Web3 critics misunderstand decentralization

March 19th, 2022

A recent criticism of web3 is that it isn’t actually decentralized, because there are centralized services in the mix, such as NFT marketplaces like OpenSea, and data availability services like Alchemy. 🧵

This criticism is based on a mistaken understanding of what web3 advocates mean by decentralization. I’ll try to explain.

There will be centralized services in web3 just as there were in web1. The key question in web3 is whether the network effects accrue as private goods (as they did in web2) or public goods (as they did in web1).

Network effects are the main source of user and developer lock-in. If I try to leave Twitter, I lose the followers I built up over the years. That’s because in web2 the network effects accrued to private companies like Twitter.

But I can leave my web hosting provider and still keep my inbound links, search rankings, etc by switching my DNS record.

That’s because in web1 the network effects accrued as a public resource thanks to open protocols like HTTP and SMTP and community-owned services like DNS.

The killer app of the internet is networks. The web and email are networks. Social apps like Instagram and Twitter are networks. Marketplaces like Uber and Airbnb are networks.

Network effects are what enable the corporate-owned networks like Facebook and Twitter to accrue dominant positions and command very high take rates (web2 take rates range from 30%-100%).

Blockchains provide a powerful new way to build networks where the network effects accrue as public goods, as they did in web1.

For example, Ethereum NFTs can be viewed as a network where users and NFTs interoperate and form connections. This network is built on the Ethereum blockchain. Users have full control of their data.

There are a variety of centralized services that let you access this network including OpenSea, Zora, LooksRare, etc.

But the network effect doesn’t accrue to these service, as I explain here

There will be centralized services in web3. Indeed, as we saw in web1, publicly-owned networks unlock a massive wave of innovation and entrepreneurship around them.

Entrepreneurs and investors in the 90s knew they could freely build and invest without fear that the network would change their economics or deprecate their access, as happens regularly with privately-owned networks.

A public park in a city attracts passersby which in turn helps nearby restaurants and other businesses. In the same way, publicly-owned networks unlock entrepreneurship and opportunity that wouldn’t otherwise exist.

Web2 is dominated by incumbents who spend vast sums of money extracting data and money from the networks they control. There is a serious risk the internet turns out like, say, broadcast TV with a few dominant corporate-owned services, and no room for startups and new ideas.

The good news is we are still early in the development of the internet and there will likely be many new networks built in the coming decades.

web3 provides a new way to build networks that combines the publicly-owned network effects of web1 with the advanced functionality of web2.

(I was referring to Zora’s website above; note that Zora is also a decentralized protocol).

Side note: DNS is the unsung hero of web1. A mapping between network layer (domain name) and physical layer (IP), controlled fully by users, enabling them to keep centralized services in check with a credible ability to exit.

What’s interesting is how little was actually kept (in web3 terms) “on-chain.” Possible architectural insight for web3 builders (?).

DNS was asked to step up one more time during the RSS vs closed social wars (circa 2007-9), but asking users to purchase and setup a domain didn’t work against ultra low friction web2 social networks.


Originally published on January 23, 2022

Arweave TX
Ethereum Address
Content Digest