culture war rows will go on forever, because there is no right answer: we can simply change what we’re talking about, whenever we want. It’s a forever war; until we learn to say what we actually mean, there is no way out.