Why advertising in cyberspace is bad for us and what to do about it.

Tony Curzon Price Feb 2024

In which I outline 3 classic economic arguments against advertising - 2 from Kaldor, 1950 and 1 from Becker and Murphy, 1993; a whole list of more specific externalities related to advertising in cyberspace; a pointer towards what people are saying we should do about it; and why I think that data unions are needed to reclaim our attention

2 visions of what it is to be a consumer

Are we fish swimming in a sea of tempting offers? Could we become masters of our attention, always auctioning it just right? Or is that too much to expect of all-too-human citizen-consumers? If so, what is the solution?

Online advertising absolutely dominates and powers consumer-facing cyberspace today. In 2018, the CMA reported that Google and Facebook accounted for 80% of a total £14bln or so online advertising industry.[1] That would be about £400 per year per annum that is, effectively, our “subscription fee” for their services. The total subscription fee to the free-to-view web that advertising funds would be about £500/year/household in the CMA’s reckoning. According to the IAB industry association, this figure has approximately doubled since.[2] Our subscription to the free-to-view web is costing every household in the UK about £1000 per year. Of course, to generate these revenues, the industry channels funds to content creators via their ad networks and the platforms spend on their own product development. But also, a big chunk of the sum goes to shareholders, or in the case of meta Facebook, the main shareholder ’s pet projects. Google has margins of 64%.

The turnover that Facebook and Google report comes almost exclusively from advertisers, from people wanting to sell other people stuff, people looking for customers. And they pay search engines, they pay social media publishers, they pay other publishers, they pay influencers, they pay app stores, apps, and all of that for placement in front of eyeballs. So almost all of that turnover is payment for eyeballs, or more politely and informatively, it is payment for attention.

The point of these numbers is to give a sense of the scale of the industry we are talking about. Before Putin's war, energy bills were about £1000 per year per household, and our energy system is examined and regulated and understood in miniscule detail by the public sector, with its own dedicated regulator and ministry. Advertising in cyberspace, on the other hand, is remarkably opaque. This note asks whether the industry is working in the interest of the public good, or whether it needs reform.

On the face of it, it might seem like a very good thing. We get a vast amount of free to view entertainment and information in exchange for our attention. The days when you were at the newsagent and thought "oh, I can ’t afford to buy this magazine which really looks nice", and you found yourself standing and reading, thinking "wouldn ’t it be nice to take home and sit in the corner of my sofa and enjoy this publication" ... those days are well and truly gone. Thanks to advertising in cyberspace, we have the best magazine rack in the world at our disposal, everywhere, at every moment of the day, instantly refreshed.

The free to view entertainment boom is unbelievable. Content creators have been able to offer free-to-view material and earn a living from advertising. Many creators complain they get too little ... but surely it was ever thus with content creators. We all of us have our favourite examples of content creators who would never have existed without the online advertising bonanza, and who have made our lives better. I have my own DIY pretensions to confess to, and finding that incredibly fluent and skillful mechanic taking apart a carburettor and telling me exactly how I can clean this little injection hose - that has enhanced my quality of life. I look at his YouTube stats - usually it is a 'him' in this category - and he has 170,000 views for this video. He ’s made a little bit of money from it, and if that is what keeps him making these videos, I am very willing to trade some of my time, attention and cash (in the form of higher end-prices) for it.

The tech firms themselves have invested in ever more attractive and attention grabbing tools. Just as an example, the entire Android operating system really was conceived as being an adjunct to Google ’s advertising business. About 2006 or so, Google could see that web viewing was going to move online, so it needed mobile data to maintain its gusher of good sales leads. What better way to ensure this than to provide the so-called open source alternative to Apple ’s iOS. So we get the Android operating system. And out of the advertising incentives, of course, we get the search engine, we get maps, we get the organisations that support social media and keep them moderately clean and safe, we get online email that is pretty good at filtering spam...

In other words, this advertising revenue supplies products, which are certainly attention grabbing, often useful, and in some cases, genuine additions to human capability. So that ’s the case for.

