Around 2021, 'Web3' became one of the hottest terms in the tech world: venture capital flooded in, companies rebranded, and Twitter erupted into heated debates. Yet, listening closely reveals that people are using the same name to discuss two nearly opposite ideas. The Wikipedia entry for 'Web3' still displays a disambiguation note at the top: 'Not to be confused with Web3D or Web 3.0'. The introduction adds: 'This term is sometimes confused with the Semantic Web, as both have been referred to as "Web 3.0"' ([Wikipedia Web3 entry](https://en.wikipedia.org/wiki/Web3)).
One name, two bets. This is not terminological pedantry: the two bets offer opposite diagnoses of 'what’s broken about today’s internet' and place trust on opposite layers. Failing to see this leads to mistaking evidence for one side as a promise from the other.
## Two Origins, Two Diagnoses
The first bet originates from Tim Berners-Lee, inventor of the World Wide Web. In 1999 he proposed the Semantic Web vision; in May 2001, he co-authored '[The Semantic Web](https://www.scientificamerican.com/article/the-semantic-web/)' with James Hendler and Ora Lassila in *Scientific American*, whose abstract states: 'A new form of networked content that enables computers to understand meaning will unleash a revolution of new possibilities.' Detail: The page text accessible here (bibliography, abstract, and visible portions) contains no mention of 'Web 3.0' or 'Web3'; the full article remains behind a paywall and was not read. Calling it 'Web 3.0' is a later retroactive label—the vision itself predates the label.
This later usage is well-documented. On March 12, 2021, Dr. Setrag Khoshafian published '[Can the Real Web 3.0 Please Stand Up?](https://web.archive.org/web/20210312133316/https://www.rtinsights.com/can-the-real-web-3-0-please-stand-up/)' on RTInsights, with a dedicated section titled 'Web 3.0: The Semantic Web', stating that Web 3.0 represents Berners-Lee’s vision for web evolution—centered on understanding meaning (semantics), not merely hyperlinking content.
In his [open letter](https://web.archive.org/web/2017id_/https://webfoundation.org/2017/03/web-turns-28-letter/) marking the World Wide Web’s 28th anniversary on March 12, 2017, Berners-Lee listed three challenges, the first being 'We’ve lost control of our personal data': data is stored in proprietary silos, invisible to users, who thus cannot reap the benefits of controlling their own data. The other two challenges are the ease with which misinformation spreads and the need for transparency in online political advertising. The core diagnosis is 'loss of control': the data still exists—but no longer belongs to you.
The second bet comes from Gavin Wood, co-founder of Ethereum. On April 17, 2014, he published '[ĐApps: What Web 3.0 Looks Like](https://web.archive.org/web/20150816021358id_/http://gavwood.com/dappsweb3.html)' on his personal blog. Its opening paragraph delivers a post-Snowden judgment: 'entrusting our information to organisations in general is a fundamentally broken model'. What he seeks is a 'zero-trust interaction system'. The core diagnosis is 'loss of ownership': it’s not that someone took your data—it’s that you never owned it to begin with.
In a [Wired interview](https://web.archive.org/web/20220210233332/https://www.wired.com/story/web3-gavin-wood-interview/) published on November 29, 2021, Wood compares the Web2 trust model to pre-internet village society: for 500 years, people traded with acquaintances, relying on social networks to guarantee 'these apples aren’t rotten, this horseshoe won’t break in three weeks'. It works in small circles of familiarity but fails catastrophically in a global network of strangers. His prescription is blockchain consensus plus token incentives: platforms and applications should no longer be owned by centralized gatekeepers, but by users—who earn ownership shares by participating in development and maintenance.
## Where Trust Is Placed
This article’s central claim: the two Web3s are not different stages of the same vision, but opposite answers to the question 'At which layer should trust reside?' Existing sources: Wikipedia documents the naming confusion; Berners-Lee’s open letter and Wood’s 2014 blog post each articulate distinct diagnoses. This article’s contribution: reframing the confusion as a design problem.
