Buying Brexit: the politics of Google ads & the general election

By Pip Thornton

I have been doing some work about the Tories and Labour buying Google adverts for ‘dementia tax’ in the run up to the General election. It is fascinating to me as it shows so clearly what I am trying to critique in my work about the politics and economics of the search engine, and the neoliberalistion of language and discourse through Google’s AdWords marketplace. The potential tainting of the search engine with advertising is of course something Google’s creators Sergey Brin and Larry Page anticipated and indeed initially tried to avoid. This is evident in an appendix to their 1998 paper, on The anatomy of a large-scale hypertextual Web search engine, where they noted that ‘advertising funded search engines will be inherently biased towards the advertisers and away from the needs of the consumers’. Last week’s debate about the use of paid Google adverts by all…

View original post 1,120 more words

How the Tories wrote my thesis: the political economy of a large-scale hypertextual Web search engine

brinpageScreenshot 2017-05-28 00.40.45

Sergey Brin and Larry Page invented Google as students at Stanford in 1998. They knew from the beginning how advertising could interfere with the efficiency and integrity of their proposed search engine. In an appendix to their paper, on The anatomy of a large-scale hypertextual Web search engine, they noted that ‘the goals of the advertising business model do not always correspond to providing quality search to users’, and that ‘advertising funded search engines will be inherently biased towards the advertisers and away from the needs of the consumers’. They weren’t wrong, yet they had to fund it somehow, and after nearly two decades of conflict between advertising dollars and organising the world’s information (as per Google’s mission statement), a typical search engine results page (SERP) today is surely unrecognisable from what Brin and Page could ever have imagined, infused as they have now become with the neoliberal logic of a linguistic marketplace. There is of course an ever-expanding literature critiquing the political, economic and epistemic implications of search engines, initiated in part by Introna & Nissenbaum back in 2006. This body of scholarship is one which I aspire to contribute towards in my final thesis. In the meantime, I saw a search result this week that pretty much sums up that thesis, and I think I might have Theresa May to thank for that.

I have been researching Google Search and its various distortions for a few years now, but until last week I have never come across an example that has so crystallised the potential political implications and social consequences of a monetised and market-dominating ‘large-scale hypertextual Web search engine’. This example brings out the many powerful themes present within my research; organic vs paid SEO, linguistic capitalism, digital democracy, neoliberalism, power, discourse, politics, money, social inequality, digital economics, philosophy, epistemology, alternative facts and more…

THIS is the search result I am tempted to submit in lieu of my PhD thesis:

3partyScreenshot 2017-05-22 14.19.22

The background

On Monday morning (22nd May 2017), Google search users first started noticing that the Conservatives had taken out an advert with Google about the dementia tax. Despite dismissing the term ‘dementia tax’ as a ‘so-called’ phrase in the hyperlinked advertising copy, the Tory media machine apparently had no problem embracing, harnessing and indeed purchasing the phrase in order to exploit its commercial capabilities on Google’s AdWords platform. As a Tory spokesman said:  ‘It is quite right we take steps to tackle the misinformation and fear being spread by Jeremy Corbyn and the Labour Party’.

The first I heard about the whole thing was when a friend and colleague Andrew Dwyer suggested on Twitter that if the Tories were paying Google per click (PPC, or Pay Per Click is how the AdWords system usually works) in order to control and dominate the narrative around their controversial policy, then we should all be continuously searching for ‘dementia tax’, and clicking on the link the Conservatives had paid for. The AdWords platform is the original and fundamental source of the wealth and power Google enjoys today, but advertising this way can be expensive – and ultimately unsustainable – if your advert attracts clicks that you are unable to convert into sales. So hacking the system by clicking on the Tory sponsored dementia tax advert could theoretically cost Theresa May’s campaign dearly. The process is in fact much more complicated than that, however. For example, the more successful click-throughs an advert generates, the higher its ‘quality score’ becomes, thus driving down the cost of each click.

Although AdWords is based on an auction model, quality scoring and other algorithmic ranking factors also help to determine which bids ‘win’ the top spots on the search page, and it is not necessarily the highest bid that comes out top. Advertisers can also buy bundles of ad placements for a fixed price (called PPM – Pay per Impression). This would normally be used for customers who just want their brand or message to have more exposure, and do not necessarily need people to click-through. In addition to this, Google has systems in place to detect and counter apparent click fraud, whether automated or part of a physical campaign, and there are numerous independent anti-click fraud companies too. Google AdWords is a very complicated and confusing economy, which is why an enormous multi-million dollar Search Engine Optimisation (SEO) industry has grown up to sustain and perpetuate it.