However, I am going to argue that on some sort of utilitarian calculus, the industry as it ’s currently structured, and with all appropriate counterfactuals taken into account, creates, I believe, substantial harms and that the benefits don ’t pay for it. I also think that we have the classic structure here of market failure: the benefits accrue privately, and the harms are socialised. This is a set-up guaranteed to produce too much of a social bad.

Three classic economic arguments about advertising

I ’m going to start by making three classic and quite general arguments, they ’re arguments about advertising rather than advertising in cyberspace. And after those three arguments I will turn to some much more specific cases about the advertising industry as it ’s grown up in cyberspace.

The first argument is one that I attribute to Kaldor - he makes it in a 1950 article in the Review of Economic Studies.[3] He points out that advertising is a "subsidised commodity" - it never enters on its own in consumer choices, but its costs are spread across all consumers, regardless of whether the advertisement was of use to them or not. It is not simply that advertising is an overhead. Nissan ’s accounting software is a fixed cost, and I pay for a bit of it with the car I buy. I don ’t exercise consumer sovereignty over the choice of accounting system Nissan has. However, as a Nissan customer, I am a beneficiary of Nissan ’s existence, and there are some costs that are required for it to exist. Fine. And perhaps the market provides some ultimate sanction against management teams who make bad choices of accounting software. But the situation with advertising is subtly different: I may be in absolutely no sense a beneficiary of some advertising spend - perhaps I am an absolute Nissan loyalist to whom persuasive messages are always a waste of funds. And yet I still have to pay for Nissan ’s advertising, like everyone else. The benefit of advertising - so far as there is any - is a private affair. But its cost is socialised. In cyberspace today, it is very hard for me to avoid paying that £1,000 per year subscription fee to the free-to-view web.

So what? As Kaldor points out, this feature means that the mere existence of the profitable supply of advertising in a moderately competitive context does not provide any prima facie case that it is a net contributor to consumer welfare. He writes:

"For advertising, being a subsidised commodity, is not supplied in response to consumers ’ demand; the scale of expenditure on advertising-unlike the scale of expenditure on goods and services which are not subsidised-is not determined by the preferences of the consumers, as registered through the price-mechanism, but by purely extraneous considerations [...]

This does not necessarily mean that the expenditure, from a social point of view, is wasted (in the sense that it brings no utility-or a utility considerably less than the cost); it means, however, that it needs to be justified by considerations other than profitability. In an analogous manner to the goods and services provided through the public purse, the question whether it is wasteful or not must be determined by reference to the presumed social utility of the service which the expenditure provides."[4]

Kaldor makes a second argument in his article - that “persuasive” advertising that shifts demand from one producer to another - as opposed to “informative” advertising which provides genuinely useful information beyond the mere existence of a good - has no economic value: in a crude case, the two goods are in all important respects identical, so it does not matter from the perspective of allocative efficiency which producer is the supplier. Advertising simply adds to the cost of the goods. Note that this argument against persuasive advertising has nothing to do with the inherent problems of judging welfare when preferences can be manipulated by sellers - it is simply a point about allocative efficiency on the assumption that the post-exposure preferences are the right measure of welfare.

This argument, of course, fails to place any value on rivalry - how can you even get competition going without attempts to take market share? However, the argument nevertheless contains an important truth: although some advertising may be necessary for an adequate amount of rivalry, there is no reason to believe that competition will lead to the right amount or type of competition. Kaldor points out that the right kind of rivalry could be induced by the consumer body reports, probably at a fraction of the cost of the persuasive advertising that actually often leads people to choose one product over another for the wrong reasons - “it'll make you irresistibly attractive” is a siren song that never delivers. Preferring products for the wrong reasons also, of course, runs the risk of leading to product development and innovation in the wrong direction from the perspective of pre-advertising preferences.