Berners-Lee places trust at the semantic layer: open standards, combined with user-controlled data storage. His Solid project [describes it](https://solidproject.org/about) thus: imagine an online storage space you control, where information is stored only once and you decide precisely who can access what, and when—like carrying a backpack with many pockets, where apps may open only the pocket you permit, leaving all others private. Wood places trust at the settlement layer: not relying on any organization’s subjective goodwill, but on consensus protocols and economic incentives. Their diagnoses of Web2’s ailment therefore oppose each other: one says 'you lost control', the other says 'you never owned it'. The shared name obscures this divergence.
The strongest counterargument is that the two are complementary—not opposite: the Semantic Web needs a value layer for incentives, while blockchain needs a semantic layer for interoperability. This argument holds merit, but complementarity does not imply identity. Evidence that would overturn this article’s claim: any publicly documented case where either approach has successfully absorbed the other’s core layer. Until such evidence emerges, treating the two as sequential phases of a single movement risks misreading each side’s evidence.
## An Audit from Early 2022
The sharpest critique of blockchain-based Web3 comes from Signal founder Moxie Marlinspike’s January 7, 2022 essay, '[My first impressions of web3](https://moxie.org/2022/01/07/web3-first-impressions.html)'. He first summarizes the prevailing narrative: Web1 was decentralized, Web2 centralized onto platforms, and Web3 will decentralize again. Then he asks: why did centralized platforms emerge in the first place? The answer is simple: 'People don’t want to run their own servers, and never will.' Not even geeks or full-time software organizations want to.
He created an experimental NFT: its image changes depending on the viewer’s IP address or User Agent—displaying one version on OpenSea, another on Rarible, and always showing a large 💩 emoji when viewed in a wallet. 'What you bid on isn’t what you get.' Days later, OpenSea unilaterally delisted the NFT; after delisting, it vanished entirely from his encrypted wallet. He names wallets like MetaMask—marketed as 'non-custodial': though keys reside with users, when connecting to the blockchain, these wallets still rely on Infura. This isn’t unique to NFTs; it’s how the standard is built: nothing specifies what the image 'should' be.
More critically, the access layer. He writes that nearly all dApps interact with the blockchain via just two companies: Infura or Alchemy. These client APIs do not verify blockchain state, and their responses are unsigned: an app asks 'What does this smart contract view function output?', and Infura or Alchemy replies with a JSON saying 'This is the output'—and the app renders it. Vast effort goes into building trustless consensus mechanisms, yet clients blindly trust the outputs of just two companies without further verification.
This article’s claim is stated as a testable proposition. Existing source: Moxie documented the 2022–01 concentration of the access layer in two firms. This article’s contribution: placing it within the same framework as Web1’s failure—'people don’t want to run servers' is a behavioral constant, not a technical flaw. A testable prediction: any Web3 application achieving mass adoption will re-centralize at the access layer—unless protocol-level UX changes alter this constant. Time-bound scope: this is an observation of the situation as of January 2022—not a final verdict on protocols.
## The Strongest Counterarguments Come From Both Sides
Auditing only one side is insufficient. The most systematic positive argument for blockchain Web3 appears in Chris Dixon’s September 26, 2021 essay, '[Why Web3 Matters](https://a16zcrypto.com/posts/article/why-web3-matters/)', published by a16z Crypto. Disclosure: a16z is a leading crypto-focused venture capital firm; Dixon is a partner; the essay explicitly discloses this interest. His definition: 'Web3 is the internet owned by the builders and users, orchestrated with tokens.'
He frames the evolution from Web1 (~1990–2005) to Web2 (~2005–2020) as an 'S-curve bait-and-switch': platforms initially attract ecosystems with open, positive-sum dynamics, then shift to zero-sum extraction upon maturity—Microsoft vs. Netscape, Google vs. Yelp, Facebook vs. Zynga, Twitter vs. third-party clients, Epic vs. Apple. Blockchain’s uniqueness lies in being 'accessible to anyone, yet owned by no one', with tokens enabling early users to share ownership: Uniswap once retroactively airdropped 15% of its governance tokens to early users.
