5 ms·

Now it gets interesting. For a long time the lack of an open-source server and the associated centralization was a big point of criticism against Ripple. Perso

Now it gets interesting. For a long time the lack of an open-source server and the associated centralization was a big point of criticism against Ripple. Personally I don't think we'll see much grassroots adoption of any cryptocurrency technology that can be categorized as "pre-mined" or anything less than full decentralization, and that includes both Ripple and Mastercoin. Now if they're able to get the existing banking system on board they could have some chance. It sounds like they're handing out large chunks of XRP (the Ripple cryptocurrency) to companies that implement Ripple integration as an incentive.

wmf 13y ago
Ripple is still semi-centralized. If Alice and Bob trust different nodes they may end up with different ledgers and thus they may be unable to transact. The only way to keep Ripple unified is for everyone to trust the same nodes, which is the definition of centralization.
- jaekwon 13y ago
  Yup. And this is why it won't gain the kind of adoption that Bitcoin is getting. I don't trust it any more than I trust a human.
- kylebrown 13y ago
  Everyone doesn't have to trust the same nodes, the network will actually come to consensus with minimal overlap of trusted validators in the UNLs (and more importantly it will detect and halt when the network isn't coming to consensus, instead of forking like the bitcoin blockchain). Besides, the definition of centralization is that there's a central authority with arbitrary power, such as changing balances or freezing accounts.
  - marcell 13y ago
    > minimal overlap of trusted validators Does that mean it does an intersect of sets of trusted validators? How does it determine which sets of trusted validators to use? What if the intersection is the empty set?
    - kylebrown 13y ago
      What do you mean by "it"? There is only a p2p network, with each node acting independently. Nodes which are on multiple other nodes UNLs could be considered the intersection of a set. But it makes more sense to think of it as a directed graph. There's no such think as an "untrusted validator", if you run rippled and your node isn't on anyone else's UNL, then you aren't really part of the network because your votes aren't considered in anyone else's consensus decisions (but you would get real-time access to the proposals and candidate sets). So the set of of validators which gets used is exactly the union of nodes in the UNLs. If you have two networks without any overlap in the UNLs, then you'd get two separate ledgers. http://bitcoin.stackexchange.com/questions/13330/what-is-the-exact-consensus-protocol-ripple-uses http://bitcoin.stackexchange.com/questions/13330/what-is-the... http://bitcoin.stackexchange.com/questions/10227/why-does-ripples-consensus-process-have-a-80-threshold http://bitcoin.stackexchange.com/questions/10227/why-does-ri...
- alwillis 13y ago
  This is incorrect. All Ripple servers go through a consensus process to validate transactions; after which the Last Closed Ledger (LCL) is generated and sent to every server. Every server has a copy of the LCL, so there can't be a case of ledgers being different anyway. Anyone on Ripple can send XRP to anyone else regardless of who they trust; in order for Alice and Bob to transact in a currency other than XRP, the Ripple network will find a path between them to allow the transaction. There a video from Ripple Labs that explains how consensus works: http://ripplefederation.org/videos/how-ripple-works-gateways-and-pathways http://ripplefederation.org/videos/how-ripple-works-gateways....
  - maaku 13y ago
    The video is misleading. There are possible network configurations where nodes keeping different trust lists will end up partitioned from each other (separate lost closed ledgers). OpenCoin, to my knowledge, has done nothing to fix this as it is a problem inherent in the consensus mechanism they are using. The way you fix it is using some other deterministic method for decentralized consensus... like proof of work.
    - kylebrown 13y ago
      Do you have a source for that claim? There is no reason why nodes with different trust lists would become partitioned. Even if a majority of the nodes were malicious, all they could do is prevent the honest nodes from reaching consensus. The honest nodes would effectively stop operating, as every ledger confirmed by the malicious group would be marked as unreliable by the honest group.
      - maaku 13y ago
        Yes, it's in the original bitcointalk thread where Ripple.com was announced. A note about your example: it's not a majority of all nodes, it's a majority of your trusted node list. Which is different for every node.
  - makomk 13y ago
    Nope, it's entirely correct. During the consensus process, each server only listens to servers that its operator has marked as trusted - there is no way for servers outside the trusted clique to affect in the consensus within the clique. This has a few consequences: - Currently, there's a small clique of servers run by OpenCoin Inc and their partner businesses which only trust each other and which everyone uses. The ledger closure process and transaction acceptance are effectively centrally controlled by that group; if your server closes the ledger itself rather than waiting for the official ledger from on high, it will be permanently stuck on its own fork of the network. - Because of the way the ledger closure process works, if you set up your server wrong and end up in a part of the trust network that's poorly connected to the official servers you will end up permanently stuck on your own fork of the transaction history. (In practice, since you can't disagree with the official OpenCoin servers and you can't influence their consensus, you should just set your trust list to the OpenCoin-approved list exactly and forget about peer-to-peer stuff.) - There's a mandatory transaction fee in a virtual currency, XRP, and unlike in Bitcoin the XRP used in that transaction fee is destroyed. The fee is set as part of the ledger consensus process, which is controlled by OpenCoin Inc. They also control the supply of XRP. In effect, there's a central organisation with the power to charge arbitrarily large fees for the use of Ripple.
- alwillis 13y ago
  Nope, that's not how it works; perhaps you should do some research then get back to us? https://ripple.com/wiki/Consensus https://ripple.com/wiki/Consensus is a good place to start; there's also http://bitcoin.stackexchange.com/questions/13330/what-is-the-exact-consensus-protocol-ripple-uses http://bitcoin.stackexchange.com/questions/13330/what-is-the....

