[Intermediate] — the non-financial applications, what’s genuinely working, and a method for judging the next proposal.
Gaming
The theory: players own their items, trade them without the publisher’s permission, and carry them between games. Items outlive the game.
What happened: the 2021 “play-to-earn” wave, led by Axie Infinity, produced games where the earning was the gameplay. Player growth was needed to pay existing players — the reward economy depended on new entrants, which is a structure with a known ending. When growth stopped, earnings collapsed, and with them the player base. Axie’s daily active users fell by well over 90% from peak.
The lesson is specific and useful: if the economy is the game, the game ends when the economy does. Players arrived to earn, not to play, and left when earning stopped.
What’s genuinely promising:
- Ownership in games people would play anyway. The blockchain is a settlement layer for a trading economy, not the point of the game. The games taking this approach are slower and better
- True item scarcity and open markets — trade without a publisher taking a cut or banning your account
- On-chain worlds (Dark Forest, autonomous worlds) — the entire game state on-chain, so it’s modifiable, extensible, and outlives its creators. Genuinely novel, and Dark Forest’s use of zk proofs for fog-of-war — proving you moved legally without revealing where — is a legitimately elegant piece of design
- Session keys and sponsored gas finally make on-chain interaction playable → account abstraction
The honest constraints: blockchains are far too slow and expensive for real-time gameplay, so only ownership and settlement go on-chain; cross-game interoperability is mostly aspiration, because it requires games to agree what an item means, which is a game-design problem no chain solves; and players have been consistently hostile to NFTs in mainstream games, with several major publishers reversing announcements after backlash.
Social
The pitch: own your social graph, so you can leave a platform without losing your followers.
Farcaster is the most credible attempt — a sufficiently-decentralised protocol with on-chain identity and off-chain message storage. It has a real, active community, which puts it ahead of everything else here. Lens Protocol puts the graph on-chain more fully, with profiles and follows as NFTs.
The genuine insight: platform lock-in comes from the social graph. Make it portable and the platform must compete on quality. That’s a real diagnosis of a real problem.
The genuine difficulty: network effects are the product. A protocol with 50,000 users doesn’t compete with one that has hundreds of millions, and “you own your data” has repeatedly failed to move ordinary users. Moderation is also unsolved — a censorship-resistant network is also a network that can’t remove abuse, and every deployment discovers this.
Real-world assets
Tokenising treasuries, real estate, invoices and commodities. This is where serious institutional money has actually gone — tokenised money-market funds run into the billions, and BlackRock’s BUIDL is the flagship.
Why it works when other things haven’t: it’s a genuine efficiency gain — 24/7 settlement, programmable compliance, fractional ownership, instant transfer — applied to assets whose value comes from outside the chain rather than from the chain’s own speculation.
The unavoidable caveat: a token representing a house is only worth something if a legal system agrees. The chain settles the token; the courts settle the asset. These systems are therefore permissioned, KYC’d and centralised by necessity — which is fine, and is not what the field originally promised.
Supply chain, credentials, provenance
The perennial enterprise pitch: track goods on a blockchain for tamper-proof provenance.
The unavoidable problem is the oracle problem in physical form. A blockchain guarantees that a record wasn’t altered. It cannot guarantee the record was true when written. Someone scanning a QR code onto a pallet can lie, and the blockchain faithfully preserves the lie forever → oracles.
This is why the 2016–2020 enterprise blockchain wave largely ended in signed database projects. Where the participants are known and a consortium runs the nodes, a replicated database with signed writes gives the same guarantees at a fraction of the cost. Most of those pilots quietly concluded exactly that.
Where it does work: credentials whose issuer is the authority — a university issuing a degree, where the issuer’s signature is the truth and the chain provides permanent, verifiable, revocable publication. That’s a genuine fit.
The test to apply to anything new
When someone proposes a blockchain application, four questions resolve it:
- Who are the participants, and do they distrust each other? If they trust each other, use a database
- Is there an administrator who could be trusted to run it? If yes, use a database
- Does the data originate on-chain, or does someone type it in? If someone types it in, the chain guarantees nothing about its truth
- What breaks if a company shuts down? If nothing, you didn’t need it
Applications that pass all four are rarer than the field pretends, and genuinely valuable when they exist → what blockchains are actually good for.
Key insight
The non-financial applications that work share one property: the thing being tracked is born on-chain. ENS names, on-chain game state, DAO membership and tokens are created and live entirely within the system, so the chain’s guarantees fully apply. Everything that tracks a physical or legal object depends on someone honestly typing it in — and at that point the chain is an expensive, permanent record of whatever they typed.
Related
- what blockchains are actually good for — the full argument
- oracles — why supply chain doesn’t work
- game development — how games are actually built
- identity and naming
Software engineering
Infrastructure & security
Computer science
Hardware & engineering