Citing figures from September 2021, he states people sell approximately $100 million worth of ETH-denominated NFTs daily. This is self-reported by the author—not independent audit—and the two figures are not merged. The sharpest rebuttal comes from the other side. On December 21, 2021, Twitter co-founder Jack Dorsey tweeted: 'You don’t own “web3.” The VCs and their LPs do. It will always be beholden to their incentives. It’s ultimately a centralized entity with a different label.' ([The Verge reported](https://web.archive.org/web/20211224164513/https://www.theverge.com/2021/12/21/22848162/jack-dorsey-web3-criticism-a16z-ownership-venture-capital-twitter) the exchange the same day; Dorsey added a jab at a16z: 'It’s between a and z.') Dixon and Dorsey agree on the risk of re-centralization—but disagree on the remedy: one believes token ownership can rewrite incentives; the other believes the VC structure inevitably prevents such rewriting.
Protocol layers are also evolving. Ethereum’s [ERC-4337 Account Abstraction specification](https://eips.ethereum.org/EIPS/eip-4337) (created September 29, 2021, now Final) explicitly states: 'Account abstraction without consensus-layer protocol changes, instead relying on higher-layer infrastructure.' It improves UX using smart contract accounts, paymasters, and bundlers—targeting precisely the same access-layer UX issues Moxie criticized. Note timing: the spec was created before Moxie’s essay—not as a reactive response, but as an existing direction of protocol evolution.
It also introduces new service layers. Whether these layers will themselves re-centralize remains untested at scale; absence of evidence is not evidence of absence—this remains an open question, and a potential source of counterexamples to the above testable prediction.
## The Semantic Web Side Also Has Business Models
Symmetry check: the Semantic Web/Solid side likewise features commercialization. Inrupt is a company [co-founded by Berners-Lee, who serves as CTO](https://www.inrupt.com/), offering Enterprise Wallet Infrastructure. As of this article’s writing (November 2026), its homepage prominently advertises 'AI That Actually Knows Your Customers', paraphrasing its description: using each customer’s data to build a personalized assistant. The behavioral constant 'people don’t want to run servers' is not unique to blockchain pressure: Semantic Web applications may similarly outsource data hosting to commercial services.
The distinction between the two visions lies not in whether one side uses intermediaries and the other doesn’t—but in where trust resides and what users actually control. Taking Inrupt’s current marketing as representative of the entire Solid ecosystem would be inaccurate: this is a time-bound observation—not a final verdict. But it suffices to remind us that re-centralization is a shared challenge both bets must confront.
## Three Questions to Untangle the Shared Name
This article offers an original analytical tool. Existing sources: terminology histories document naming confusion and advise checking authors’ definitions. This article’s contribution: turning that advice into three concrete, operational questions. Upon encountering any claim invoking 'Web3/Web3.0', ask first: Who grants authority—standards bodies, consensus protocols, or platforms? Second: Where is state stored—on user-side data pods, on-chain, or in platform databases? Third: What does the user actually run—a browser, a node, or nothing at all? The combination of answers to these three questions locates the author within one narrative or the other—without requiring prior rejection of any usage.
A more honest attitude toward terminology: there’s no need to endorse one usage while invalidating others—provided readers are clearly informed about each usage’s origin and scope. Applying this tool to this article’s own sources: Berners-Lee’s open letter and Solid place authority in open standards, state in user-side pods, and require users to run applications—not servers; Wood’s 2014 blog post and Wired interview place authority in consensus protocols, state on-chain, and ideally require users to run nodes—though in practice most rely on Infura/Alchemy; Moxie’s audit directly addresses the reality of the third question. These three questions stably distinguish the two narratives. A counterexample would be a claim whose answers span both narratives coherently—requiring revision of this framework.
## What You Can Take Away
For infrastructure and standards builders: a name is not a roadmap. 'Decentralization' is a concrete design choice about *which layer*—not an adjective. To evaluate any proposal, first answer the three questions—then discuss vision. For general readers: next time you encounter 'Web3', ask first which bet the author means—is it 'returning data control to you', or 'distributing ownership via tokens'? Their diagnoses of 'what’s broken about the internet' are opposite; evidence from one cannot be borrowed to support the other.
