[abiro]

The new thing with blockchain tech is that we can now enforce invariants on state transitions in a decentralized manner. Paired with non-interactive zero-knowledge cryptography, we can also enforce those invariants while keeping the confidentiality of transactions.

The tech is rapidly improving and it’s pretty easy to see where things are going from here: soon we’ll have general purpose decentralized databases where data is open as open source code is open. In fact, state-of-the-art smart chains can be viewed as special (financial) purpose decentralized databases.

As to why decentralized databases are desirable: imagine if you could fork databases like you can fork code in a completely permission-less manner. This is how web development would look like:

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

[quickthrower2]

This somewhat assumes blockchain can also deliver on privacy. Perhaps full encryption of the data is enough. But who is paying for all the nodes?

Another way to do this without the blockchain is open standards (think… HTML for example!) and software that lets you export. Where possible run this software open source on the desktop or as an open source web app. Think Photopea as the pinnacle here.

The blockchain or no blockchain approachable both suffer when you have a data format per app locking you into that app anyway.

[abiro]

Re costs: it’s decreasing rapidly. Solana is within 100x of DynamoDB prices so costs are soon becoming negligible. At that point costs can be taken over by the app developer or paid by users on a subscription basis. There are also decentralized consensus mechanisms that don’t rely on paying fees (FBA, delay functions). Idk which will succeed.

Re open standards and “software that lets you export”: we’ve had this for the past 30 years and see how it turned out.

Re lock in: no, in a decentralized database, the data schema is public which means it’s trivial to write an app that works with another app’s data.

[quickthrower2]

Thanks. Keeping an open mind! This is the best blockchain argument I have heard. At least in terms of what I value.