Subscriber OnlyInnovation

Truth is being written out of the internet equation

Web algorithms favour power of repeated lies when it comes to generating revenue

Click bait: Advertisers are not overly fussy about which   websites and Facebook pages their ads actually end up on
Click bait: Advertisers are not overly fussy about which websites and Facebook pages their ads actually end up on

‘A lie told often enough becomes the truth.” The internet reverberates with catchy populism. The more attention an article attracts, the more people promote it to their friends and community. The more an article is amplified, the higher credence given to it by web algorithms. Then the higher an article appears in search rankings and social network feeds, the more attention it receives. Positive feedback can frequently amplify a lie to become a convenient truth.

Such popular hit pieces can generate a lot of money. Channel 4 broadcast interviews on November 24th last with teenagers from Veles, in Macedonia. Each is earning up to €200,000, or even more, from advertising associated with articles they have published. How can this be?

Well most publishers of online material notify Google and Facebook of empty slots in their content – these are placeholders for adverts. Advertisers pay Google and Facebook to insert their ads into these slots, alongside content that matches particular keywords associated with each advert. Advertisers also specify target demographics and local geographies for their ads. The advertisers know that only a few readers will be interested. But with millions of readers, having just a few per cent clicking through on the ads is still significant. Within acceptable limits, advertisers are not overly fussy about which specific websites and Facebook pages their ads actually end up on. They key metric is a steady stream of clicks to the ads.

So how then are these Macedonian teenagers making quite so much money? Actually, it is surprisingly easy. Design a website or Facebook page formatted to emulate global news brands. Sprinkle in a few outrageous headlines and pieces, provided you are unfazed by any defamation laws. Something like "Michelle Obama is a Man"; "Pope Francis Endorses Donald Trump"; "CNN Promotes Fake News about Mike Pence" – you get the idea. Then sign up to Google and Facebook and tell them you are willing to be paid for any adverts they want to insert into empty slots alongside your piece. Then count the euros hitting your Paypal account as your fake news goes mainstream. Freedom of speech can indeed trump truth.

READ MORE

Revising history

There are a few tricks of the trade. Various heuristics can fool the algorithms to promote your content ahead of others. If you become proficient you can even persuade Google to revise world history. The Guardian newspaper has published a number of recent articles illustrating how Google results can assert racial hatred, because certain right-wing groups have been more tactically exploitative of Google's algorithms than politically correct opponents.

How did Google and Facebook in particular get it so wrong? Google used to have a motto of “Don’t be evil” to encourage staff to abide by virtuous corporate values. After some 16 years, it dropped that maxim last year. In a letter alongside its 2004 flotation prospectus, Google asserted that “Google users trust our systems to help them make important decisions”.

Yes, indeed: this year, many users trusted Google to help them decide which candidate to vote for in the US presidential elections by prioritising which headlines they should read.

The 2004 letter went on: “Our results are the best we know how to produce.” But now in 2016, those results can be blatantly misleading. In 2004, Google asserted its results are “unbiased and objective, and we do not accept payment for them”. But now instead, Google shares advertising revenue generated by deliberately biased and unobjective results specifically written – by the aforementioned cohort of Macedonian teenagers among others – to attract readers.

Google and Facebook have grown as a consequence of their employees developing algorithms to rank content. Both companies have sought to minimise human curation, instead favouring automated approaches to efficiently process the vast volume of material published each day. Whatever truth or crassness such content may or may not contain appears to be entirely irrelevant. What appears to matter above anything else is generating revenue. This is achieved by prompting as many advertising clicks and read-throughs as possible, regardless. Regardless of the damage to the advertising income of professional news organisations and public service broadcasting. Regardless of the truth.

Business models

The internet is still in its infancy. Many of its algorithms and business models are frankly asinine. Automated content analysis is but the first step to the invention of algorithmic fact verification and editorial precision. A minority viewpoint, however factually incorrect, need not necessarily be suppressed, but neither should it be asserted as an undeniable truth to generate clicks and corresponding revenue. Current digital advertising business models carpet bomb millions of internet users with irrelevant and untimely ads, hoping that at least a portion will read them. Precision advertising, with near 100 per cent success rates, has yet to be invented.

The astute observation on the power of repeated lies, attributed to Russian revolutionary Vladimir Lenin but in fact first published in 1869, carried a a prescient warning for the internet age. It is both astonishing how little effort can now leverage an outsized reaction, and that doing so can be financially lucrative. Has there ever been a case in history when a society made itself vulnerable by creating a technology which pays amateurs and enables more sinister forces to both undermine its own cohesion?