The Blockchain Survived. The Product Didn't.

One of the stranger promises of Web3 was that products could outlive the companies that made them.

A normal digital product is obviously temporary. Spotify exists because Spotify keeps paying engineers, licensing music, operating servers and answering support tickets. A bank account works because the bank keeps maintaining the system around it. Even a plain website survives only as long as someone renews the domain and remembers where the database is hosted.

Web3 tried to offer something more permanent. The asset would live on-chain. The smart contract would continue running. Ownership would no longer depend on a company's database or permission. The company might disappear, but the protocol would remain.

Technically, this was sometimes true. The problem was that the product still died.

A token might remain inside a wallet after the marketplace closed. A smart contract might continue to exist after the development team stopped maintaining the interface. An NFT's ownership record might remain perfectly intact while its images, community, utility and cultural meaning slowly vanished. The blockchain survived, but almost everything the user actually experienced was gone.

This is where the language of permanence becomes misleading. Web3 often confused the permanence of data with the permanence of a product. These are not the same thing.

A product is not merely a record stored somewhere. It is the continuous relationship between a system and the people trying to use it. It includes explanations, interfaces, maintenance, security decisions, compatibility updates, customer support and someone paying attention when something goes wrong. That attention is easy to overlook because it usually remains invisible until it disappears.

When a Web3 product shuts down, its founders can sometimes say that the assets are still there. The code is open source. The contracts remain accessible. Users technically still own what they bought.

This can sound reassuring, but it is a little like closing a railway station and telling passengers that the tracks still exist. The tracks may be real. They may even be in good condition. But the ticket system is gone, the timetable has disappeared, the lights are off and nobody knows whether another train will arrive.

Rally was built around the idea that creators could issue their own social tokens and digital assets. When its sidechain shut down, the tokens did not necessarily vanish from existence. But the infrastructure and product environment that made them usable did. The ownership record continued, while the product around it stopped functioning.

Nike's RTFKT presented digital fashion as a new kind of cultural object, connected to identity, community and the future of clothing. When Nike decided to wind down the operation, collectors still had their tokens. Yet what exactly did they continue to own?

A digital shoe is not valuable only because a blockchain says that one wallet owns it. Its value also depends on the designer continuing to develop the world around it, platforms continuing to display it, communities continuing to care and the company continuing to treat the object as part of an active product rather than an abandoned experiment. Once that care disappears, ownership becomes strangely hollow.

The same problem appears in more technical projects. A decentralised bridge may consist of smart contracts spread across several chains, but the usable service can still depend on a small team maintaining validators, interfaces, keys and operational infrastructure. Multichain became an especially dramatic example. The system appeared decentralised from the outside, but essential access and knowledge were concentrated around a single person. When that person disappeared from the organisation, the protocol's supposed independence became much less convincing.

Aztec Connect offers a quieter version of the same problem. Aztec stopped accepting deposits in 2023 and later retired the sequencer it had been operating. The relevant code remained open source, and another party could theoretically run the infrastructure, but ordinary users were still told to withdraw before the official service ended. The protocol could persist in principle while the supported product disappeared in practice.

This is not just hypocrisy. It may be more accurate to say that Web3 underestimated how many layers of a product cannot be decentralised simply by putting transactions on a blockchain.

Users do not interact directly with abstract protocols. They interact with buttons, dashboards, wallet prompts, warnings, documentation and support messages. They need someone to decide what happens when an upgrade fails, when a transaction gets stuck, when a browser changes its security rules or when a new scam starts imitating the interface. The blockchain can preserve a state. It cannot automatically preserve understanding.

This is important because Web3 often treated maintenance as a weakness. If a project required an active company, central interface or responsible operator, then perhaps it was not truly decentralised. The ideal protocol was meant to become autonomous, almost natural, as if it could continue existing without anyone having authority over it.

But products need care precisely because the world around them changes. Operating systems update. Regulations change. Attackers discover new vulnerabilities. Users forget passwords. Standards become obsolete. Other services disappear. Even if the smart contract remains untouched, every connection surrounding it becomes less reliable over time. A permanent contract inside a changing environment may actually become more fragile, not less.

This is where Web3's promise starts to resemble a familiar design mistake. It optimised for the part of the system that was easiest to formalise and ignored the parts that depended on continuous human responsibility.

The transaction could be made immutable. The database could be distributed. Ownership could be represented cryptographically. But who would keep the experience understandable?

Who would explain an unusual permission request? Who would restore confidence after a hack? Who would update the interface when the original wallet stopped working? Who would stop a community from filling with fraud? Who would make sure that the thing still worked for a person who had not spent the past five years learning how blockchains operate?

These questions were often treated as temporary inconveniences that better infrastructure would eventually solve. But they were not peripheral problems. They were the product.

The strongest Web3 projects were often the ones that quietly rebuilt the institutions they claimed to replace. They hired support teams. They operated official interfaces. They created foundations, multisignature committees, grant programmes and governance processes. They moderated communities and formed relationships with regulators. In other words, they rediscovered organisation.

The weaker projects tried to survive on a token, a Discord server and the assumption that the community would somehow take over. Sometimes the community did continue the code. More often, people moved on to the next project with active developers, rising prices and a more energetic social feed.

This also reveals something uncomfortable about digital ownership. We tend to think ownership is an individual relationship between a person and an object. But many digital objects exist only through a surrounding system of recognition. A game item matters because the game continues accepting it. A profile badge matters because the platform continues displaying it. A membership token matters because someone continues organising the membership.

Even physical ownership depends on systems of care, but the object remains physically present when those systems weaken. A chair is still useful after its manufacturer shuts down. A digital sneaker may become a line of metadata pointing toward a file that no interface wants to show.

Web3 did not eliminate dependency. It rearranged it and made parts of it harder to see.

The dependency moved away from a single company database and into wallet software, indexers, RPC providers, metadata storage, token liquidity, browser extensions, governance processes and social consensus. This was sometimes a real improvement. It also created a product whose continued existence depended on many separate systems remaining compatible. Calling the asset permanent could obscure how temporary the experience still was.

This matters beyond the projects that have already shut down. AI and blockchain companies are beginning to make similar promises about autonomous agents that can own assets, make payments and act independently across networks.

Again, the technical record may persist. An agent might have an on-chain identity, a wallet and a history of transactions. But the agent itself may depend on a model provider, a hosted inference service, a specific prompt, a company-controlled permission layer and a team continuously updating its behaviour.

What happens when the model is retired? What happens when the API price changes, the company pivots or the safety system blocks the agent's old behaviour? The wallet might survive. The identity might remain on-chain. But the intelligence connected to it could become unavailable overnight.

We may end up with permanent financial shells surrounding temporary software personalities.

The question is therefore not whether a blockchain can keep a record alive forever. It probably can, at least in some form. The harder question is whether anybody will still take responsibility for making that record useful.

Products do not survive because their data cannot be deleted. They survive because someone keeps noticing when reality changes and adjusts the product accordingly. Someone keeps interpreting, repairing, explaining and deciding that the experience is still worth maintaining.

Web3 promised permanence without having to trust that someone would continue caring.

Perhaps caring was the part that could never be automated away.

Connected reading

• Digital Product Assurance
• The Product Is Becoming an Organization
• The Interface Arrived Before the Institution
• When a Website Loses Its Memory

AI authorship note

This essay began with Kris Haamer asking why so many Web3 projects had shut down and noticing that an on-chain asset can remain technically present after the product around it disappears. AI supported research, drafting, editing and implementation. Kris chose the framing, developed the argument around permanence and care, and approved the article for publication.