But despite this, the potential financial implications of the Tory’s advert (if the clicking went viral and unchecked, for example) along with a nagging scepticism about the possible manipulation and loop-holing of free ‘charity status’ Google advertising through the AdGrants scheme, initially made me think that the overtly political adverts for ‘dementia tax’ had somehow bypassed the standard commercial (and ostensibly market driven) AdWords procedure. As I have written about elsewhere, it used to be the case that the free AdWords ‘donated’ to Not-For-Profits (NFP) through the AdGrants scheme could not be used for political or religious purposes, but Google dropped that caveat when it started allowing NFP groups to take part in the ‘Redirect Method’. Born out of criticism from government over the spread of extremist literature and sites through platforms such as Google search, The Redirect Method is a scheme spearheaded by Google’s innovation centre Jigsaw (but the method is open to anyone with the right charity status) which uses free AdWords to ‘buy up’ keywords and phrases which might be used by would-be extremists to search for sites, videos, forums or manuals relating to terrorism. Keywords such as ‘Join ISIS’, or ‘Jihad’, would be simple examples, although the whole list is far more detailed and nuanced than that. Instead of being returned a page of ‘organic’ results relating to their search terms, the would-be jihadist instead sees adverts at the top of the page which link to specially curated sites and YouTube footage which aim ‘to confront online radicalization’ by ‘redirecting’ the user to alternative sites. As its brochure states, The Redirect Method:

focuses on the slice of ISIS’ audience that is most susceptible to its messaging, and redirects them towards curated YouTube videos debunking ISIS recruiting themes. This open methodology was developed from interviews with ISIS defectors, respects users’ privacy and can be deployed to tackle other types of violent recruiting discourses online

It is a similar narrative-altering motive and methodology to the Tories’ rationale for using AdWords in order to ‘tackle the misinformation and fear being spread by Jeremy Corbyn and the Labour Party’, which is why my cynical mind wouldn’t have been surprised if the dementia tax adverts turned out to be exploiting some charity status loophole to obtain such a dominant position on Google, or indeed had been initiated through a non-official channel. This scenario would of course have meant that Google was not profiting from click-throughs, although the company can and does use these ‘in kind’ charity donations against their tax bill. On a more practical level, AdGrants accounts take many weeks to set up – and almost certainly longer than the period of time it took for the phrase ‘dementia tax’ to gain enough political traction (thanks in part to a Financial Times front page) to warrant the Tory’s need to ‘tackle’ it.

Winners and losers in Linguistic Capitalism

I have been following the adverts for several days now, and am fairly certain that it is plain old ‘linguistic capitalism’, rather than any hidden hand, that is mediating the current narrative of the ‘dementia tax’. But this is not to say that the current politicisation of the word ‘dementia’ has not had an impact on its use in a social environment. On the contrary, shortly after the Conservative advert on Monday morning, a Labour one appeared, and then some time after that, a Liberal Democrat one. (NB – the screenshot above shows the paid-for Tory and Labour adverts, and an ‘organic’ Lib Dem one… more on that later…). And it was not just political parties getting in on the act – privately funded adverts began to appear too, all of them presumably bidding for the keyphrase ‘dementia tax’. One anonymous advert links to www.dementia.tax, a site created last Monday (22nd May 2017) ‘by a voter that really doesn’t want mayhem in power’. As I mentioned before, Google AdWords is a strange and opaque marketplace, but whatever its distortions, the bidding wars going on for ‘dementia tax’ this week have presumably significantly driven up the price of the words. This is potentially great news for Google, who get paid the winning bid price on the pay-per-click system (which is in effect 1 cent more than the second highest bid), but it is less good news for the dementia charities who rely on free AdGrants words (which are capped at $2 a click), and who may in effect have been priced out of the market.