The third general argument that I’d like to draw attention to is one due to Becker and Murphy, ultra-free market economists from Chicago, published in 1993, at the dawn of the cyberspace era.[5] They continue with Kaldor’s general characterisation of advertising as being analytically equivalent to a demand complement - a good whose supply together with another good increases the demand for that good - and consider the question of advertising being what a consumer “pays” for free-to-view content (at the time of their writing, this was television and radio). They are relatively comfortable with advertising in paid-for media, because they consider that in these cases there is a price-mediated choice being made by consumers about the whole product as a bundle. Competition might work adequately to point producers to make consumer-sensitive choices. But they consider free-to-view content to be more problematic, since the conclusion is inescapable here that viewers are being compensated with content for their attention - the advertising is not a direct supplier of utility. They do not say it, but the case is made perhaps stronger by the behavioural economics literature that points to the specialness of zero as a price: behaviours really do change at something that looks “free”.

For free-to-view media funded by advertising, they offer the following striking comparison:

And just as death, divorce, unemployment, and similar utility-reducing events often induce greater drinking, smoking, overeating, and similar changes in consumption, we believe so too do many advertisements lower utility and yet raise demand for the advertised goods. These ads produce anxiety and depression, stir up envious feelings toward the success and happiness of others, or arouse guilt toward parents or children

To be placed in the same category as "death, divorce and unemployment" is perhaps not the ringing endorsement that the advertising that funds free-to-view content might have wished for… (Kaldor has a fascinating footnote in a similar vein - he points out that advertising creates needs and wants that no one was ever aware of, and he mentions the recent invention of Body Odour as a problem that needed fixing.)

In other words, advertising pays for itself by increasing the demand for a good which does carry a utility - it must, because that good is tested in the market - but there is no a priori reason to believe that it itself carries a positive utility. It is simply not put to a market test. (Imagine the market test for a cigarette advertisement, rather than for a cigarette: “Go on. Have a peek at this. It will make you feel inadequate, and will offer an unhealthy and fake solution to the problem it just created, it will get you hooked and you’ll probably die early. But go on. Have a look. Be tempted”. A tough sell on its own…)

Those are three general arguments about advertising, and they all point towards its problematic nature from the point of view of a fundamental utilitarian calculus. At the very least, they should alert us to the fact that advertising supporting free-to-view media (as is the case, in various ways, with search, display, social media and app advertising) needs to be examined on its merits. And therefore, I now consider the specific harms that the advertising-entertainment ecosystem has produced in cyberspace.

These can be split into four large buckets, which I consider in turn:

Direct Economic Harms

Furman focuses on two in particular: the monopoly rents that have accrued to the platforms in cyberspace, and the “data hoarding” that ad-funded platforms engage in.

The rents point to a failure of competition: when a sector persistently maintains the kind of margins that we see in the advertising-funded platforms without entry reducing those rents, competition has failed to do its usual disciplining work.

Data hoarding is perhaps a less expected economic harm. Data, once collected, is a near zero marginal cost good that is non-depletable. Ignoring privacy considerations for now, that data, from a social welfare point of view, should be widely available at almost no cost for anyone to find value in. However, the de facto guardians of huge amounts of the data, the platforms, have an interest in restricting its use to their own private good as an advertising signal. To let it out for other purposes would cannibalise that revenue stream, since inferences could be made for any purpose, including rival advertising sales.

The surveillance consumerism harms

Here, there are harms related to privacy violations, in the sense of people sometimes finding that the general availability of data about them has been used in ways they would rather not. An iconic example is the case of the Grindr priest, as told in this Washington Post story.[7]

There are also the more abstract harms pointed to by Soshana Zuboff in her magisterial “Surveillance Capitalism”, where she wants to point to the underlying harm caused by “extraction of human futures”, a sort of cyberspace analogy to the concept of extraction of surplus value in the Labour Theory of Value.[8]

