Hacker News Reader

Comment by abiro | Hacker News Reader

For those genuinely curious about Web3:

The new thing is that we can now enforce invariants on state transitions in a decentralized manner for a limited set of applications (mostly financial).

With all the capital inflow thanks to the various crypto bubbles there is an insane amount of cryptography and distributed systems research going on to let us build general purpose decentralized databases in the future.

Once we get there, it’ll obviously change all of our assumptions about the internet which are based on companies controlling centralized databases.

View on HN

If I have an IP address I can host anything and anyone anywhere on the internet can access it right now. There's no company controlled centralized database I have to get permission from to host an apache server with cat photos for example. Isn't this problem already solved?

The interesting part is state transitions (mutations). How would your Apache server coordinate mutations to your cat database with other servers when you need to enforce invariants like the maximum number of kittens a cat can have per year?

I... don't?

I'm hosting cat photos for friends to view. There are no invariants. There are no rules. It's purely content I'm posting for other people to consume. It's no different than if I printed photos and mailed them to all the people.

Then you're talking about a different problem.

It sounds like you're offering a solution for a non-problem.

https://news.ycombinator.com/item?id=30683308

This is not a problem anyone has ever had. The entire world is running on either no coordinated mutations, or batched mutations that are spreading over time. If you're found not respecting the mutations that my Apache server has done, the only thing that happens is that you get banned or fined. Cryptobros see the world as absolutely unable of trust, when it's ran for millenia on it.

If you’re running Postgres you have this problem and you gave it a centralized solution.

I can run Postgres with no cryptobro crap required. Huge advantage. If people don't want to trust my instance, then we can start signing blobs so my instance only shovels dumb data around that anybody can verify being authentic. If relationships are required to be verifiable, we can build a PKI.

If you could build PKI in a decentralized, privacy preserving manner for all internet users that’d be huge. So far no-one succeeded.

> The entire world is running on either no coordinated mutations, or batched mutations that are spreading over time.

Every financial transaction that requires settlement between institutions is a “coordinated mutation.”

And while it is true that financial transactions are often synced between institutions via a “batched mutation” approach (especially when coordinating settlement), the delay that batched settlement represents, and the complexity that it introduces to the overall system, is extremely costly.

This is one reason why blockchains have the potential to be extremely valuable—both from a utilitarian point of view as well as from a financial point of view.

Imagine if you could get rid of batched settlement between institutions, because you just didn’t need to do it anymore. What would that be worth?

> the delay that batched settlement represents

Bitcoin transactions can take anywhere from 10 minutes to upwards of 16 hours for minuscule volumes of transactions. It's not solving that problem for certain.

> extremely costly

The cost is built into the transactions. Just like it is with blockchain transactions. Now then, which has higher per transaction costs?

Assessing the future prospects of blockchain technology based on the performance characteristics of Bitcoin is like assessing of the future prospects of general computation based on the performance characteristics of the UNIVAC III.

As a factual matter, settlement costs are not fully embodied by transaction fees, and include such things as, for instance, financing and opportunity costs and the cost of supporting an organization that is more complex than it otherwise needs to be. If settlement delays were entirely eliminated, new business models would emerge that are not even possible under the current regime.

If you want to share control of that data with me—and say that the data represents something valuable (or is itself valuable somehow) and we don’t really trust each other not to tamper with it—then the classic “single server hosting data on the internet” doesn’t really do the trick.

If the number of people needing to collaborate is arbitrarily large, and includes people who don’t know each other and don’t trust each other, then the only option for storing and updating shared state in a tamperproof and durable way is to use a consensus-replicated blockchain.

Or just use certificate signing.

It’s notable that all blockchain systems use cryptographic signatures to authenticate state transitions.

But how do cryptographic signatures protect you from surreptitious deletions?

On their own, they do not. They are a necessary, but insufficient part of the solution.

Does that happen a lot?

Does what happen a lot? People who might not know or trust each other needing to collaborate in updating and maintaining some kind of shared database? Yeah, it happens every day, everywhere, all the time.

In fact, it is exactly what you and I are doing right now. In our case, though, we happen to trust Y Combinator to mediate our exchange.

If our exchange were hosted on a blockchain, we wouldn't need to rely on Y Combinator to keep track of the upvotes and the downvotes, and to preserve our comments in their original form.

We also rely on the mediator for moderating. Imagine the amount of spam if there was no trusted authority to moderate the database.

I'm fine with trusting Ycombinator with this exchange. These interactions are not important enough to warrant a costly amendment to a permanent record.

Can you provide examples where trustlessness is really necessary?

Except, we already have decades of research into general purpose decentralized databases that aren't tied to shitcoins.

I'd be curious to hear about any solutions to enforcing invariants on state transitions in a decentralized manner pre 2008.

You are curious but you are not able to search for it? Oracle worked on it in the 90s.

Worked != solved.

Worked doesn't imply that they didn't solve it. You can still be CURIOUS and look up the patents :) Just because something is new for you (and lets be clear 14 years and counting isn't new) doesn't mean it's new for everyone.

“Curious” was my courteous way of saying that your and the first commenter’s claims are BS.

Well, snarky comments like that are forbidden on HN better stay with the I am to gulible to search argument. The patent was filled in 1998 but I am not helping your CURIOUS mind any further. ;0)

Markov chains (cryptographic signatures on a linked list of transactions) have been around for much longer than 14 years. There's your "decentralized invariants on state transitions".

The "magic" provided by bitcoins was the "proof of work" part - basically trying to create only one "real" trusted branch of the markov chain at any point in time. But that "reality" has been broken multiple times by both the miners and the programers with voluntary and involuntary splits, double spends, etc.

Assuming you mean a Merkle tree (a Markov chain is a statistical concept). A Merkle tree provides an efficient way to compute hashes (not signatures) of data. It provides neither properties that you claim for Markov chains.

You're correct - Merkle tree. Words are hard some days.

> A Merkle tree provides an efficient way to compute hashes (not signatures)

A signature is a hash, in that it is effectively unique for the content. It simply has the additional property of being generated using cryptographic keys, proving that not only is the hashed content what's expected, it was generated using the specified key.

Signing and hashing are specific cryptographic operations with well defined properties. You can’t use them interchangeably if you want to be taken seriously.

A merkle tree that uses cryptographic signing instead of hashing is still a merkle tree.

Though if we really feel it's necessary, we could simply say "a merkle-tree like implementation that uses cryptographic signatures in place of a hash function".

Pedantry about the difference between hashes and signatures doesn't change the validity of the structure as a response to the original question.

I challenge you to build a Merkle tree using a digital signature scheme in place of a cryptographic hash function where a Merkle tree constructed by Alice can be used by Bob without revealing Alice’s secret key.

OK. Challenge accepted.

Alice and Bob both use asymmetric keys, and use one of a thousand different methods to exchange their public keys (hey, they can even go in the Merkle tree). Alice signs a transaction using her private key and publishes it. Bob can now use Alice's public key to verify the signature on the transaction. No private/secret key exposure.

Asymmetric encryption is a fairly well explored field, which is what makes the above both easy to reason about and secure.

Ok well played, you’ve trolled me masterfully.

>The new thing is that we can now enforce invariants on state transitions in a decentralized manner for a limited set of applications (mostly financial).

Lol, this has to be ironic.

Reply on news.ycombinator.com