Hacker News Reader

Comment by proto-n | Hacker News Reader

Proof of stake has little to do with transaction costs. The roadmap to scaling is sharding + layer 2 tech (optimistic and zk rollup, some of it is already live like arbitrum).

View on HN

Why does PoS have little to do with transaction costs? I thought it enabled massive throughput without the need for block computation? Am I wrong?

Layer 2 is a hack given the current constraints, but it comes with its own trade offs that I think are worse.

But storing things off-chain kinda kills the entire concept, doesn't it? If I hack into DigitalOcean I can mess everybody's NFTs by tampering with one database.

> But storing things off-chain kinda kills the entire concept, doesn't it? If I hack into DigitalOcean I can mess everybody's NFTs by tampering with one database.

It's not about storing off-chain, you compute off-chain and post to the network a cryptographic proof of the computation.

Rollups provide something akin to lossless compression proofs of computation. You execute the computation off-chain with high throughput and post a proof to the Ethereum network, this proof contains everything necessary to verify the correct execution of the computation. Sounds mind-blowing but we can thank smart mathematicians for figuring out the cryptography that makes it possible.

This means that we can scale off-chain while enjoying the security guarantees of the blockchain.

These techniques allow scaling the network throughput towards the 1000-4000 transactions per second range. Which together with data-availability sharding will push the throughput towards the 100K transactions per second range.

To be noted that this is not anymore a theoretical construction, these things are live running today on Ethereum. You can check the different solutions as well as detailed explanations of their trade-offs, limitations, etc... Here: https://l2beat.com

NFT definition is stored off-chain. Which means the security you defined on the chain protects only the URI to a random file or some database - which can be modified any time DigitalOcean or the hacker decides to modify. So you're right, we can create quickly and safely keys to the metal front door, while the back door is propped with a stick.

What you are saying is that an certificate of authenticity is useless because you can burn the artwork with a torch. Those two things are mostly unrelated.

I say the certificate of authenticity can be easily forged or destroyed with no way to know what it once contained. And because the entire reason to exist of an NFT is based on this certificate of authenticity (unlike a real world artwork which can be touched and estimated by a specialist), I see a problem right there. Of course this is a problem only if the NFT has an artistic value, which is an entire discussion per se.

But you can't really touch the certificate (the blockchain entry), only the "artwork" behind it (png stored on digitalocean).

What is and isn't art and artistic value, and whether an indistinguishable copy of the mona lisa should be valued the same is an entire discussion, yes, but it isn't tied to NFT-s. By the way when I say indistinguishable I don't neccessarily mean by world class experts, but for example my guests who happen to spot my (hypothetical) fake mona lisa in my living room.

There is no certificate on the chain. There is only an URI to the certificate, see ERC-721.

Edit: yes I have only NFTs as defined in Ethereum in mind. If there are other solutions working better/safer, I am not aware of them...

We may have differing concepts of what a certificate of authenticity is. To me, the certificate of authenticity is what guarantees that it can't be copied. That is the NFT with the uri that lives on the blockchain.

You are right, it cannot be copied, and the URI lives on the blockchain. But the resource pointed by that URI does not. So the server doing the redirection via that JSON can redirect to whatever else, because even in the JSON we only have the asset URI, and a blockchain navigator can only check whether the JSON is according to the schema. So yes, nobody can steal your NFT, but your NFT might tomorrow be instead of a Beeple, a picture of my cat. Rickrolling on a whole new level.

The recommended way to handle this is to use a URI which uniquely identifies the content, such as an IPFS address. Then, while the data is still stored outside the blockchain, anyone who has a copy can trivially prove that it's authentic and you aren't reliant on a particular server. (Just make sure you save/pin a copy of the file on your own system so it doesn't disappear.)

I would agree that an NFT pointing at some plain HTTP(S) URL on some random server without so much as a hash to identify the original content is pretty much worthless.

Okay this is the first time I hear about the IPFS bit. Somehow all I read is missing this point - but it makes fully sense, thank you!

There are also NFTs where the art is “onchain”. See autoglyphs and artblocks

If the whole world uses DigitalOcean for NFTs sure, but maybe we end up in a world where there are many providers. And not one that controls everything (by which point it's not centralized). Similar to right now many crypto is stored in exchanges but not one exchange holds all.

I can hack into those providers as well, right? My worry was that the chain of trust is broken by definition - as soon as you store things outside the chain. I really don't understand why one would dilute the value proposition of the blockchain like that. If I have a Ferrari I shouldn't put monster truck wheels on it only because it can't drive over speed bumps - it kills the "ferrari-ness".

> My worry was that the chain of trust is broken by definition - as soon as you store things outside the chain.

From the chain perspective they never really leave the chain. They are either locked (most L2 solutions such as Bitcoin Lightning) or they are held by someone else (most centralized custodians, including exchanges). So the worst that can happen is that the locked/held coins get "stolen".

> My worry was that the chain of trust is broken by definition

As I understand it, the idea is you put the data itself off chain but keep hashes on chain. Anyone who downloads the data can verify it by checking that the hash matches what is stored on chain.

Uh, no, you store nothing of sorts. Which is the problem. According to ERC-721 (NFT standard) you store exactly three things: the name, a cute pic and the URI to its definition file somewhere else (not even that definition has a hash, not that it would help anything).

PoS is much cheaper computationally than PoW so it absolutely does increase the throughput of a network.

Not really, the computational cost of PoW is not related to the throughput. PoW is not the work that the nodes put into processing transactions, PoW is by definition a useless computation, and it being useless is neccessary for the game theory behind it to work.

You can have the same amount of PoW on very little or very high throughput networks, the two are almost completely orthogonal.

You're right that the specific computation of PoW is supposed to be useless for game theory reasons [0], but the computation requirement itself still delays updating the network (while block size constrains how many transactions can happen between each block update). It takes time for those updates to get published back across the network.

If you drop the computation requirement and move enforcement to validators and slashing you can easily 10x-100x this without off chain layer 2 hacks. I was originally skeptical of this because it gives up the computational security guarantees of PoW in favor of game theory guarantees, but I was ultimately persuaded after reading more about it.

I think there's a place for both, digital gold like BTC with very low throughput and slow transactions for holding value and rarely moving currency, and ETH (and tokens backed by ETH's protocol) for everything else. With maybe some special situations for something like Zcash.

[0]: https://bitcoin.stackexchange.com/questions/97030/beginners-...

> the computation requirement itself still delays updating the network

Yes, but that's intentional and configurable, e.g. the ten minutes of average time between blocks on the btc chain. You could very easily lower that value by modifying the difficulty adjustment formula. The same amount of pow would be done for a unit of time (say 1 hour), but with higher throughput through more frequent blocks. At that point this becomes a discussion about propagation times, network quality and node speeds, all unrelated to pow.

Or am I misunderstanding what you said?

Reply on news.ycombinator.com