I do not know the full amount the Tories, Labour, the Lib Dems or the independents are paying Google for their political exposure, and neither do I know for definite that charities such as Alzheimer’s Society, MindCare and Carers Trust receive free AdGrants advertising. Google’s data and systems are of course proprietary and opaque. What I do know is that according to the Google Keyword Planner (KWP), which suggests appropriate bid prices of words and phrases to advertisers entering the AdWords auction, the phrase ‘dementia tax’ has been valued at zero from 22nd May 2017 to time of writing (29th May 2017). From work I have done on the KWP in the past, however, suggested bid prices – especially those for words or phrases fairly new to the market – take a while to start reflecting changes. I saw this when the suggested bid price for the word ‘Chilcot’ rose from zero to £1.86 the week after the Iraq Inquiry was released in 2016, which I suppose is both a testament to the ingenuity of the SEO industry, as well as a depressing confirmation of the postmodern condition (work forthcoming). I will be keeping an eye on the KWP price data on dementia/dementia tax over the next few weeks. Given what I found in the historic data on the price of ‘brexit’ (see below), I suspect any jump in price may already have occurred, although that would not explain why the suggested price for ‘dementia tax’ is still zero.

Suggested bid prices are of course not necessarily indicative of the actual PPC, and access to that data is only obtainable through a funded campaign. For many reasons, I am reluctant to start getting involved in fake Google advertising for the purposes of research, even if if would afford me better data, especially when the arena is both so political and potentially detrimental to charity organisations. If I can find a way of doing so unproblematically then I will (I have tried contacting the dementia charities concerned to see if they might share their AdWord data), but until then, my own methodology is to use suggested bid prices and an artistic intervention called {poem}.py to critique the system.

Of course, it is important to remember that the coveted phrase here is ‘dementia tax’, and not just the word ‘dementia’, which when searched still returns dementia charity results, and has only risen in price on the KWP by a penny over the course of this week. But there is a significant ‘bleed’ in the way Google monetises language. If the phrase ‘dementia tax’ is not enclosed in inverted commas in the search bar as a ‘verbatim’ search, then the component words become triggers in their own right. This is easiest to see if you search a phrase (without inverted commas) on Google. If there is no discrete match (or at least if none is selected by the algorithm), then the snippets of text under each result will highlight in bold the word that has triggered the hit. Thus the advert below for a dementia charity is served on the search term ‘dementia tax’, despite the word ‘tax’ not appearing in the ad:

Screenshot 2017-05-28 18.56.31

What is really both interesting and worrying is that this advert for Alzheimer’s Society appears alongside other dementia charities at the bottom of the second page of the search results, which in eyeball terms is pretty much oblivion. While a Huffington Post article speculated that ‘the best place to hide a dead body is page two of Google’, there have also been other more rigorous studies into browsing habits that show that people rarely scroll beyond the first page. The adverts for the 3 main political parties, however (and the Lib Dems had taken out a paid ad out at the time of this screenshot) have pride of place at the top of the first page. This economic and political hierarchy of results not only takes up valuable real estate space from the organic results, but reveals the real-time effects this politicisation of AdWords might be having.

Returning to the original thesis-writing screenshot (thanks Theresa)… I hope to have illustrated how all the issues I have discussed in this post are implicit in that one search result (and I haven’t even begun to decipher the whole ‘alternative fact’ elements of the actual content of the ads). It tells some many stories. As well as what I have already discussed, the screengrab was taken from Page 2 of the search results. So not only does it reconfirm the Tories’ bigger spending power over the Labour Party (and that both parties have budgets big enough to reach beyond the first SERP), but that the organic Liberal Democrat result didn’t show up on the first page at all – it had been pushed to the second page by the paid ads. When only the Tory advert showed up at the top of Page 1, the organic Lib Dem result took up the last slot on the first SERP. Much the same seems to be happening to charities and NFPs who cannot compete against well-funded political campaigns. Political parties have of course always paid for advertising, but this new way of harnessing linguistic capitalism through Google AdWords speaks volumes not only about the state of digital democracy; a new fusion of politics and proprietary technology with strong and far-reaching collateral effects, but also about the nuances of individual campaigns and the political economy (and indeed the anatomy) of the search engine itself.

twit

The Death of the Reader | UNDER THE INFLUENCE MAGAZINE

My essay The Death of the Reader has been published in Under the Influence magazine. It’s a special issue on the subject of SPEED, so the essay reflects on the implications of machine reading, and includes a brief description of my own intervention against the quantification of language (specifically by Google’s AdWords platform) in my {poem}.py project. A PDF of the essay is available here: Death_of_the_Reader_THORNTON.

undertheinfluence18-6
Nasan Tur (UTI 2017)

I haven’t laid hands on the actual magazine yet, but it also features some artwork by Nasan Tur that I particularly like, an interview with Professor Judy Wajcman on Digital Capitalism, Speed and The Capturing of Imaginary Futures, and a variety of other fascinating and fantastic art, photography and writings.

