Google’s ambition “to make Google an institution that makes the world a better place” stands in stark contrast with allegations of privacy violations, bias, self-preferencing, ad fraud, censorship, anti-competitive behavior, and nonreciprocal exercise of market power. In my previous blog, I presented the Google’s ecosystem and revenue models. In this blog I sketch the contrast between Google’s ambition and its practice, and point at a way forward.
Our fully documented analysis of Google can be found in a companion White Paper.
To improve the world is a mission to serve the public
In their IPO letter to Google shareholders, dated 2004, Larry Page and Sergey Brin express the ambition “to make Google an institution that makes the world a better place.” Their goal was “to develop services that significantly improve the lives of as many people as possible.” This sounds like an NGO with a mission to improve the world.
As an example of their good intentions they explain that to have a positive impact on the world, Google makes its services “as widely available as we can by supporting 90 languages and by providing most services for free.” However, the reason to make search freely available to as many users as possible is not to serve the public, but to enlarge the market for paid advertisement, which serves shareholders.
Since its founding, Google has built a global public infrastructure for online search, geographic information, scientific information, flight information, video and music streaming, and more. It has entered our office with Google Cloud and Google Workplace. It has entered our homes with Nest smart home devices, Google TV and the Stadia gaming service. Since its acquisition of FitBit, it accompanies us when we do our running. Google has entered schools with Google Classroom and it sits in the pockets on almost 90% of smartphone owners world-wide. Waymo will drive us where we want, and the Android Auto operating system will follow us in other cars. What could possibly go wrong? Google does no evil.
Google’s share value has increased from US$ 85 at IPO in 2004 to almost US$ 3000 today. Its market cap has exploded by a factor of 1000. The company is now worth almost US$ 2000B. Its revenue in 2020 was US$ 183B, up from US$ 162B in 2019. This is bigger than the government revenue of Denmark and rising. Google has a near-monopoly on online search, dominates the online advertising market, and is under scrutiny from regulators for monopolistic behavior. Advocacy groups protest its violation of privacy and intellectual property rules.[i]
And still it sounds like an NGO. In its 2018 SEC report it says that “Our vision is to remain a place of incredible creativity and innovation that uses our technical expertise to tackle big problems” (pages 3-4).
A year later: “We’re focused on building an even more helpful Google for everyone. We aspire to give everyone the tools they need to increase their knowledge, health, happiness, and success.” (SEC 2019, page 5).
Another year later: “We see tremendous potential for devices to be helpful, make your life easier, and get better over time, by combining the best of our AI, software, and hardware.” (SEC 2020, pages 5, 6).
These statements are muted by US$ 147B revenue from advertising alone. Services are free for users, but Google is definitely not free for advertisers. You can try to improve the world with the best of intentions, but with a US$ 2T valuation you tend to feel more responsible to shareholders than to the non-shareholding public. Google’s revenue does not come about without generating social and economic cost.
Purely technical solutions don’t exist
Part of the cost of doing business for Google follows from a second aspect of Google’s origin myth, after the professed goal to improve the lives of as many people as possible: Google solves big problems by purely technical solutions.
This is unbelievably naïve. One characteristic of technical problem solving is that any solution will consist of a technical and a social component. If you introduce online search in society, this will impact people’s behavior. Big Tech can introduce social networks, games and streaming in our homes, and this affects the way we communicate with each other and entertain ourselves. This is not different from the way that computers and networks changed our economy in the past 50 years and the introduction of railways and the telegraph transformed the economy in the 19th century.
The technical component of a solution can be designed, but the social component cannot, at least not completely, and not in the same way. We can try to influence people, we can guide them, instruct them, or force them, but people will always surprise us in their responses. We cannot design social systems as we can design technical systems.
We can try to predict what people will do with devices and algorithms, but these predictions are usually inaccurate, and we do well to study the impact of innovations after they are implemented so that we can improve our solutions to deal with undesirable “side” effects.
And when we do that, we usually find unforeseen and undesirable behavior. The Campaign for Accountability found that Google places ads against fake news. Fake news spreads faster and wider than real news and so tends to generate more ad impressions, which generates ad revenue for Google.
In addition to placing ads against fake news, there are ways to generate fake ad impressions. Fraudsters acquire legitimate apps to train bots on the way humans use apps. The trained bots are then used to generate ad impressions on fraudulent web sites, as if human beings rather than bots viewed those sites, thus raking in advertising dollars for the fraudster — and for Google.
Google Search has been accused of delivering results biased against women, black people, or political conservatives. This has been attributed to bias among algorithm designers, bias in user behavior summarized by the ranking algorithm, or bias induced by advertisers who want their products visible on top of the answer page.
This takes Google into difficult discussions about politics, morality and censorship that are a lot more specific than the slogan “Don’t be evil”.
Wikipedia defines bias as “disproportionate weight in favor of or against an idea or thing” but this shifts the problem. What is disproportionate? Since we do not know the total set of URLs out of which Google Search selects and ranks the ones we see on the answer page, we cannot know if the selection is disproportionate with respect to the total set.