Advertising/Media bundles are themselves harmful products

This is the collection of harms that has preoccupied campaigners for child and teen safety & mental health, for example in the work done by Beeban Kidron that has led, in the UK, to the passing of the Online Safety Bill. In versions of the harms already identified in general terms by Kaldor and Becker/Murphy in the old-media age but supercharged by transposition to cyberspace, we have serious adverse consequences on body image, feelings of inadequacy, self-harm, various media addictions, including ad-funded pornography. All of these have been created in very large part by advertising ’s bottom-line imperative of keeping us online, giving attention to ad-funded products, for longer. This is part of the story beautifully told in Johann Hari’s Stolen Focus[9], including revealing interviews with some of the inventors of the technologies that have had these effects. For example, there is a moving account of the life-journey of Aza Raskin, the inventor of the infinite scroll, who now regrets the code he built.

Wider socio-economic harms

Perhaps, as Hari argues, advertising in cyberspace has changed our psychology on a mass scale, destroying attention, imposing cognitive distraction costs and even perhaps contributing to the productivity puzzle. (The release of the iPhone comes in the same year as the Global Financial Crisis, the date usually taken to be the start of the productivity “anomaly”, at least in the UK).

One of these wider harms is that advertising in cyberspace has destroyed what used to be a positive externality of old-media technology in creating a relatively broadly-tented public sphere. Habermas and Benedict Anderson emphasise how important broad-based media have been to the development of democratic politics.[10] The creation of “broad tents” in newspapers, then radio and television stations, was a direct result of the fixed costs of publishing: content had to be shared widely. Given that content both responds to citizens’ and consumers’ demands and shapes them, these broad tents very plausibly created “imagined communities” with some degree of shared value and ethos. These communities are argued to be prerequisites for democratic politics inside the nation state. The reduction of fixed publishing costs to almost zero in cyberspace has dismantled the old externality and, according to the logic of these theories, is the underlying cause of the fragmentation of our politics. It is competition to supply the advertising imperative to maintain attention that has led to very different types of “imagined communities” establishing themselves online, often demarcating in-groups and out-groups within the boundaries of the older imagined community of the nation state. These are the sorts of harms that were considered in the 2019 Cairncross Review.[11] However, of the trio of forays into cyberspace policy of the last 10 years, this one has had the least tangible impact - the other two led to the creation of the Digital Markets Unit and the Online Harms Bill.

It’s all about attention

That concludes the canter through the specific harms that could be laid at the door of the advertising model’s over-dominance in cyberspace. It perhaps seems a little bitty - a list of charges against, with no underlying logic. However, I think there is a unifying thread to all of this which relates to the power of getting another person’s attention. When you have someone’s attention, you have the opportunity to lead their mind somewhere. It is an option that is easily fluffed - how often do you get distracted and then pull back from the next ask of your attention? - but if you handle the other’s attention carefully, it can be very powerful. You can plant seeds in their minds, make them see the world in ways they had not before envisaged, etc. This power of guiding a mind is at the heart of human sociability; it is a power that constitutes us as a group. And like all powers, it can be used well or badly. All of the harms considered above with the exception of the privacy harm are related to privately gainful, publicly harmful ways of capturing attention. And, as we will see below, the exception of the privacy harm is interestingly the one to which the dominant platforms want to offer a solution.

Four families of remedies

So what if all of the above points to something really big that is wrong in cyberspace? What should we actually do to fix it?

The solutions that I have seen discussed fall into 4 categories, I think. I have a distinct preference for the last, but try to summarise fairly each of them.

