Jaron Lanier, who popularized the term "virtual reality," founded VPL Research in the 1980s, one of the first companies to produce commercial VR devices, including the DataGlove. Since 2006, he has held the position of lead interdisciplinary scientist at Microsoft Research while remaining a vocal critic of Silicon Valley's business model, which relies on offering free services in exchange for users' personal data.
ForkLog explored the concept of "siren servers," why Lanier views AI as a product of collective human effort, and how his ideas about compensating creators relate to the $1.5 billion settlement concerning Anthropic.
Who is Lanier?
Lanier is an American scientist, musician, and author born in 1960 in New York and raised in New Mexico. In 1984, he co-founded VPL Research with engineer Thomas Zimmerman, the first company to bring commercial VR devices to market: the DataGlove and the EyePhone headset.
Source: Wikimedia Commons.Although the term "virtual reality" had appeared earlier—coined by French writer and director Antonin Artaud in 1938 in reference to theater—Lanier is credited with popularizing it in the tech lexicon during the 1980s.
He has authored several books, including "You Are Not a Gadget: A Manifesto" (2010), "Who Owns the Future?" (2013), and "Ten Arguments for Deleting Your Social Media Accounts Right Now" (2018). In these works, he critiques the data collection and advertising business models of major platforms and advocates for companies to pay users for the information they profit from.
Lanier has been with Microsoft since 2006, currently serving as the Office of the Chief Technical Officer Prime Unifying Scientist, abbreviated as OCTOPUS. He is notably passionate about the neurobiology of cephalopods, earning him the nickname "Microsoft's octopus" within tech circles.
A Dome in the Desert
At the age of nine, Lanier's mother died in a car accident, prompting him and his father to move to the New Mexico desert, where they lived in a tent for some time. When he was eleven, he designed a geodesic dome that would become their home. The construction took several years, with plans sourced from the Domebook catalog.
Source: Domebook One.By thirteen, Lanier convinced the University of New Mexico to let him attend classes despite not having a high school diploma. In the early 1980s, he created experimental games such as Alien Garden (1982) for Atari and Moondust (1983) for Commodore 64, while also working as a researcher at Atari's lab. He co-founded VPL Research in 1984 using royalties from Moondust with Zimmerman. VPL produced the DataGlove and EyePhone, establishing the first commercial devices for immersive digital experiences. However, financial difficulties led to VPL being acquired by the French company Thomson-CSF in the early 1990s, and later by Sun Microsystems in 1998.
Throughout his writings, Lanier emphasizes that technology's direction is determined by human choices and argues that the existing rules of the digital economy should be reevaluated.
You Are Not a Gadget
Lanier's first book, "You Are Not a Gadget: A Manifesto," was released in 2010, coinciding with Facebook's rapid growth toward its first half-billion users and Wikipedia being touted as a triumph of collective intelligence.
The book introduces the concept of "lock-in." To illustrate this, Lanier references the MIDI standard developed with Dave Smith in the early 1980s for synchronizing synthesizers, which established a rigid framework for millions of musicians despite its limitations. He concludes that a single engineer's arbitrary decision has become a restrictive model for creativity.
Lanier believes users gradually adapt to describing themselves in terms dictated by platforms, which impose a limited set of categories that fail to capture the diversity of human interests and relationships.
Earlier, in his 2006 essay "Digital Maoism," Lanier criticized the blind faith in the "wisdom of the crowd," arguing that collective contributions diminish the value of individual perspectives and accountability. This essay sparked significant debate, with responses from figures like Clay Shirky, Cory Doctorow, and Wikipedia founder Jimmy Wales.
Wales asserted that the belief in an all-knowing collective among Wikipedia editors is a myth, emphasizing that articles are written and edited by identifiable individuals, not an anonymous mob.
Source: Edge.This contrasts sharply with Richard Stallman's ideas, where free software is an ethical imperative. While Lanier acknowledges the contributions of Linux and Wikipedia, he argues that the slogan "Information wants to be free" primarily benefits tech giants profiting from unpaid user labor. This notion became central to his subsequent work.
