Hacker News new | past | comments | ask | show | jobs | submit login
Data Immutability, Verifiability and Integrity Without the Blockchain Overhead (techtake.info)
42 points by sidcool 14 days ago | hide | past | favorite | 11 comments

Really interesting article - would be very fascinated to hear any detail on use cases for hashing an immutable database.

In TerminusDB we have built an immutable DB based on these (our storage implemented in Rust [1]) principles and are now moving towards content addressable hashing and encrypted layers [2]; however, our user community haven't really found many use cases for these features. I'm sure financial services and other such heavily regulated industries need the governance and confidence, but don't have any detail of the likely implementations.

1. https://github.com/terminusdb/terminusdb-store 2. https://github.com/terminusdb/terminusdb-store/blob/master/d...

>Any party explicitly trying to cheat would face the court...

This is the exact issue that currency-based Blockchain systems avoid, at great cost - the essential requirement for a centralized third party to resolve all disputes

So no one seems to answer this when I ask: what about the necessity for security services, to be able to implement and act, such as the Magnitsky Act allows for - punishing known bad actors financially to try to sway bad actors to behave, before perhaps having to escalate things to physical violence-destruction?

So a hash chain. This prevents someone from arbitrarily changing a particular historical value in isolation, but there's nothing preventing the database operator from making a change and regenerating all the hashes from scratch.

The article didn't address that threat. One defense would be to make the entire hash chain available, and let a third party verify that it hasn't changed. Of course you could also replicate the entire database to the third party, so you're just gaining some efficiency and privacy (which might be worthwhile in some cases).

A blockchain on the other hand makes data really immutable. You can't change the data without doing an enormous amount of work, or (more recently) losing a large amount of staked currency.

Aren't there lots of "private blockchains" that work this way? It's basically just a database that is made available for others to audit.

Yes, it’s just old companies like IBM trying to ride a buzz wave. The whole point of a blockchain is that it’s public and verifiable.

IMHO they've correctly assessed the situation that the buzz is the most important feature.

Touche. It’s so hard to get past the buzz in crypto because people are making way too much money on the price of tokens. It’s so distracting. I think we’ll get there eventually.

Exactly. This is no fundamentally different than a traditional time-stamping service. If the time-stamping service is compromised, all of the previous timestamps are invalid.

The only alternative to a blockchain is to publish the latest hash widely in a verifiable place such as in a well-known newspaper.

This is literally a blockchain, one of the four components of bitcoin. I believe it was Stonebreaker’s blockchain dB project that the sci.crypters were referencing, although it wasn’t unique (e.g. git). But we do need a reminder every once in a while that we don’t need to use all components together for every use case.

Where would you see using this over a distributed ledger, or where do you see the distributed ledger still being the better choice?

Applications are open for YC Summer 2021

Guidelines | FAQ | Lists | API | Security | Legal | Apply to YC | Contact