Search

Tuesday 13 December 2016

Parental Advisory

Though they are routinely yoked together as two of the horsemen of the post-truth Apocalypse, the leave campaign in the UK's EU referendum and Donald Trump's campaign in the US Presidential election were very different in terms of the means they used to circulate "fake news". In the UK, the standout moments in the propaganda war involved media that have been around for over a century: newspaper headlines ("Queen backs Brexit", varieties of "millions more migrants on the way"), billboard posters ("Breaking Point"), and even a big red bus assuring us that £350 million could be diverted weekly to the NHS. In the US, attention focused on the new media of the Internet, notably Trump's Twitter account, Facebook's inadvertent creation of neo-Nazi safe spaces, and (though their importance has been over-stated) out-and-out black propaganda sites like Breitbart News. This led to the odd sight of US cable TV, which was once the new kid on the media block, misjudging the popular mood and being artlessly played by Trump.

The sense of the two countries inhabiting different centuries - possibly the 19th and the 21st - was obscured by the commonality of the "mainstream media" doing a poor job in countering the nonsense or interrogating (rather than witlessly amplifying) popular "concerns". This might not appear to reflect particularly well on the UK, but it actually points to a greater resilience, or at least inertia, in the institutions of British politics. We still have a career politician in Number 10, UKIP aspires to be a party just like the others, and the judiciary doesn't seem minded to rubber-stamp the will of the executive. While institutional rot affects both countries, it looks to have spread further in the US, accelerated by the removal of restraints on political financing and the strategic vandalism of the Republican Party in Congress since the mid-90s.

This cultural difference goes some way to explain why the Guardian Media Group (GMG) appears to be far more worried about the impact of the Internet than most other UK news organisations. Having consciously set out to be the global liberal's trusted source, it has invested heavily in both online and the US, and is acutely aware of the power of the Internet giants to dominate advertising budgets. It has become simultaneously more American (and thus prudish) in style and more fretful about market imperfections. This has led to an increasingly moralistic critique of both individuals (e.g. the prominence given to online harassment and the defensive focus on identity politics) and organisations deemed to be eroding journalistic values. This often produces hyperbole. For example, Facebook's filter bubble is presented as tantamount to a system of mind control when it is essentially just confirmation bias.

Despite the high profile given to its ability to amplify dodgy news, Facebook's chief feature is that it reinforces existing mainstream media preferences through sorting. If you're a teacher in Hackney whose mates are mostly teachers, you'll see plenty of Grauniad articles. This makes it a poor channel to grow readership beyond already saturated groups. In contrast, Google can provide much wider exposure, particularly if you can appear in "natural" search results rather than via Ad Words. While adverts of any sort have a low return, the benefit to a news provider of appearing in a self-directed search is considerable. If it is going to grow its global audience, GMG probably needs to rely more on Google than on Facebook, but the lack of predictability inherent in search produces anxiety: "New platforms have put a bomb under the financial model – advertising – resources are shrinking, traffic is increasingly dependent on them, and publishers have no access, no insight at all, into what these platforms are doing in their headquarters, their labs".


That statement featured in what appears to be a planned campaign, launched in The Observer this month by Carole Cadwalladr,  to warn us that "tech-savvy rightwingers" are "gaming" the Google algorithm. The gaming of Google for political rather than commercial ends is not new. The Google bomb dates from the late-90s, and was made famous by the association of "miserable failure" with George W Bush, while the autocomplete function of Google Search has been around since 2004. For years Google has been editing both search rankings and autocomplete predictions to remove objectionable results, though it keeps this quiet because the system's credibility depends on its comprehensiveness - i.e. reflecting what users in aggregate are searching for and which pages the Internet as a community of content-providers considers most authoritative on any given topic.

Cadwalladr's article conflates two separate issues: autocomplete and search result rankings, though this is forgivable since Google launched Instant Search in 2010, which presents cached results based on predicted search terms. Autocomplete tells you what other people are searching for. If you enter "are jews" and this produces "are jews [evil]", then that is a reflection of actual usage (which is also changing, so don't expect the same results each time), not of gaming. Neither does it reflect an upsurge in antisemitism. The word "Jews" is likely to produce negative stereotypes in autocomplete predictions because while antisemites tend to bandy it about, most people tend to be sensitive to the charge of inadvertent antisemitism and are therefore circumspect in using it. To put this in context, consider some leading autocomplete predictions for a variety of other groups: are the french [rude]; are lawyers [rich]; are man utd fans [glory hunters]; are the japanese [a cruel race] (autocomplete is also localised, so this may reflect heavy usage by Daily Mail readers in my neck of the woods).

