Skip to content

Why Web3 Is Inevitable, How It Can Go Wrong, and What to Do About It


We believe that general purpose decentralized databases will be ubiquitous in the future, but we see two major obstacles to realizing truly decentralized web applications: The immediate problem is that it's very difficult to experiment with new dapps due to safety and UX issues and this slows down progress. The longer term problem is that wallets need to be reinvented, otherwise the majority of internet users won't own their keys. We try to fix both with SealVault.

What Is Web3

Web3 refers to the emerging shift in internet value chains from permissioned, centralized databases to permissionless, decentralized databases1 where users own2 their data. The driving force behind Web3 is that we can now enforce invariants on state transitions in a permissionless, decentralized manner for a limited set of applications (mostly financial). For an even more limited set of applications, we can also enforce invariants while preserving the confidentiality of transactions.

With hindsight, the two technological breakthroughs that proved decentralized databases possible were Bitcoin and Zerocash. Bitcoin showed that state-machine replication (SMR) is solvable in a distributed, permissionless, partially synchronous setting with Byzantine fault tolerance. Zerocash showed that zero knowledge succinct non-interactive arguments of knowledge (zk-SNARKs) can be used to ensure the confidentiality of transactions in this SMR setting.

While state-of-the-art blockchains in 2022 are best seen as special (financial) purpose decentralized databases, we are fairly certain that general purpose decentralized databases with confidential transactions will emerge in the 2020s, because:

  1. Iterative improvements on the state-of-the-art can likely get us to general purpose decentralized databases.
  2. Since there is an established way to monetize decentralized databases,3 the incentives to build them are very high, and thanks to the recent crypto bubbles, funding is abundant for cryptographers and distributed systems researchers.

Web3 Is Inevitable

We believe that the shift in internet value chains to permissionless, decentralized databases is inevitable, once they become available, because:

  1. Decentralized databases enable permissionless innovation with data.
  2. Synthetic data can be valuable.
  3. Composability is infectious.

Permissionless Innovation

We have already seen the flurry of financial innovation4 in decentralized finance ignited by smart chains, and we'll see a similar innovation explosion in all web applications5 once general purpose decentralized databases become available thus drawing more and more internet users to Web3 in a virtuous cycle. Another way to look at this is that data becomes open as code is open with all its implications for innovation.

Valuable Data

While the value of ownership is already clear to crypto native users, data ownership will draw other users to Web3 once the wealth it creates becomes apparent. Synthetic data can be valuable as proven by thriving black markets for MMORPGs, and given a choice between investing time in applications where data is owned by a third party, users will naturally choose Web3 applications even if the initial user experience is inferior. Viewing blockchain technology as disruptive to centralized databases is a good framework for thinking about this.6

The Law of Potential Composition

Composable data is user data that can be combined with data from untrusted counterparties to produce composable data subject to some invariants. Decentralized databases store and produce composable data.

Fungible tokens are examples of composable data. Eva Lu Ator can send tokens to Alyssa who can add some of her own tokens and send them to Ben and so on ad infinitum.

Non-fungible tokens (NFTs) are also composable data. An on-chain game may allow Eva to use a cartoon NFT to assume control of an advanced character and defeat Alyssa's advanced character to gain experience. Lem E. Tweakit may grow irate with the game and fork it to build a better game which Eva can try out with her improved character and play against Ben's character that was leveled up in the new game. And so on.

In the blockchain world all on-chain data is permissionlessly composable by default. In the Web2 world composable data manifests in user data that is accessible through permissioned APIs.7 So at a minimum we can say that composable data in Web3 will be all on chain data + all user data currently accessible through Web2 APIs.

How do we decide what data to store in a decentralized database? Well, we can rule out certain classes of data like private keys, but if we are uncertain, then it makes sense to store data in a decentralized database in case it becomes composable and thus more useful.9 This is the Law of Potential Composition which is a major driving force behind the eventual ubiquity of composable data and decentralized databases.

How is composable data produced in Web3?

First, a programmer takes a look at a public data schema (e.g. a smart contract that implements the ERC-721 interface) and decides to build on top of it. Then, a user, who has already interacted with what the programmer built on, decides that they like what the programmer built and lets the programmer's app use her data. The programmer can be sure that the data they built on remains available, and the user can be sure that they'll be able to port the data produced by the programmer's app into new apps.

How Can Web3 Go Wrong?

Being an internet user in 2022 is like being a guest in one's home. While we believe that the shift in internet value chains to decentralized databases is inevitable, this will not bring material improvement in the status quo if users don't own their keys.

