Two things can be true at once. A great deal of what was sold as Web3 was speculation with a technical vocabulary attached. And a few of the underlying ideas were sound, and have persisted after the money left.
Separating the two requires a specific question, asked of every proposal: what does decentralisation buy here, and what does it cost? Applied consistently, that question sorts the field quickly.
The one thing a blockchain is actually good for
A blockchain is a database that multiple parties who do not trust each other can agree on without a referee. That is genuinely novel, and it is the whole of the technical contribution.
The cost of that property is severe: every participant stores everything, throughput is a tiny fraction of a normal database, and each write costs real money. You accept those costs to remove the trusted party.
Which yields a simple test. Is there a trusted party, and is removing them worth a thousandfold reduction in efficiency?
For most applications the honest answer is no. There is a trusted party, everyone is fine with them, and a normal database is better in every measurable way. Sorting the ideas below is mostly a matter of applying that test honestly.
Ideas that survived
Portable identity and credentials
The strongest surviving idea, and the one furthest from cryptocurrency.
Today your identity online is a collection of accounts held by companies. Your professional history lives on one platform, your reputation as a seller on another, and none of it moves. If an account is closed, it is gone.
Verifiable credentials invert this. An issuer — a university, an employer, a government — signs a statement about you. You hold it. You present it to whoever needs it, and they verify the signature without contacting the issuer.
Two properties make this genuinely valuable. Selective disclosure: prove you are over eighteen without revealing your birth date, or prove you hold a degree without revealing your grades. And no phone-home: the issuer does not learn every place you used the credential, which is a real privacy improvement over current federated login.
This is being standardised and deployed, in digital driving licences and academic credentials. Most implementations do not use a blockchain at all — signatures and a public key directory are sufficient, and that is a sign the idea is healthy rather than a sign it failed.
Content addressing
Instead of naming a file by where it is — a URL pointing at a server — name it by what it is, using a cryptographic hash of its content.
The consequences are useful and unglamorous. The name verifies the content: if a byte changed, the hash does not match. It can be fetched from anywhere, including a peer, because any source can be checked. And it deduplicates automatically, since identical content has one name.
This is not speculative. Content addressing is fundamental to how modern software builds work, how container images are distributed, and how version control identifies commits. It predates the Web3 vocabulary and outlasted it.
Local-first data with real sync
Not blockchain-derived, but part of the same reaction against everything living on someone else's server.
The idea: your data lives on your device, works offline, and syncs when it can — with conflict resolution that does not lose your edits. The enabling technology is a class of data structures that merge concurrent changes deterministically.
This has produced genuinely good software. Several well-regarded editors and note applications are built this way, and the user-visible benefit is simple: it works on a plane, it is fast because there is no round trip, and your files are still yours if the company folds.
Open protocols instead of platforms
The observation that social platforms are more valuable to their owners than to their users, because the users cannot leave without losing their audience.
Federated and protocol-based social networks address this directly: your identity and follower graph belong to a protocol rather than a company, so moving servers does not cost you your network. Adoption is modest compared with incumbents, but these systems are running at meaningful scale, and the underlying protocols are being standardised.
Again, no blockchain required. Just an open protocol and portable identifiers.
Every idea on this list solves a real problem — lock-in, verifiability, offline capability, privacy — and almost none of them turned out to need a blockchain. What survived was the goal of decentralisation, not the specific mechanism sold alongside it.
Ideas that did not survive
Worth being direct, because a lot of money and attention went here.
Tokenising everything. Adding a tradeable token to an application was presented as alignment of incentives. In practice it attracted people interested in the token's price rather than the product, and the resulting incentives were misaligned with anything useful. Most such tokens are now worth very little.
On-chain gaming. Games need thousands of state updates per second at zero marginal cost. Blockchains offer a handful per second at meaningful cost. The mismatch is not a matter of engineering effort; it is four orders of magnitude. The projects that shipped kept the game off-chain and put only ownership records on it, at which point the blockchain is doing very little.
Digital collectibles as an asset class. Cryptographic proof of ownership is real. What it establishes is ownership of a token that references a file, which is not the same as any exclusive right to the work. The market discovered this.
Decentralised organisations as governance. Token-weighted voting concentrates control with whoever holds the most tokens, which is a shareholder structure with worse turnout. Most such organisations ended up controlled by a small group, which is what they were designed to prevent.
Storing application data on-chain. Storage on a blockchain is thousands of times more expensive than ordinary storage. Nearly every real system stores a hash on-chain and the data somewhere conventional — which means the data has the availability guarantees of that conventional store, and the blockchain is providing integrity only.
The honest assessment
| Idea | Needs a blockchain? | Status |
|---|---|---|
| Verifiable credentials | No | Being standardised and deployed |
| Content addressing | No | In production everywhere, invisibly |
| Local-first sync | No | Shipping in good products |
| Federated social protocols | No | Running, modest but real adoption |
| Cross-border value transfer | Yes | The original use, still the clearest one |
| Tokenised applications | Yes | Largely abandoned |
| On-chain games | Yes | Technically infeasible at present |
The column that stands out is the middle one. Almost everything that worked did not need the technology it was marketed with.
What to take from it
If you are evaluating something in this space, three questions do most of the work.
- Who is the trusted party you are removing, and does anyone object to them? If the answer is nobody, you are paying a large efficiency cost for nothing.
- What happens when the token price falls? If the system stops functioning, it was a financial product with a technical description.
- Could this be built with signatures, hashes and an open protocol? If yes, it probably should be, and the version that is will be faster and cheaper.
The lasting contribution of this period was not a technology. It was a renewed and reasonable suspicion of building your life inside systems you do not control, and a set of concrete techniques — portable credentials, content addressing, local-first data, open protocols — for doing otherwise. Those are quietly showing up in software people actually use, without the vocabulary, which is usually how you can tell an idea was any good.