Another example reported in The Observer a week later was "did the hol[ocaust really happen]". It is easy to mistake the salience of certain terms in autocomplete as evidence of a conspiracy, but the mundane truth is that predictability largely depends on the probability of alternatives. "Did the hol[e in the wall gang really exist]" or did the "hol[iday inn invent holidays]" are questions that are rarely asked. "Did the hol" is not a commonly used prefix, even among members of the Hollies fan club. This means that the number of searches required to feature in the autocomplete top four may be quite small, certainly compared to something like "the cheapest price for". It is also likely that many of the (possibly few) people who entered the predicted search are already primed to question the reality of the Holocaust. We regularly use search engines to validate existing beliefs or confirm prejudices, not just to fill in the gaps in our knowledge, hence the idea of Google as a "prosthetic memory".

While there are plenty of ignorant people in the world, it would be wrong to assume that they are empty vessels that will be filled by the first propaganda they encounter. That is a reactionary prejudice with a long history, combining a contempt for the common sort and the demonisation of malign forces set on undermining our cherished social order. To put this in the terms of an earlier technology, it is like the debate over whether public libraries should stock copies of Mein Kampf, or whether certain "dangerous" literature should be held under lock and key and only be made available to responsible persons bearing a letter signed by a bishop. Significantly, Wikipedia, as an edited content platform, is not seen as a problem, just as the Encyclopaedia Britannica wasn't, though a case can be made that both are riddled with errors and dubious editorial judgements. Of course, Wikipedia is not in the business of providing news, so it is not a competitor in the eyes of media organisations like GMG.


In demanding that certain search terms be curated the liberal media is exercising its own bias. Nobody seems too bothered that the term "was Hitler" produces: a socialist; married; German; a dictator (the last of these leads to a result set that uniformly confirms that he was and that this was a bad thing). The insistence on curation goes beyond autocomplete predictions to actual search results, which is the main prize for the likes of GMG. The top-ranking result for "did the holocaust happen" is a comment thread on the site of Stormfront, a neo-Nazi group, in which a convinced denier recycles various myths. The reason for its prominence is the form of words used in the search. If Hannah Arendt had chosen that phrase as her title instead of Eichmann in Jerusalem, the first result today would probably link to her book on Amazon, no matter how much "gaming" white supremacists got up to. If you search for just "holocaust", you'll get Wikipedia first. On a lighter note, the first page of results for "are jews evil" currently includes articles about the Observer story and Google's response. Irritatingly for all at GMG, the top link is to a report in The Telegraph.

Cadwalladr makes the relationship between the Internet and democracy explicit, casting Google in the role of a privileged interest that must be restrained by the people: "Are Jews evil? How do you want that question answered? This is our internet. Not Google’s. Not Facebook’s. Not rightwing propagandists. And we’re the only ones who can reclaim it". Though this sounds like a clarion call to liberty and the defence of the commons, what GMG is demanding is that Google be obliged to exercise editorial control. Their practical hope is that this will give greater prominence to traditional media businesses in search results, while their ideological aim is to reinforce the gatekeeper model central to the liberal notion of civic responsibility. I'm not one for giving Google a free ride - their claim that they merely reflect the wisdom (or stupidity) of the crowd is disingenuous - but the value of a near-universal search engine with minimal editorial direction far outweighs the downside of a bunch of muppets at Stormfront denying the Holocaust.

There is a lack of logic in insisting on editorial control without addressing the question of ownership, which is particularly obvious in a week when Rupert Murdoch revived his bid for Sky plc (something you can be sure he wouldn't be doing unless he had already squared the government). The nutter who recently turned up with a gun in a Washington pizza parlour looking for Hillary Clinton's child sex ring was primed not just by far-right websites but by the longstanding virulence displayed towards both Clinton and paedophiles by cable TV, radio and tabloid newspapers. The press coverage of paedophilia changed in the 1980s, with Murdoch's outlets very much to the fore, with the trope of an organised conspiracy to target children overtaking the traditional image of the seedy loner. This was fed not just by "ritual satanic abuse" panics and government campaigns against the "promotion of homosexuality" in schools (both in the context of an assault on the public sector), but by bonkers claims that drug dealers thought kids a more lucrative market than adults ("just say no") and rock groups were urging teenage suicide in backmasked recordings.

At times, Cadwalladr sounds like Tipper Gore, whose objection to Prince's Little Nikki famously led to the introduction of Parental Advisory stickers. Consider the emotional register of the following: "The right is on the rise everywhere. And that includes on the internet. It is creating more content that is travelling wider and further. It has changed both the questions being asked – did the Holocaust actually happen? Are Jews evil? Should Islam be destroyed – and answered. It is in the process of remaking the world, rewriting history, rewiring minds, changing the conversation, reframing the questions and answers. It’s our world. Our internet. Our history. And we have to wake up to what is happening right now on the laptop on our desk, the phone in our pocket, the tablet in our children’s bedrooms. This is our choice: do something. Or accept the truth according to Google. That six million didn’t die. That the Holocaust never happened. That we didn’t care enough to remember."


No comments:

Post a Comment