The magazine can be purchased here.

The key to data processing

I had an unexpected (and puzzling) email a few weeks ago which said how much the author had enjoyed my article in the Wiener Zeitung (The Vienna Newspaper). This was news to me. I’m having trouble publishing anywhere, let alone in an Austrian broadsheet – but after some investigation I found the article in question. It is a lovely piece by Adrian Lobe entitled Stichwortgeber der Datenmaschinerie (The Key to Data Processing), in which he really engages with my work critiquing Google’s exploitation of language for profit and also the methodology I have been developing through the {poem}.py project.

The article is written in German (obviously) – but here is a translation* of the last two paragraphs:

The criticism that Google reduces everything to its marketability and sees in information above all advertising potential, is not new. Many Internet critics have already referred to this, and it is not what Thornton’s project is about. It’s about what Google does with our language. What techniques does Google use to highlight specific terms, which does it hide? How is language structured within algorithmic orders? Is a Google search about communication at all? Any language or search input lands immediately in a server farm and is there evaluated by algorithms and deprived of semantics and ambivalence. Are our search queries perhaps only keywords for data processing?

Thornton reverses this non-dimensional information flow (and thus the advertising logic) by making machine output itself the starting point of (her) own, human investigations. It is also about the eminence of our own narrative. Frank Schirrmacher once wrote that we are no longer described in words but in mathematical models and formulas. Thornton’s project makes an important contribution to reflect our own speechlessness in the face of algorithmic agenda setting and automated text production.

I particularly like the last line, and will continue striving “to reflect our own speechlessness in the face of (an) algorithmic agenda” as I begin to write up my thesis.

*as a (very relevant) aside, this episode also highlights just how indispensible Google has become for so many of our daily tasks. I tried translating the article through a few different on-line tools, but Google had by far the best translation. It wasn’t perfect, however, so I dredged up my rusty A-Level German and dusted off my Collins Gem to fill in the gaps. I also would have had trouble locating the article without Google News. I know Google is not the Internet, but it really does feel like it sometimes…

http://www.wienerzeitung.at/nachrichten/kultur/medien/882818_Stichwortgeber-der-Datenmaschinerie.html

 

reclaiming poetry from the algorithmic marketplace

cropped-fullsizerender_1I’ve been playing around with some ideas for my upcoming presentation at the Research in Real-Time – Practice in Progress conference at NUI Galway next month. I’ll be using some material from the Engineering Fictions workshop I did with CONNECT writer in residence Jessica Foley in Dublin in February, where we experimented with writing the most expensive and cheapest love poems using Google AdWords suggested bid prices. We then sent the poems to each other via GMail in order to ‘expose’ the words to the algorithms, before processing them through the AdWords keyword planner and the {poem}.py code, and printing them out on a receipt.

EF_Ode-Love_Poem
Ode-Love: Winner of the cheapest love poem prize (photo: Jessica Foley)

The workshop was great fun and also extremely interesting methodologically. By creating poetry manually, and in the firm knowledge that the words which pass through Google’s advertising and search platforms are always already infused with algorithmically and economically mediated ‘values’, it felt like we were able to reclaim some of the artistic agency from the algorithms that increasingly second guess our linguistic intentions. We were able to second guess the second guessers, and it felt really liberating! Jessica and I plan to write the experience up in full very soon.