  1. Step up state control of cyberspace, civilise the platforms and regulate out of existence the bad actors
  1. Cut out the risks and the rogue players who are over-sharing data with strong ICO enforcement of data regulation, and do not worry too much if the platforms decide to implement privacy-enhancing solutions themselves
  2. Strengthen conduct regulation of the platforms as permitted by the legislation setting up the Digital Markets Unit,
  3. Clamp down heavily on online harms by content-regulating the platforms - here again, if there are fewer big players to discipline, that makes regulatory life easier
  4. Beef up advertising standards
  5. Offer the mainstream press a cut of the ad revenues they have lost through final offer arbitration
  6. Revive the idea of a platform or advertising tax to recycle rents
  1. Create a competitive advertising market
  1. Do all of the above, but worry a great deal about the market power consequences of allowing the platforms to use the harms outlined to regulate away their competitors
  2. Do not worry excessively about rents in advertising - as long as the market is contestable, they should be dissipated into more and better free-to-view content
  1. Develop the tools and regulations required to turn the tables on the advertising-entertainment complex and create an “intentional economy”
  1. Take inspiration from Doc Searls’ early thinking on these questions, try to empower consumers to announce their intentions, and for producers to respond to these by pitching offers.
  1. Create the conditions for the flourishing of civil society data unions (also called fiduciaries or trusts or clubs) that can relieve the cognitive burden to the individual of life in cyberspace while negotiating on our behalf with its big beasts.
  1. This last, my favourite remedy, is the only one that I believe has the power to address most of the harms outlined while preserving the enormous power of cyberspace to be good for humanity.

Very briefly, the disadvantages with the first three are the following:

  1. The first does not actually solve the market power of the platforms and probably even exacerbates it. This puts a huge burden on the other regulatory tools working effectively to constrain the dominant platforms and possible rogue players. The platforms are advocating for solution one because they do not believe that regulation will be effective if they keep their dominant positions, and I fear they might be right.
  2. The second purports to solve the market power problem, but does nothing to address the externalities that are due to advertising in general. Indeed, by increasing competition in the provision of such “bads”, it may well exacerbate the problems.
  3. The third is attractive but places a great cognitive burden on the individual. The early web hopes for the development of an intentional economy have foundered because the power of defaults and of convenience are very great. Cognitive burden is what underpins the market power of the platform and the harms due to attention capture.
  4. That leaves the fourth option as the only viable solution: it is the one that takes what is good from the intentional economy solutions, recognises the reality of uneven power in the face of cognitive constraints, and solves those problems by delegating the minute-to-minute choices, permissions and interactions to institutions whose values are aligned with the data-subject.

[1] https://assets.publishing.service.gov.uk/media/5fa557668fa8f5788db46efc/Final_report_Digital_ALT_TEXT.pdf

[2] https://www.iabuk.com/adspend

[3] https://diigo.com/0vcxzc

[4] Advertising in cyberspace provides technological advances that lead to an interesting thought experiment: imagine the the practice of attribution, by which marketers track the effectiveness of an advertisement in leading to a purchase, were sufficiently good for sellers to be able to reflect the cost of the advertisement in the price of the good. I never reacted to any Nissan ad, according to the ad-tracking system, so I get a discount. My neighbour, for ever on ad-filled free-to-view sites and easily distracted, was attributed as having chosen a Nissan because of an advertisement, and he therefore pays more for the car. My claim is not that this would ever be the result of a plausible competitive dynamic, but rather I ’d like to imagine what would be consumer reactions to the free-to-view web in this case. They might become more aware of the cost of their entertainment and might make different media purchasing choices.

[5] https://diigo.com/0vd4kg

[6] https://assets.publishing.service.gov.uk/media/5c88150ee5274a230219c35f/unlocking_digital_competition_furman_review_web.pdf

[7] https://www.washingtonpost.com/religion/2021/07/21/catholic-official-grindr-reaction/

[8] I have a lot of sympathy with the tone of the Zuboff argument but often find it frustratingly hard to pin down the harms. I have a go at that over here - https://tonycurzonprice.tumblr.com/post/188247156791/the-economistic-take-down-of-surveillance. For economists at least, I find her conversation with Russ Roberts useful, and yet still unsatisfactory. One day, I would like to rework Zuboff’s central concept of “human futures” into an account that is based on the power that one group can have over other people through guiding their attention. As in all such power discussions, power is both positive and constitutive and often exploitative.

[9] https://stolenfocusbook.com/

[10] https://en.wikipedia.org/wiki/Imagined_community

[11] https://www.gov.uk/government/publications/the-cairncross-review-a-sustainable-future-for-journalism