Join the discussion in the comments: Which 'Web3' claims you’ve encountered map to which narrative under the three-question test? Have you seen any claim whose three answers span both narratives yet remain coherent?
Scope and boundaries: accessed on November 21, 2026; all sources cached that day. Terminology history relies on the Wikipedia Web3 entry (marked 'needs updating' as of May 2025) as a secondary hub; primary sources are the original texts. The full *Scientific American* 2001 article remains behind a paywall and was not read; only bibliographic and abstract metadata were used, with the absence of the term 'Web 3.0' or 'Web3' noted per accessible-text criteria.
Moxie’s critique is time-bound to January 2022; Dixon’s figures are self-reported for September 2021 and are not merged with independent audits; the Inrupt observation reflects its November 2026 homepage marketing copy; ERC-4337’s impact on the behavioral constant lacks large-scale empirical evidence and is treated as an open question. Only one set of Q&A from the *Wired* interview is archived; the rest of the article’s body was not verified—only the cached, field-by-field excerpt is cited. The strongest counterevidence is already embedded in the main text: Dixon’s affirmative argument, Dorsey’s criticism, ERC-4337’s new service layer, and the symmetric Inrupt case.
## Sources
The Semantic Web, Tim Berners-Lee, James Hendler, Ora Lassila, Scientific American, May 2001, https://www.scientificamerican.com/article/the-semantic-web/ (Bibliographic data and abstract verified against page metadata; full text behind paywall not read; retrieved page text contains no mention of 'Web3' or 'Web 3.0')
Three challenges for the web, according to its inventor, Tim Berners-Lee, Web Foundation, March 12, 2017, full text (Wayback snapshot from 2017) https://web.archive.org/web/2017id_/https://webfoundation.org/2017/03/web-turns-28-letter/
ĐApps: What Web 3.0 Looks Like, Gavin Wood, April 17, 2014 (self-published originally on gavofyork blog), full text (Wayback snapshot from August 16, 2015) https://web.archive.org/web/20150816021358id_/http://gavwood.com/dappsweb3.html
What Is Web3, Anyway? (Gavin Wood interview, main headline: "The Father of Web3 Wants You to Trust Less"), Gilad Edelman, WIRED, November 29, 2021, full text (Wayback snapshot from February 10, 2022) https://web.archive.org/web/20220210233332/https://www.wired.com/story/web3-gavin-wood-interview/
My first impressions of web3, Moxie Marlinspike, January 7, 2022, https://moxie.org/2022/01/07/web3-first-impressions.html
Why Web3 Matters, Chris Dixon, a16z crypto, September 26, 2021, https://a16zcrypto.com/posts/article/why-web3-matters/ (Author’s self-disclosed perspective; conflict-of-interest statement included at end of article)
Jack Dorsey says VCs really own Web3, Jacob Kastrenakes, The Verge, December 21, 2021, full text (Wayback snapshot from December 24, 2021) https://web.archive.org/web/20211224164513/https://www.theverge.com/2021/12/21/22848162/jack-dorsey-web3-criticism-a16z-ownership-venture-capital-twitter
Can the Real Web 3.0 Please Stand Up?, Dr. Setrag Khoshafian, RTInsights, March 12, 2021, full text (Wayback snapshot from March 12, 2021) https://web.archive.org/web/20210312133316/https://www.rtinsights.com/can-the-real-web-3-0-please-stand-up/
About Solid, Solid Project, https://solidproject.org/about
Inrupt homepage, https://www.inrupt.com/ (marketing copy as captured on November 21, 2026)
ERC-4337: Account Abstraction Using Alt Mempool, Ethereum Improvement Proposals, Created September 29, 2021, Status Final, https://eips.ethereum.org/EIPS/eip-4337
Web3, Wikipedia, https://en.wikipedia.org/wiki/Web3 (marked in May 2025 as requiring update)