Siren Servers
In his 2013 book "Who Owns the Future?", Lanier introduces the term "siren servers" to describe large digital systems that collect user data and exploit their informational advantage for profit, shifting significant costs and risks to other market participants. This category includes companies like Google and Facebook, as well as algorithmic trading funds and insurance corporations.
A notable historical comparison in the book contrasts Kodak's peak workforce of 140,000 employees with Instagram, which Facebook acquired for $1 billion in 2012 when it had only thirteen employees. Critics of this business model argue that users essentially work for these platforms without compensation, receiving only free access to the service.
Lanier contends that such an economy exacerbates wealth concentration among major tech companies and undermines the middle class. Despite the ease of copying digital information, it still holds value.
Thus, he rejects the simplistic interpretation of Stewart Brand's phrase "Information wants to be free." Advocates for free content often neglect that Brand himself viewed information as a valuable resource.
Long before the Web3 boom, Lanier diagnosed a core issue: he proposed the concept of "data dignity" with a micropayment system where corporations would pay users royalties for their digital footprints. Today, this model could address the primary conflict between content creators and AI developers.
Dignity of Data
Lanier's solution is not to prohibit data collection but to require companies to compensate users for their data. This model, termed "data dignity," was initially outlined in "Who Owns the Future?" and further developed in a 2018 article for the Harvard Business Review, co-authored with economist Glen Weyl.
The structure of "data dignity" includes:
- Each data unit contains a link to its source. If a photo, text, or travel route contributes to something valuable—like targeted advertising or model training—the source receives a micropayment.
- Individuals form "mediators of individual data" (MID), akin to a union or guild. This intermediary negotiates with platforms on behalf of thousands of participants and ensures that data is not undervalued.
- The result is a new type of labor market, where every uploaded photo and written comment becomes a compensable contribution.
Lanier is clear about the origins of this idea: Ted Nelson described bidirectional links and micropayments for citations in the 1960s in his Xanadu project. He argues that the current web, with one-way links, has gone astray; a link knows where it leads but not where it came from, preventing authors from benefiting from their work.
Ten Arguments
In 2018, Lanier published a manifesto against Silicon Valley, summarizing social media's structure with the acronym BUMMER. The main thesis is that platforms do not profit merely from attention but from the ability to subtly manipulate our habits, selling this capability to anyone willing to pay.
However, Lanier does not advocate for a boycott of technology per se—neither smartphones nor social media. His more pragmatic advice is to avoid services that operate on an advertising model, encouraging platforms to shift towards subscription models or direct payments to users for their data. As long as people continue to use these services for free and generate content, companies have no incentive to change their established business models.
No AI Exists
In April 2023, just months after the launch of ChatGPT, Lanier published an essay in The New Yorker titled "There Is No AI." His main argument is that what we call artificial intelligence is essentially a form of social collaboration. The model is trained on the texts, images, and code of millions of users, and without human input, it would not exist.
This leads to a practical conclusion: artificial intelligence is a tool, not an autonomous entity. An entity can be ascribed will or inherent danger, but a tool is merely a product in a supply chain, where one link remains unpaid.
In 2026, Lanier continues to defend this perspective in key discussions. For instance, during a lecture at Brown University in April, he noted that AI is often discussed as if it were an object, whereas the system is entirely composed of human contributions.
In the May podcast StarTalk, Lanier emphasized that the metaphor of a "black box"—which cannot be examined—is convenient for developers, but in reality, this box is filled with people and their data. He also explained why low-friction digital networks tend to centralize.
In the Existential Hope podcast, Lanier described a "third way" between mass unemployment and universal basic income—creating new categories of creative labor where authors receive fair compensation for their contributions to training models.
On July 24, Irish TechCentral columnist Marie Boran referenced Lanier's ideas in a piece about his StarTalk appearance, noting that his views are becoming increasingly relevant over time.