In the meantime, I have returned to the poem that started this whole project off – my favourite poem – William Stafford’s At the Bomb Testing Site (1960), which was the first poem I ‘valued’ as part of my {poem}.py project. When I first ran the poem through AdWords last year, I used the Ad Groups function of the keyword planner to try to reverse engineeer the logic of the bid prices Google suggests for each word, and for the poem as a whole. This feature suggests other keywords, phrases and topics which might help enhance your advertising campaign, but it also provides a glimpse into what the algorithms ‘think’ you are trying to advertise and as such is a fascinating insight into what words ‘mean’ to Google. Last year, references in Stafford’s poem to a curved desert road, hands gripping and tense elbows, generated suggestions that I was trying to advertise road biking. The phrase ‘ready for change’ had the algorithms thinking I was planning a well-being or recruitment campaign.

stafford
http://www.poetryfoundation.org

This year, the road biking suggestions have gone, but other even more fascinating semantic assumptions have appeared. To Google, at this moment in time, At the Bomb Testing Site conjures up Carl Jung, gastric bands and Californian Republicans. So in the spirit of reclaiming the poetry from the algorithmic marketplace, I decided to rewrite Stafford’s poem using only the suggested advertising categories and potential related search queries offered to me by Google when I put the poem through the keyword planner. With apologies to the estate of William Stafford, here is the result:

At the Bomb Testing Site (2017)

By Google AdWords (after William Stafford)


I’m feeling stuck.

Atomic trinity:

anger, depression, ego

and archetype elbow pain after fall.

California republican

delegates latest nuclear test.

Popeye syndrome.

Who invented the hydrogen bomb?

Carl Jung’s shadow?



I don’t like myself.

Business goals,

data entry jobs,

weight loss surgery in mexico.



I am ready to change my life

Self referral mental health

define psyche.

Inner self crossword clue.

Feel joy! Wellbeing,

core beliefs,

gastric bypass,

bikini island.

Ready steady:

be yourself.

 

Politics, Poetry and Google: The Value of Words in an Age of Linguistic Capitalism

An alternative angle to the fake news and Google Ads debate. It’s more poetic, but by no means less political.

“Control of language equals control of thought, and it is private capital and tech companies who are at those controls.”

penguin-books-twenty-seventeen-george-orwell-complete-unabridged-13497898

 

I recently published a short piece on The Medium about the similarities between language in the age of Google and Orwell’s Newspeak in 1984. It uses Orwell’s text to critique linguistic capitalism and the political power of language, and imagines the rise of Google as a neoliberal thought police.

The full article is here:  Politics, Poetry and Google

 

View story at Medium.com

View story at Medium.com

SUBPRIME LANGUAGE: The Precarious Value of Words in an Age of Linguistic Capitalism, Digital Advertising and #FakeNews

As the value of words shifts from conveyor of meaning to conveyor of capital, has Google become an all powerful usurer of language, and if so, how long before the linguistic bubble bursts?

I’m giving a talk at Trinity College Dublin next week as part of the CONNECT centre and Engineering Fictions. I’ll be using a lot of the material from the talk I gave at NUIG a couple of weeks ago, but I also want to try out some of the new ideas I’ve been developing around the idea of subprime language and linguistic liquidity. Below is an extended abstract/intro for the new stuff. It is work in progress – any thoughts are welcome…. I hope also to develop these ideas at the AAG in Boston and at the RGS-IBG in London later this year. 

As tech companies such as Google increasingly mediate and monetise the informational landscape through search and advertising platforms such as AdWords and AdSense, the ongoing effects on and of the language they auction, sell and exploit are becoming more and more palpable. In the viral spreading of fake news and political click-bait, and in the daily battles for exposure, it seems that words are being lent against a narrative so tenuous as to make their linguistic function negligible. Infused with a neoliberal logic which favours advertising dollars over truth and the systemic bias of algorithmic processing, the discursive side-effects of this semantic shift reveal a deep-rooted weakness in the linguistic marketplace which reaches far beyond the linguistic sphere and into the political, with powerful and potentially devastating consequences. Were it not for an overriding metanarrative of neoliberal logic, this evolution in the ontology of digital language may seem like an obvious manifestation of the postmodern condition. But as the value of words shifts from conveyor of meaning to conveyor of capital, should we be thinking of Google as the all powerful usurer of language, and if so, how long before the linguistic bubble bursts?

In this paper I set out some recent thoughts about the idea of subprime language – asking questions such as how much and how often language can be bought, sold or ‘borrowed’ before it becomes exhausted of meaning and restrictive of expression and understanding. How resilient is language to a quasi-capitalist operating system, and what happens if/when linguistic capitalism crashes? And finally, knowing the historical and cultural power that a control of language can have, the fragility and unpredictability of the economic system which now seems to underpin it, and with a growing awareness of the power wielded by technology companies such as Google, should we not be more aware of the the potential dangers in these techno-linguistic shifts?

