4 ms·

Why? What purpose does a slow, expensive, and energy intensive linked list serve?

by olliej 4y ago

hot_gril 4y ago
It's not all that energy-intensive. Tezos is proof-of-stake.
- olliej 4y ago
  Ah ok, so just slow and expensive? Also what happens if someone decides to do a 50% attack? All the registrations get erased?
  - hot_gril 4y ago
    Slower and more expensive than a regular DB, yeah. But still reasonably quick and cheap for something infrequent like a car title operation, which would normally cost you a lot more in DMV fees. > Also what happens if someone decides to do a 50% attack? All the registrations get erased? I'm not sure exactly what a 51% stake gets to do to contracts. Probably you could show verification of a transferred title then undo that transfer, but not erase titles created before the attack or force new transfers. Anyway, if someone managed to pull that off (which is a big if), Tezos would be sufficiently broken that the DMV/Tezos would have to snapshot the blockchain before the attack and take things back to centralization. There'd be a big disruption, and whatever happened after the attack would be undone.
    - olliej 4y ago
      No, it is definitionally more expensive than a normal db. The entire security model of anything blockchain is “it’s too expensive to get to 50%”, and as such is littered in large losses in the small blockchains where 50% was not too expensive. There is nothing that a block chain does that has any value over a database of you are assuming trusted participants. The only thing block chains do is - questionably - resolve that problem. Now let’s go and look at a dmv example: do we believe that the dmv is going to allow anyone to write whatever they want in this nonsense? Let’s look at other constraints: vehicles can and are seized - when that happens the car owner has no say in the ownership, and so this idiotic application of blockchain nonsense can’t do the other thing it’s ostensibly good “the owner of the real thing controls the blockchain representation of that thing”. Similarly, mistakes are common in the real world, and a typo in the dmv can’t make a car permanently ownerless, so reversing changes has to be supported, so there goes “immutable” (which is always a nonsense goal). Seriously, the sooner these blockchain grifters are gone the better. The only thing blockchain has enabled consistently is crime: fraud, grifting, and ransomware. Certainly the latter is arguably the only example of something that is impossible without blockchain, so I guess blockchains did do at least one thing new.
      - hot_gril 4y ago
        > No, it is definitionally more expensive than a normal db. The entire security model of anything blockchain is “it’s too expensive to get to 50%”, and as such is littered in large losses in the small blockchains where 50% was not too expensive. You're describing a proof-of-work setup. Tezos is proof-of-stake, which tends to be cheaper than PoW. To put numbers on it, a Tezos TZIP-12 (NFT) transfer lately costs $0.02. Minting costs like $0.10. Ethereum on PoW was like $5 IIRC, idk what it is now. Yes, more expensive than a regular DB in any case, that's what I said before. > There is nothing that a block chain does that has any value over a database of you are assuming trusted participants. There are side benefits that have more to do with standardization and transparency. You get a small server/DB that everyone can see and verifiably tell what it's doing, and it's distributed, and it's already running. If you haven't used Ethereum or Tezos, it's worth playing around with. Yes someone could implement that without blockchain, and maybe someone has a proof of concept, but nothing like that has become mainstream like the blockchains have. And the DMV isn't going to do it. So you're right, they don't need a blockchain for this because the DMV is a centralized point of trust in the end (though they're probably sorta federated on the inside). I don't imagine people will be trading titles like NFTs and the DMV will just honor whatever the contract says. I can still imagine this being an viable path forward for them for the other reasons, or just a big gimmick, depending on what they do. > Similarly, mistakes are common in the real world, and a typo in the dmv can’t make a car permanently ownerless, so reversing changes has to be supported, so there goes “immutable” (which is always a nonsense goal). Contract datastores are always mutable in terms of what the last block says. The immutability of blockchains refers to past blocks. If the DMV put titles in a contract, the authorized editors (might just be several DMV offices) would be able to update it, with the edit history publicly visible.