Price of a Book
On July 20, 2026, a federal court in California approved a $1.5 billion settlement in the Bartz v. Anthropic case, where the company agreed to compensate rights holders for books illegally copied from pirate libraries.
This settlement marks the largest collective resolution in U.S. history regarding copyright infringement. However, the dispute is not entirely settled, as an appeal has been filed concerning the attorney's fees. The settlement covers 482,460 works, with payments averaging around $3,000 each before expenses. Typically, the amount is split evenly between the author and the publisher, with funds distributed in two installments: the first payments are expected by November 15 for works with agreed distributions, while the remainder will follow later.
Publishers are already calculating their shares. Bloomsbury, known for publishing "Harry Potter," reported that 14,087 of their titles are included in the settlement, with estimates suggesting they will receive about $18.7 million—before legal costs.
Yet, from Lanier's perspective, this victory is ambiguous. In June 2025, Judge William Alsup deemed Anthropic's model training as fair use in the case but noted that downloading pirated copies to create a permanent library does not fall under this protection. Thus, Anthropic paid not for commercializing others' ideas but for downloading them from shadow libraries instead of stores. The monetization aspect is tangential: authors received compensation for infringement, not royalties for their intellectual contributions.
The one-time payment of $3,000 seems modest; authors receive it once, while models trained on their texts will generate income for years. Anthropic claims that pirated datasets were not included in commercial models and is obligated to destroy them under the settlement, but it is impossible to independently verify what Claude's weights retain.
Meanwhile, big tech is establishing a second layer of protection—licensing. Reddit is selling access to its API to developers of large language models, while News Corp, Axel Springer, and AP are licensing content directly. The profits accumulate for corporations, leaving users, who have generated terabytes of comments and articles over the years, with nothing.
This agreement highlights the difference between the current copyright protection system and Lanier's proposed model. Rights holders can receive compensation for infringement, but ongoing remuneration for their contributions to AI training has yet to become standard practice.
Has Web3 Found a Solution?
Some Web3 projects are offering mechanisms that align with Lanier's ideas: paying for data provision, aggregating information owners, and automating licensing payments. Since 2019, Brave has shared advertising revenue with users via the BAT token. Ocean Protocol builds a decentralized marketplace for datasets, closely mirroring the MID mediators concept that Lanier and Weyl described in the Harvard Business Review.
Story Protocol allows for the registration of IP assets on the blockchain, setting licensing conditions and automating the distribution of royalties among system participants. Even NFTs at the peak of hype sold the idea of perpetual on-chain royalties for creators upon resale in secondary markets.
Lanier is critical of the cryptocurrency economy, fearing it reproduces capital and influence concentration rather than redistributing benefits as promised.
His criticism later extended to tokenomics: if a token's utility boils down to speculation and anticipation of growth, the claimed "payment for data" becomes a fiction. On StarTalk, Lanier went further: in zero-friction networks—where transactions and copying are nearly free and access is open—the ecosystem inevitably collapses towards the center. This reflects the same gravitational effect of "siren servers" he described in 2013.
Decentralization of a protocol does not guarantee decentralization of profits. Lanier suggests changing the reward model: users should be compensated for their data, not just gain access to free digital services.
Weyl, co-author of the soulbound token concept, is attempting to navigate between paradigms, seeking a hybrid that combines blockchain infrastructure, quadratic funding, and non-transferable reputation tokens.
Insider vs. System
Lanier critiques the business model of the AI industry while remaining a Microsoft employee. For years, he has advocated for a system where people are compensated for the data and content used by digital platforms. He has repeatedly emphasized that working for a corporation does not prevent him from publicly criticizing tech companies.
Lanier continues to uphold his stance: the capabilities of AI are dependent on the labor of people whose contributions often go unnoticed. Legal disputes and licensing agreements illustrate that the issue of compensating content creators is becoming increasingly important. However, the industry has yet to adopt his proposed model of ongoing payments for data usage.