What is data dignity?

The data we produce online is valuable raw material for AI and advertising. Data dignity is the idea that, instead of letting companies soak it up for free, we should recognize it as the work of the people who made it, and bargain over and manage it collectively through intermediary organizations, rather like labor unions.

The review you wrote, the photo you posted, the text you translated. Data like this is the raw material that makes AI smarter and delivers advertising. Yet you are almost never paid for it. "Data dignity" is an idea that tries to change this.

In short

Search histories, posts, translations, photos: the data we produce online is valuable raw material for AI and advertising. Data dignity is the idea that, instead of letting companies soak it up for free, we should recognize it as the work of the people who made it, and bargain over it and manage it collectively through intermediary organizations, rather like labor unions [1][2].

For example

The translator's case (2018). In an essay in Harvard Business Review, the computer scientist Jaron Lanier and the economist E. Glen Weyl pointed out a contradiction in machine translation. Machine translation threatens the work of human translators. Yet to keep up with new words and trends, it needs large amounts of fresh data every day from those very translators [1].

Someone who subtitles foreign films for their friends has no idea how much value they are giving away. If people were honestly told that they are needed, the two argue, they would be paid and would take pride in providing better data [1].

Taiwan's "data coalitions". According to the book ⿻ Plurality, Taiwan's Presidential Hackathon (a contest for solving social problems with technology) "convened mixed teams of civil servants, academics, activists and technologists to propose tools, social practices and collective data custody arrangements that allowed them to 'collectively bargain' with their data", supported by the government's program of "data coalitions". Examples given include air quality monitoring and early warning of wildfires [3].

Data coalitions in the 6-Pack of Care. The 6-Pack of Care by Audrey Tang and Caroline Green treats existing community institutions as data coalitions. In its glossary's words, "an existing community institution — a neighbourhood association, tribal council, union, craft cooperative, or religious congregation — acts as steward of knowledge its members authorise it to manage. It bargains the Engagement Contract, decides what may be shared and what stays offline, and would receive settlement collectively rather than per person" [4]. → Kami

In more detail

The other side of "free"

The HBR essay criticizes how today's big internet companies work. They let people use services for free and collect the cost from third parties who want to influence those users. As a result, the essay says, society is being built not on consensual dialogue but on manipulation poured in behind the scenes [2].

Giant platforms such as video sites also control both the people who make videos and the people who watch them. A single company decides from above how much to pay which creator, the two say [2].

What data dignity means

The two propose a real market economy in which people are paid for their own data, and pay for services that use other people's data [1].

Weyl has called this "data as labor", and Lanier "humanistic digital economics". In this essay they choose the less politically loaded term "data dignity". They explain that it carries "human dignity", which was central to standing up to 20th-century totalitarianism, over to the present situation of protecting data from new concentrations of power [1].

An economics paper from the same year, "Should We Treat Data as Labor?", also argues that treating users' data as a company's capital ignores the role of the users who generate the data, and distributes the gains unfairly [5].

In a 2023 article in The New Yorker, Lanier describes a world in which digital things flow while staying connected to the people who made them, so that even when they are recombined by large AI models, their makers can be paid [6].

Not individuals but organizations in between (MIDs)

The gap in power between giant companies and individuals is huge. Individuals demanding data dignity on their own, or asking governments to regulate, will not work, the two say [7].

So they propose a middle layer of organizations they call "mediators of individual data" (MIDs). A MID is a voluntary group with its own rules. It negotiates fees for the use of its members' data, sets quality standards and builds a brand, and also handles accounting and legal work. The two also write that the idea itself is not new, since companies, labor unions, farm cooperatives, universities and publishers have played the same role in the physical world [7].

How Plurality develops it

Chapter 4-4 of the book does not use the term "data dignity", but it takes up this line of thought as a question of property and data sharing. The book says that "organizing data rights and control at the individual level is impractical", for reasons including these three [8]:

Social leakage. As with genetic information, one person's data reveals things about others.
Management challenges. "It is nearly impossible for an individual alone to understand the implications, both financial and personal, of sharing data in various ways."
Collective bargaining. To negotiate on equal terms with giant companies, data creators have to act collectively.