In recent weeks the fake news debate has been evoking numerous references to Newspeak, the language of thought control and state propaganda employed to further the ideology and control of English Socialism (INGSOC) in George Orwell’s 1984. It is an interesting analogy, but I think rather than a straight forward comparison to the misinformation and alternative facts seemingly employed during the Trump campaign, there are deeper problems within today’s informational infrastructure that a more thorough reading of Orwell’s text draws out. Firstly, there is the assumption in Newspeak that “thought is dependent on words”, a somewhat problematic yet entirely relevant causal linkage when it comes to debates about search results, auto-predictions, filter bubbles and algorithmically generated social media newsfeeds, which can be instrumental in the cultivation of extreme views and hate crime.

The second issue concerns the limitations and restrictions of language that is so important to the idea of Newspeak, a language which “differed from most all other languages in that its vocabulary grew smaller instead of larger every year”. We can see echoes of this in the shrinking of creative vocabulary of digital language in favour of words which might be cheaper, easier to find, or more alluring either to algorithms or to human readers.

The third point I want to explore takes the culmination of the first two – i.e. that words have a real effect on how we think, yet the way information flows through the digital spaces encourages the shrinking of our online vocabulary and discourages non-normative language – and complicates this already worrying formula with an overriding motive not of state political control (as in Orwell’s dystopia), but of private capital gain (as in advertisers and tech/media companies). In the digital networks of information and communication we have created, the potential for political control comes often as a side effect of the economic incentive, or as a manipulation of a system which allows language, and therefore thought, to be so dependent on and subject to a neoliberal logic which is itself so precariously mediated by algorithmic systems and networks.

 

 

PODCAST: Pip Thornton – Critiquing linguistic capitalism, Google’s ad empire, fake news and poetry

Algocracy and the Transhumanist Project

slide1.jpg

My post as research assistant on the Algocracy & Transhumanism project at NUIG has come to an end, and I will shortly be returning to Royal Holloway to finish writing up my PhD. I have really enjoyed the five months I have spent here in Galway – I  have learned a great deal from the workshops I have been involved in, the podcasts I have edited, the background research I have been doing for John on the project, and also from the many amazing people I have met both in and outside the university.

I  have also had the opportunity to present my own research to a  wide audience and most recently gave a talk on behalf of the Technology and Governance research cluster entitled A Critique of Linguistic Capitalism (and an artistic intervention)  as part of a seminar series organised by the  Whitaker Institute’s Ideas Forum,  which…

View original post 513 more words

Talk at NUIG 25th Jan – Linguistic Capitalism – technology & governance research cluster

slide1

I’m giving a talk at NUI Galway on Wednesday 25th January as part of the Whitaker Institute Ideas Forum seminar series.

It will be an explanation and exploration of all things Linguistic Capitalism, with a demonstration of my {poem}.py  project, as well as some new ideas about the role of Google advertising in the fake news debate.

Most exciting of all is a guest appearance from Galway poet Rita Ann Higgins who will be reading some of her poem Killer City, to help illustrate the talk.

Being Human | Human Being: a panel discussion of Ex Machina

Ex Machina panel: if Ava was trained on search data, how come she doesn’t try to sell Nathan the pair of trainers he googled months ago? And other insights….

Algocracy and the Transhumanist Project

exmach-poster

Back in March I co-curated a Passengerfilms event in London which used Alex Garland’s 2015 film Ex Machina to provoke a panel discussion about what it means to ‘be human’ in a world in which the digitally -or algorithmically – processed ‘virtual’ is increasingly experienced in the actualities of everyday life. I wrote a post on my own blog about the event at the time, but have now had the chance to edit the audio recording of the panel discussion, which features thoughts on the film and on the wider discourse from John Danaher (NUI Galway) and myself, as well as Lee Mackinnon (Arts University, Bournemouth), Oli Mould (Royal Holloway) and Mike Duggan (Royal Holloway).

We held the event in the downstairs area of The Book Club, an East London club venue, so some of the audio is accompanied by a booming bassline from the upstairs bar. I have tried…

View original post 314 more words