What Is Web3? The Pitch, What Shipped and the Critiques

In December 2021, three weeks after stepping down as chief executive of Twitter, Jack Dorsey posted his verdict on the movement half the tech industry was calling the future of the internet: "You don't own 'web3.' The VCs and their LPs do." It came from a man who co-founded one of the defining platforms of the era web3 was meant to replace, and it has been quoted ever since.

He had a point. He also skipped the parts that work. Web3 is the name for a proposed third act of the web, one where your money, your identity and your data live in a crypto wallet you control rather than in a platform's database. Some of it runs today and gets used daily, some of it is still a pitch deck. This guide is for telling those two apart: an explainer, not financial advice.

The pitch: read, write, own

The term comes from Gavin Wood, an Ethereum co-founder who coined it in 2014 and later started Polkadot. His history of the web runs in three acts. Web1 was read-only, pages you looked at, and Web2 is read-write: platforms where you post, follow and message, and where the platform keeps the account, the audience and the data. Web3 adds owning: your login is a wallet, your assets are tokens, and the record lives on a blockchain no single company can edit. Wood made that case in 2014, before Ethereum itself had even launched.

One housekeeping line: this is not Tim Berners-Lee's older "Web 3.0" semantic-web idea. The lineage is different, the name is the only overlap.

The sales pitch runs like this: instead of renting an audience from a platform, you carry your account with you, and if an app turns hostile you walk, taking your name, your followers and your money along. Anyone whose account has ever been locked or deleted can hear the appeal.

What has actually shipped

Strip away the conference talk and a working list remains. It is not nothing.

• Dapps and wallet login: a dapp is an app whose back end runs on a blockchain, so you do not create an account, you connect a wallet, and standards like Sign-In with Ethereum turn one wallet into a username that works across many apps.

• Stablecoin transfers: tokens pegged to a currency, moving between wallets at 3am on a Sunday, no bank involved.

• On-chain exchanges: Uniswap has matched trades with code rather than an order desk since November 2018.

• Portable usernames: the Ethereum Name Service launched in 2017, and an ENS name turns a 42-character address into something like yours.eth.

• DAOs: online groups coordinating shared treasuries through multisig wallets that need several keyholders to approve a payment.

• NFTs as provenance records: a public entry showing who issued a digital item and which wallet holds it now.

The money arrived before the users

Venture capital flooded in through 2021 and 2022, Andreessen Horowitz alone raising a $4.5 billion crypto fund in May 2022, the largest to date. All that money built infrastructure quickly, and it also armed the first big critique.

Dorsey's objection: who ends up owning it

Back to that December 2021 post: Dorsey was not saying the technology fails. His claim was that ownership had concentrated before most users arrived: venture funds and insiders hold outsized slices of many tokens, so "the users own the network" drifts towards "the early money owns the network". His own enthusiasm stayed with bitcoin, an older system with no venture allocation at all.

Taken seriously, it partly stands: token ownership is real, you can hold an asset in your own wallet that no platform can confiscate. But distribution in many projects is lopsided, and voting power in token-governed systems follows the same curve. Owning a sliver of something a few large holders steer is ownership, just not the kind on the poster.

The Signal founder's experiment

The second critique landed in January 2022, when Moxie Marlinspike, the cryptographer behind the Signal messaging app, built a couple of dapps to see the machinery for himself, then published what he found. Most apps never talk to a blockchain directly, they call a couple of centralised API companies, chiefly Infura and Alchemy, and trust whatever those servers say. His reasoning was blunt: people do not want to run their own servers, and never will. Running blockchain infrastructure is work, so apps outsource it, and the outsourcing pools in a few hands.

His test NFT sharpened the point: OpenSea delisted it, and it promptly vanished from wallets that relied on the platform's API. The token still sat on-chain, the wallets just stopped showing it. Decentralised in principle, choke-pointed in practice.

The honest scoreboard

Token ownership works, and it is concentrated, exactly as the December 2021 jab said.

The daily-use layer, stablecoins, Uniswap, ENS names, DAO treasuries, runs now and has for years.

The grand version has not arrived: Sign-In with Ethereum works, yet most people still log in with Google, and carrying your whole social graph between apps remains mostly demos and small networks. The plumbing problem is improving slowly: more independent infrastructure exists than when Marlinspike wrote, but the choke points he named are still there.

The part worth learning either way

Whichever way the label goes, the skills underneath it keep their value: holding your own private keys, reading addresses carefully, checking a token approval before you sign it. Those work on Ethereum today and will work on whatever ships next. The wallet that signs into a dapp is the same wallet that holds coins.

Funding that wallet is the entry step: an on-ramp converts ordinary money into crypto, and Banxa has run that plumbing since 2014, with more than 100 payment methods across 100-plus countries, in the markets it serves.

Start with the wallet, the label can mind itself.