Why? What purpose does a slow, expensive, and energy intensive linked list serve?

by olliej 4y ago

Why? What purpose does a slow, expensive, and energy intensive linked list serve?

hot_gril 4y ago

It's not all that energy-intensive. Tezos is proof-of-stake.

olliej 4y ago
Ah ok, so just slow and expensive? Also what happens if someone decides to do a 50% attack? All the registrations get erased?
- hot_gril 4y ago
  Slower and more expensive than a regular DB, yeah. But still reasonably quick and cheap for something infrequent like a car title operation, which would normally cost you a lot more in DMV fees. > Also what happens if someone decides to do a 50% attack? All the registrations get erased? I'm not sure exactly what a 51% stake gets to do to contracts. Probably you could show verification of a transferred title then undo that transfer, but not erase titles created before the attack or force new transfers. Anyway, if someone managed to pull that off (which is a big if), Tezos would be sufficiently broken that the DMV/Tezos would have to snapshot the blockchain before the attack and take things back to centralization. There'd be a big disruption, and whatever happened after the attack would be undone.
  - olliej 4y ago
    No, it is definitionally more expensive than a normal db. The entire security model of anything blockchain is “it’s too expensive to get to 50%”, and as such is littered in large losses in the small blockchains where 50% was not too expensive. There is nothing that a block chain does that has any value over a database of you are assuming trusted participants. The only thing block chains do is - questionably - resolve that problem. Now let’s go and look at a dmv example: do we believe that the dmv is going to allow anyone to write whatever they want in this nonsense? Let’s look at other constraints: vehicles can and are seized - when that happens the car owner has no say in the ownership, and so this idiotic application of blockchain nonsense can’t do the other thing it’s ostensibly good “the owner of the real thing controls the blockchain representation of that thing”. Similarly, mistakes are common in the real world, and a typo in the dmv can’t make a car permanently ownerless, so reversing changes has to be supported, so there goes “immutable” (which is always a nonsense goal). Seriously, the sooner these blockchain grifters are gone the better. The only thing blockchain has enabled consistently is crime: fraud, grifting, and ransomware. Certainly the latter is arguably the only example of something that is impossible without blockchain, so I guess blockchains did do at least one thing new.
    - hot_gril 4y ago
      > No, it is definitionally more expensive than a normal db. The entire security model of anything blockchain is “it’s too expensive to get to 50%”, and as such is littered in large losses in the small blockchains where 50% was not too expensive. You're describing a proof-of-work setup. Tezos is proof-of-stake, which tends to be cheaper than PoW. To put numbers on it, a Tezos TZIP-12 (NFT) transfer lately costs $0.02. Minting costs like $0.10. Ethereum on PoW was like $5 IIRC, idk what it is now. Yes, more expensive than a regular DB in any case, that's what I said before. > There is nothing that a block chain does that has any value over a database of you are assuming trusted participants. There are side benefits that have more to do with standardization and transparency. You get a small server/DB that everyone can see and verifiably tell what it's doing, and it's distributed, and it's already running. If you haven't used Ethereum or Tezos, it's worth playing around with. Yes someone could implement that without blockchain, and maybe someone has a proof of concept, but nothing like that has become mainstream like the blockchains have. And the DMV isn't going to do it. So you're right, they don't need a blockchain for this because the DMV is a centralized point of trust in the end (though they're probably sorta federated on the inside). I don't imagine people will be trading titles like NFTs and the DMV will just honor whatever the contract says. I can still imagine this being an viable path forward for them for the other reasons, or just a big gimmick, depending on what they do. > Similarly, mistakes are common in the real world, and a typo in the dmv can’t make a car permanently ownerless, so reversing changes has to be supported, so there goes “immutable” (which is always a nonsense goal). Contract datastores are always mutable in terms of what the last block says. The immutability of blockchains refers to past blocks. If the DMV put titles in a contract, the authorized editors (might just be several DMV offices) would be able to update it, with the edit history publicly visible.