Users care about products that serve their needs. Decentralized databases enable novel applications that let users port data and make transactions without permission from application developers. This is an extremely strong value proposition, but it is not necessary for users to own their keys to realize it. Consider the local and cloud Web3 value chains sketched below for key management:

graph BT
  local_client(Local Client)
  --> user_need[Port data to novel apps/Transact with valuable data]

  local_tx_mgmt(Local Transaction Management)
  --> local_client

  local_key_store(Local Key Store)
  --> local_tx_mgmt

  cloud_tx_mgmt(Cloud Transaction Management)
  --> local_client

  decentralized_db(Decentralized Database)
  --> local_tx_mgmt

  --> cloud_tx_mgmt

  cloud_key_store(Cloud Key Store)
  --> cloud_tx_mgmt

  classDef userNeed fill:transparent,stroke:yellow;
  class user_need userNeed;

  classDef local fill:transparent,stroke:green;
  class local_tx_mgmt,local_key_store,local_client local;

  classDef cloud fill:transparent,stroke:red;
  class cloud_tx_mgmt,cloud_key_store cloud;

  classDef decentralizedDb fill:transparent,stroke:blue;
  class decentralized_db decentralizedDb;

In the cloud solution, the user's local client initiates transactions through a cloud service that stores the user's private keys. This solution still lets users enjoy the immediate benefits of decentralized databases since the cloud service will let them control their data (at least initially), but users no longer own their data in the exclusive right to transact sense thus shifting power to the cloud provider. The history of the internet has been riddled with the abuse of data lock-in, so this would be a highly undesirable outcome for Web3.

The challenge is that the cloud-based solution is easier for both the user and the provider. Users are familiar with cloud provider profiles and security mechanisms and recovery options are well-established. Similarly, with the cloud solution, there is no need to sync keys among devices and key backup is trivial. The cloud provider can invoice transaction fees through established payment methods like credit cards. For the local solution, payment for transaction costs, sync between devices, backup and recovery are all novel problems that need to be solved.

Fortunately, continued backlash from the crypto community has prevented the emergence of cloud-based key management solutions so far,8 but it's clear that unless we can replicate the ease of use of a cloud-based solution with a self-custody key store, cloud-based solutions will win out as the Web3 user base grows from early adopters to all internet users, especially if cloud providers start subsidizing transaction costs.


Wallets are the self-custody key and transaction management solution in the blockchain world. Wallets generate keys from seed phrases and assume few keys per user.

Wallets have major pain points for Web3 early adopters who regularly use social and gaming dapps and like to experiment with new projects:

  1. Users are worried about trying out new dapps, because malicious dapps can easily steal their stuff.
  2. Wallets use seed phrases for backup and portability. Seed phrases are both annoying and insecure.
  3. Users want their gaming identity, dating identity and shitpost accounts to be isolated. This is challenging with wallets since blockchain data is public and addresses are easily linked publicly with on-chain transactions.
  4. Wallets have too many popups. Most signature approval can and should be automated and when it cannot be automated, users should have simple prompts like “sign in” or “pay” where they can trust the outcome.
  5. Users often want to use the same keys on different devices, but wallets don’t support syncing keys and settings between devices.
  6. It's difficult to launch dapps and multitask in mobile wallets.

It is clear to us that in order to realize truly decentralized web applications, we need to move beyond the wallet paradigm for everyday key and transaction management. This is why we're building SealVault, a new type of key manager that combines password manager and wallet features to provide a safe, smooth and private self-custody Web3 experience.

You can learn more SealVault and the roadmap on the website.

SealVault Logo

  1. Blockchain technology is often described as a new computing paradigm. There is a lot of confusion surrounding this terminology, so it's important to stress that, in the context of web application stacks, blockchain tech impacts what is traditionally done with SQL databases, namely enforcing invariants on state transitions, not "compute" in the "execute some code" sense. The confusion stems from smart contracts that are essentially stored procedures defined in Turing complete languages, so blockchains run computation in this sense, but smart contracts are not the main differentiator in the author's opinion, and they're likely to lose ground to off-chain execution in the future for better scalability. 

  2. “Own” as in “exclusive right to transact”. 

  3. Namely, tokens and transaction fees. 

  4. Yes, there has been a lot of fraud too. 

  5. Which is to say all applications. 

  6. A similar dynamic played out in the transition from desktop apps to cloud apps: The data stored in cloud apps is always available and ready for collaboration, hence it's more valuable than the data in desktop apps. 

  7. The difference between permissioned and permissionless composability is like the difference between websites on an intranet vs websites on the internet. 

  8. For dapps and NFTs at least. For token ownership, the picture is bleaker as many users prefer to hold tokens in centralized exchanges due to the difficulties of self-custody. 

  9. Just as we default to cloud storage over local storage for free backups and the potential for collaboration.