Stop talking about fake news!

ABSTRACT

Since 2016, there has been an explosion of academic work that fixes its subject matter using the terms ‘fake news’ and ‘post-truth’. In this paper, I argue that this terminology is not up to scratch, and that academics and journalists ought to completely stop using the terms ‘fake news’ and ‘post-truth’. I set out three arguments for abandonment. First, that ‘fake news’ and ‘post-truth’ do not have stable public meanings, entailing that they are either nonsense, context-sensitive, or contested. Secondly, that these terms are unnecessary, because we already have a rich vocabulary for thinking about epistemic dysfunction. Thirdly, I observe that ‘fake news’ and ‘post-truth’ have propagandistic uses, meaning that using these terms legitimates anti-democratic propaganda, and risks smuggling bad ideology into conversations.

KEYWORDS:

• Fake news
• post-truth
• philosophy of language
• nonsense
• context-sensitivity
• metalinguistic negotiation
• propaganda
• applied philosophy of language

ORCID

Joshua Habgood-Coote http://orcid.org/0000-0003-3427-3325

Notes

1 For other advocates of abandonment, see (Oremus 2016, 2017; boyd 2017; Jack 2017; Sullivan 2017; Staines 2018; Talisse 2018; Wardle 2017; Zuckerman 2017), (Finlayson MS), (Coady MS).

2 I will use ‘ideology’ as a neutral term, meaning that there can be good and bad ideology (see section 3).

3 Mentioning quotes refer to a word (as in ‘the word ‘dog’ has three letters’), whereas scare quotes are used to express skepticism, or non-standard usage (as in ‘I believe that ’national character’ does not explain football results’).

4 For example, expressivist theories of meaning may think that expressive content is the meaning of a term, and that expressive content is explanatorily prior to descriptive content.

5 ‘Fake news’ and ‘post-truth’ occur both as adjectives and as noun phrases. Consider:

• The Froome story is fake news.

• We are living in a post-truth era.

• Fake news is a huge political problem.

• Post-truth is a complex phenomenon.

I will assume that both ‘fake news’ and ‘post-truth’ are primarily descriptive phrases (as in (1) and (2)), and that uses as noun phrases are derivative.

6 Things get much more complicated with names. For some, the meaning of a name is just its extension and descriptive content is relegated to the pragmatics (Salmon 2007), for others names have both extensions and descriptive content which matter in different contexts (Frege 1892), and for others the descriptive content of a name determines its extension (Russell 1905).

7 See, (Hunt 2016) Craig Silverman does the best job of constructing a robust set of cases, compiling listicles of 50 stories they class as 'fake news' in 2016 and 2017. However, I think it is better to think of this list as advocating for a definition rather than providing a neutral list. (Silverman 2016 ; Silverman, Lytvynenko, and Pham 2017).

8 (Gomes 2017).

9 (Waterson 2018).

10 Michael Lynch aptly calls this phenomenon the internet shell game (Lynch 2016).

11 These observations might motivate a purely expressivist treatment of ‘fake news’ (perhaps meaning something like I don’t like this story). This treatment faces some problems. An expressive semantics would have a hard time making sense of the possibility of disagreements about whether a story falls under ‘fake news’: if one person likes it, and another doesn’t, there is no room for disagreement. This proposal would also require a solution to the Frege-Geach problem, since ‘fake news’ does not have its expressive meaning in logically complex expressions (consider ‘if Prince Harry and Meghan Markle didn’t marry, then lots of stories were fake news’). I prefer to think of ‘fake news’ as a putatively descriptive term with expressive force.

12 I owe this term to (Coady MS).

13 Davies defines bullshit as "any form of communication – verbal or non-verbal – that is not the clearest or most succinct statement of the sincere and reasonably held beliefs of the communicator." (Davies 2017, xx). This definition leaves much to be desired: as stated it includes mis-speaking, sarcasm, putting forward a tentative hypothesis, and being verbose.

14 That said, some of these definitions seem problematic on their own terms. How are we supposed to measure the influence of facts and opinion on public opinion? Are we to count beliefs from different sources, or do some belief matter more? What is it for truth not to exist? And, for that matter, what is it to not believe in (the existence of) truth?

15 It’s an interesting question whether all the co-authors on this paper would agree on the extension of ‘fake news’.

16 (Farkas and Schou 2017).

