web3 is a stupid idea (🔊)

Tim's website

2020-09-08

blockchain, ethereum, metamask, usability, web standards, web3

| 4 minutes read

A long time ago, I gave a talk in front of a full audience talking about BigchainDB, a company I worked for to create a (scalable) decentralized database. As we just had released our browser-compatible JavaScript driver, enthusiastically, I told the audience: "... and so by using our driver from within the browser, your app won't need a backend anymore"!

That must have been around 2017 when I first discovered Metamask, and started drinking Ethereum's web3 cool-aid. Arguably, web3 quickly became something extraordinary. All of a sudden, users could download a browser extension and directly interact with a public network. In a sense, it still is extraordinary.

If you have an extension like Metamask installed in your browser today, you can visit sites on the web that allow you to do the craziest things with your digital money. An excellent recent example of this are DeFi (short for "Decentralized Finance") websites. They allow a user to engage in trading cryptocurrencies, providing liquidity, and peer to peer lending. With the click of a button and no mandatory signups, you're able to pool thousands of dollars. That is super cool and confirms the viability of the web3 vision.

But actually, what is the web3 vision? It may be that there never was such a thing in the first place. All I know is that someone named a library "web3.js". Developers use it to talk to remote or local Ethereum nodes when working in a browser environment (JavaScript).

On a web3-enabled website, when a user now clicks a button to, e.g., pool ether in a smart contract, most calculations are supported by the web3.js library that periodically talks to an Ethereum node. Ultimately web3.js allows the user to send the transaction to the node to transfer the user's money.

Often, a key-management program, like Metamask, is running on the user's browser. It allows the user to sign transactions with the same key on different websites.

In a nutshell, that's web3. It's supposed to be a play on words regarding "web 2.0". Web 2.0 is the upgrade of web standards that gave us modern single-page applications and dynamic AJAX loading. And Web3? An advancement towards what exactly? Money websites?

Indeed, if you were capable of cleaning your mind of specific memories, specifically, let's say you could do grep -l web3 brain | xargs rm. And then someone asked you how you'd envision a blockchain-based and smart-contract-enabled web3; you'd likely describe an ecosystem vastly different to what it is today. You'd think about peer-2-peer networks, light clients, and renewed web standards. That's precisely not web3.

In today's experience it will instead be mostly shitty react websites that crash or stop working when you've neglected to install Metamask (or other key-management plugins). Opening a web3 website's network console, you'll see that it's making an excessive amount of RPC request to an Ethereum full node. Sorry, I meant to say Infura node, a hugely-popular cloud provider hosting Ethereum full nodes. That's kinda stupid.

And since Metamask allows developers to prompt the user for specific contract calls, what's even more stupid, is that all your money may be at the risk of continually getting stolen with the accidential click of a button. Either by someone hacking the website's server. Or by the website provider becoming corrupt themselves. Or simply because a website pretends to do X when it does Y (stealing all your money).

But instead of continuing to rant, I'd now like to now point out what I think should change about web3:

We should stop building key-management plugins and start thinking about a standardizable web API. We must stop training our users to install shitty browser plugins!
We need to make light clients work as soon as possible and become independent from third-party services like thegraph and Infura.
We need to improve our client libraries (ethers.js and web3.js) by dramatically simplifying them and making them bug-free (god damn it!)!
We need to take advantage of some of the blockchain's fundamental properties. Most data is immutable so let's start caching things.

And finally, I think we should stop focusing all of our attention on bumping the web's version number. Maybe we should reconsider writing more backends. We should promote more work on permissionless networks like Open Gas Station Network that allow developers to upgrade a user's experience. And, we should start thinking of a machine network of blockchains more often. In many ways, web3 was just a cool demo. But let's come up with something better. Just imagine what happens once there's a deeper integration of money into computer systems!

Further reading