HN Explorer

Litmus tests for using blockchain vs. relational database

4 points by perditus ↗ HN
When should a business use blockchain despite its inefficiency? Honestly, I can't think of any reason to not just use a traditional relational database instead.

Instead of anecdotes I'm interested in any literature exploring this issue, ideally outlining specific features of a use-case that would make it a good fit for blockchain.

Here's an example, but it's far too vague IMO: https://dcicblog.umd.edu/wp-content/uploads/sites/13/2017/06/Smith.pdf (clickable version in comments)

5 comments

[ 3.0 ms ] story [ 31.0 ms ] thread
nostrademons
When you don't trust the other people who may be writing to the data. The advantages of a blockchain are immutability and byzantine fault tolerance: regardless of how malicious your counterparties may be, they can't rewrite the record of the state of the world, because every new record contains a cryptographic proof of the integrity of previous records. Also, this property is symmetric: there is no one person who is "hosting" the data and has full access to it by virtue of having physical possession of the bits. Even if you have physical possession of the bits of a blockchain, you can't rewrite them in any way that the other participants will accept.

Most real-world uses of blockchains (illegal dealings, adversarial relationship with the local nation-state, currency speculation & trading, gambling games) take advantage of this property. If you don't need it, you're almost always better off with an RDBMS, because you pay a heavy price in latency, transaction scalability, and data modeling for it.

Can't think of any business where you wouldn't trust your counterparties? That makes perfect sense, because people today generally do not do business with people they don't trust, because they can't. IMHO most usage of blockchains will happen in new markets doing new things that people would've considered foolhardy before, because its central value proposition enabling transactions that would've been considered foolhardy before.

aey
Generally enforcement has really high legal costs. My partners small business was ripped of by about $10k in product by a Canadian company. There is basically no recourse that wouldn’t cost more.

If a “60 day net” contract could be encoded on chain, and automatically execute as the products are sold, that would have reduced the risks and enforcement costs.

around_here
Dispute resolution will still be just as difficult.
aey
Sure, but if the manufacturers net their share as soon as the item is sold at a store, we reduce the opportunity for fraud.

This won’t work for everything, but it could work well for high mark up items that rely on brand authenticity or warranties.