What remains is our judgment whether search results are appropriate. Even if Google’s search results would be proportionate to the total set from which they are selected, the total set may be very inappropriate with respect to the norms of some of us. It may contain a lot of racist, xenophobic and illiberal results that we find inappropriate.
This is a moral discussion with political implications. For a global service such as Google’s, it turns out that there are few ways to resolve bias that we can globally agree on.
Which leads to the problem that some governments find some search results, or queries even, inappropriate. They demand censorship of search results. At the request of the Russian government, Google and Apple are reported to have removed Navalny voting apps when Russian elections began. Google was reported to develop a censored search engine for China but apparently this has been withdrawn.
Tech solutions are always part of sociotechnical solutions. The bigger the scale of implementation, the harder it is to predict and manage the social part of the solution. Big solutions to big problems may create big surprises.
Google engages in anti-competitive behavior
The goal “to develop services that significantly improve the lives of as many people as possible” was executed without bothering too much about privacy violations, fake news, ad fraud and censorship. But it was executed with a lot of attention to gaining a dominant position in search and advertising.
The report published in October 2020 by the House Committee on the Judiciary of competition in digital markets contains numerous examples of anti-competitive behavior by Google. The UK Competition and Markets Authority published a damning report about anticompetitive behavior in digital advertising, which provides material for an anti-trust case against Google. Here are a few of the findings.
Google search tends to prefer Google services by listing them on top of the answer page. Google pays smartphone manufacturers to be the exclusive default search engine, giving it a command of the marker. By a variety of exclusionary techniques, Google may rake up as much as 42% of the fee that advertisers pay for digital advertisements. This far exceeds its cost of capital, estimated to be 9%.[ii] Through its control of Android, it has exclusive access to the usage of apps that compete with Google’s apps. Through its control of all levels of the ad stack, Google controls the digital advertising market and is able to exclude competitors.[iii]
Is there a problem?
This may be viewed as market dominance necessary to improve the world for a lot of people, including shareholders and users. However, in practice this leads to higher prices for advertisers, which end up as higher prices for consumer products. It leads to lower revenues for publishers, which results in lower quality of content that they can publish. It forecloses technological rivals, which reduces innovation. And it reduces choice for consumers, even though Google has less incentive to show the best possible search results.
Google’s professed mission to improve the world through technical solutions detracts attention from privacy violations, bias, self-preferencing, ad fraud, censorship, anti-competitive behavior, and nonreciprocal exercise of market power.
Much of Googles market dominance and anti-competitive behavior comes from its presence on several layers of the search tech stack and from its presence on several layers of the ad tech stack, which gives it unfair information and control advantages in both stacks. Governments are now working hard to update anti-trust law to the age of digital platforms, and we may see Google’s dominance of the market curbed in the interest of competition.[iv]
A surveillance capitalist must be transparent
But this will not be enough. Google is not just a dominant digital infrastructure; it is a dominant digital content-delivery infrastructure. Google extracts data from people to generate revenue. As I argued elsewhere for big tech platforms, it should be transparent about the data they collect about their users, about the beneficiaries of the data, and about the way they generate revenue from this. Google should provide comprehensible meta-information about advertisements, be transparent about what they do to combat misinformation on its platforms and allow researchers to analyze its network.
Google may argue that these are trade secrets. But there should be nothing secret about what a company does with our data. Data is not like oil. Oil is extracted from rocks, but data is extracted from people and the data-extracting company is answerable to these people.
Google should also accept auditors who check the veracity of Google’s transparency reports. And a government agency should check the comprehensibility of the reports. If Google’s privacy pages are a guide, they should be forced to be comprehensible on issues they’d rather be silent about.
Google has become part of the digital content infrastructure of billions of people in hundreds of countries. With great size comes great responsibility. There is nothing wrong with a bunch of tech optimists trying to improve the world. Without technology, we would still be collecting berries on the savanna. To solve today’s problems, we need technology, optimism, and perseverance. And honesty about the impact of our solutions.
If Google’s improvements of the world require us to sit still like rabbits in Google’s headlight, then there is something deeply wrong with the way it improves the world. The cost of obedience in a surveillance society is too high.
Since Google profits from collecting data about the public, it is answerable to the public. When the idealistic drive to improve life is taken over by the monetary drive to generate value for shareholders and this succeeds on a scale never seen before, without due attention to the social part of these solutions, then public governance is required to guard the interests of the public.
[i] Rana Foroohar. Don’t Be Evil. The Case Against Big Tech. Currency, 2019.
[ii] Fiona M. Scott Morton & David C. Dinielli. “Roadmap for a digital advertising monopolization case against Google.” Omidyar Network, May 2020. https://omidyar.com/wp-content/uploads/2020/09/Roadmap-for-a-Case-Against-Google.pdf
[iii] Fiona M. Scott Morton & David C. Dinielli (2020).
[iv] Congressional Research Service. The Big Tech Antitrust Bills. https://csreports.congress.gov, August 13, 2021. Digital Markets Act, https://ec.europa.eu/info/strategy/priorities-2019-2024/europe-fit-digital-age/digital-markets-act-ensuring-fair-and-open-digital-markets_en