17 One might think all terms of natural language lack determinate meaning (Carnap 1937), are context-sensitive (Travis 1997), or are up for negotiation (Ludlow 2014). One might also think that all political terms have one or other of these linguistic defects (it is easy to come up with examples: consider ‘progressive’, ‘neoliberal’, and ‘conservative’). I hope that someone who thinks that these linguistic defects are general will either admit that there is some special linguistic problem with ‘fake news’ and ‘post-truth’, or will be persuaded by the arguments in section 3.

18 How can speakers stipulate the meaning of a nonsense term? We might think that nonsense terms that do not contribute to sentence meaning can still contribute to speaker meaning if the speaker’s communicative intention is clear (Grice 1957).

19 Do we need any epistemic slur terms? On the one hand, we might think that epistemic slurs are a useful shorthand for dismissing obvious falsehoods. On the other hand, the difficulty of ascertaining the truth, combined with group-level benefits from epistemic diversity (Zollman 2010) should make us extremely tolerant of differences of opinion.

20 See (Jack 2017; Wardle and Derakhshan 2017).

21 One might think that reaching complete abandonment would be just as hard. I agree: as noted in the introduction, the primary goal of this paper is to persuade academics and journalists to stop using ‘fake news’ and ‘post-truth’.

22 I take it that propaganda that aims to repair the conditions for open and rational public discourse is still manipulative (Stanley 2015, C3).

23 One might worry that ‘propaganda’ can easily be stripped of its descriptive content through use as an epistemic slur. I hope to assuage this worry by associating the term with a clear descriptive content, and by distinguishing good and bad propaganda.

24 For example, consider Rwandan radio presenters’ use of ‘inyenzi’ (cockroach) and ‘inzoka’ (snake) to refer to the Tutsi in the period before the 1994 genocide (Tirrell 2012). Dehumanisation is a persistent theme in violence-inciting propaganda (Smith 2011). On May 21^st 2018 the White House website published an article entitled ‘What you need to know about the violent animals of MS-13’. (Whitehouse 2018).

25 For example, Trump’s second amendment comments, which gave his supporters permission to attack Clinton (Corasaniti and Haberman 2016).

26 For example, consider the way that ‘welfare’ can cue up thoughts like ‘Blacks are lazy’ in the US context (Stanley 2015, 123–4), or the way a generic like ‘Muslims are terrorists’ can introduce an essentialising generalisation about the group targeted (Leslie 2017)).

27 (Reporters without Borders 2017 ; Head 2018 ; Olewe 2018).

28 (Beech 2017; Isikoff 2017; Sheperd 2017).

29 (The Economist 2016).

In general, I think we should be very careful about making parallels with 1930s Germany, but in this case the analogy seems justified.

30 I don’t think this term has a stable referent, but it is difficult to report the content of this ideology without it.

31 For the acceptable content of this ideology, see the PragerU primer What is Fake News? (Klavan 2017).

32 I don’t want to suggest that establishment media has impeccable epistemic credentials, just that they are closer to the ideals than alternative news sites.

33 A nice example of this phenomenon is Stefan Molyneux’s The Art of the Argument: Western Civilisation’s Last Stand, which is presented as a logic textbook, but functions to shore up patriarchal white nationalist ideology (Douglas 2017), whilst displaying basic errors about logic concepts (Chartier 2017).

34 (Guess, Nyhan, and Reifler 2018).

35 It is not difficult to see resources like the Media Bias Chart as a kind of epistemic policing.

36 ‘Centrist dad’ refers to middle-aged men who are putatively left-wing, but have failed to come to terms with political changes, leading to resistance to radical ideas. I intend the term in both pejorative and descriptive senses.

37 D’Ancona gives Winston Churchill as an example. One might think this choice of political figure is indicative of a wider project of rehabilitating problematic values via supposed threats to democracy.

38 Thanks to Koshka Duff for this phrase.

39 This idea is a theme in (Lakoff 2014). See (Lakoff and Duran 2018).

40 Thanks to Peter Allen, Zara Bain, Sebastian Becker, Thomas Brouwer, David Coady, Jennifer Corns, Koshka Duff, Lunette Elle Warren, Adam Ferner, Noah Friedman-Biglin, Karen Frost-Arnold, Axel Gelfert, Michael Lynch, Matthew McKeever, Claire Moriarty, Jakob Ohlhorst, Richard Pettigrew, Mark Pinder, Kim Pedersen, Orestis Palermos, Lauren Spence, Johannes Steizinger, and Alessandra Tanesini.