The book writes that organizations able to take on "this role of collectively representing the rights and interests of 'data subjects' have been given a variety of names: data trusts, collaboratives, cooperatives, or, in a whimsical turn of phrase one of the authors suggested, 'mediators of individual data' (MIDs)". Some could extend existing organizations: "for example, unions for creative workers representing their content, or Wikipedia representing the collective interest of its volunteer editors and contributors". Others may need new forms of organization, such as "the contributors of open-source code that is being used to train code-generation models" and "authors of fan fiction" [8].

The book adds that there must also be rules "protecting against unilateral disclosure by a member of a MIDs (analogous to prohibitions against unilateral strikebreaking against unions)" [8].

What is debated

Weyl's own second thoughts (2020). Looking back on the data chapter of Radical Markets, Weyl wrote that the metaphor of data as labor was meant to recall the history of the labor movement and collective bargaining. But, he reflects, it leaned too much toward the question of digital shares of income, and missed the real questions of power that need to be tackled collectively: the quality of information (fake news), the flow of information (privacy) and human attention [9].
The effort of thinking (Buterin, 2018). Vitalik Buterin, the founder of Ethereum, says the same criticism he made of the Harberger tax — the burden of people having to deal with countless valuations — may also apply to many markets for personal data [10].
Clashing with basic income. Worries about AI taking jobs often lead to a choice between two options: either many people become poor, or the state hands everyone a minimum income from money it controls. The HBR essay rejects both as dark futures that concentrate too much power and ignore the value of the people who create data [2]. In 2023 Lanier also criticized universal income schemes as making everyone dependent on handouts in order to protect the idea of AI as an opaque black box [6].

Common misunderstandings

That it is about paying individuals for their data. The book, the HBR essay and the 6-Pack of Care all place the unit of bargaining and payment not in individuals, but in organizations in between (MIDs or data coalitions) [7][8][4].

Where it fits in Plurality

In the book, it belongs to the organizational layer of "⿻ property" in chapter 4-4, "Property and Contract" [8]. The argument that data is social comes from contextual integrity and ⿻ publics.

In chapter 4-1, "Identity and Personhood", MIDs appear as intermediaries that reduce the number of relationships one person has to manage in order to verify identity [11]. Chapter 5-7, "Social Markets", counts "data unions" among the reforms that change how corporations work [12]. Chapter 6-4 (on the environment), discussing a digital twin of a river, describes data coalitions as "little platoons" created by the people who benefit from the ecosystem [13]. It is also directly connected to the question of AI training data (generative foundation models).

Sources

Lanier and Weyl, "A Blueprint for a Better Digital Society" (HBR, 2018), section "Data Dignity" — https://hbr.org/2018/09/a-blueprint-for-a-better-digital-society
Same, opening — https://hbr.org/2018/09/a-blueprint-for-a-better-digital-society
Plurality, 2-2, section "Hackathons, coalitions and quadratic signals" — https://www.plurality.net/v/chapters/2-2/eng/ ［CC0］
civic.ai glossary, "data coalition", and FAQ — https://civic.ai/glossary/ and https://civic.ai/faq/ ［CC0］
Arrieta-Ibarra and others, "Should We Treat Data as Labor?" (2018), abstract — https://www.aeaweb.org/articles?id=10.1257/pandp.20181003
Jaron Lanier, "There Is No A.I." (The New Yorker, 2023-04-20) — https://www.newyorker.com/science/annals-of-artificial-intelligence/there-is-no-ai
Lanier and Weyl, "A Blueprint for a Better Digital Society", section "MIDs" — https://hbr.org/2018/09/a-blueprint-for-a-better-digital-society
Plurality, 4-4, section "⿻ property" — https://www.plurality.net/v/chapters/4-4/eng/ ［CC0］
E. Glen Weyl, "Why I Am not a Market Radical" (2020), item on data as labor — https://www.radicalxchange.org/updates/blog/why-i-am-not-a-market-radical/
Vitalik Buterin, "On Radical Markets" (2018) — https://vitalik.eth.limo/general/2018/04/20/radical_markets.html
Plurality, 4-1, section "⿻ identity" — https://www.plurality.net/v/chapters/4-1/eng/ ［CC0］
Plurality, 5-7, section "Social markets tomorrow" — https://www.plurality.net/v/chapters/5-7/eng/ ［CC0］
Plurality, 6-4 "Environment" — https://www.plurality.net/v/chapters/6-4/eng/ ［CC0］

We welcome error reports and questions ("What is…?"). Submissions become public issues.

Report an error in this article ・ Ask a question