wmf 13y ago

Ripple is still semi-centralized. If Alice and Bob trust different nodes they may end up with different ledgers and thus they may be unable to transact. The only way to keep Ripple unified is for everyone to trust the same nodes, which is the definition of centralization.

jaekwon 13y ago
Yup. And this is why it won't gain the kind of adoption that Bitcoin is getting. I don't trust it any more than I trust a human.
kylebrown 13y ago
Everyone doesn't have to trust the same nodes, the network will actually come to consensus with minimal overlap of trusted validators in the UNLs (and more importantly it will detect and halt when the network isn't coming to consensus, instead of forking like the bitcoin blockchain). Besides, the definition of centralization is that there's a central authority with arbitrary power, such as changing balances or freezing accounts.
- marcell 13y ago
  > minimal overlap of trusted validators Does that mean it does an intersect of sets of trusted validators? How does it determine which sets of trusted validators to use? What if the intersection is the empty set?
  - kylebrown 13y ago
    What do you mean by "it"? There is only a p2p network, with each node acting independently. Nodes which are on multiple other nodes UNLs could be considered the intersection of a set. But it makes more sense to think of it as a directed graph. There's no such think as an "untrusted validator", if you run rippled and your node isn't on anyone else's UNL, then you aren't really part of the network because your votes aren't considered in anyone else's consensus decisions (but you would get real-time access to the proposals and candidate sets). So the set of of validators which gets used is exactly the union of nodes in the UNLs. If you have two networks without any overlap in the UNLs, then you'd get two separate ledgers. http://bitcoin.stackexchange.com/questions/13330/what-is-the-exact-consensus-protocol-ripple-uses http://bitcoin.stackexchange.com/questions/13330/what-is-the... http://bitcoin.stackexchange.com/questions/10227/why-does-ripples-consensus-process-have-a-80-threshold http://bitcoin.stackexchange.com/questions/10227/why-does-ri...
alwillis 13y ago
This is incorrect. All Ripple servers go through a consensus process to validate transactions; after which the Last Closed Ledger (LCL) is generated and sent to every server. Every server has a copy of the LCL, so there can't be a case of ledgers being different anyway. Anyone on Ripple can send XRP to anyone else regardless of who they trust; in order for Alice and Bob to transact in a currency other than XRP, the Ripple network will find a path between them to allow the transaction. There a video from Ripple Labs that explains how consensus works: http://ripplefederation.org/videos/how-ripple-works-gateways-and-pathways http://ripplefederation.org/videos/how-ripple-works-gateways....
- maaku 13y ago
  The video is misleading. There are possible network configurations where nodes keeping different trust lists will end up partitioned from each other (separate lost closed ledgers). OpenCoin, to my knowledge, has done nothing to fix this as it is a problem inherent in the consensus mechanism they are using. The way you fix it is using some other deterministic method for decentralized consensus... like proof of work.
  - kylebrown 13y ago
    Do you have a source for that claim? There is no reason why nodes with different trust lists would become partitioned. Even if a majority of the nodes were malicious, all they could do is prevent the honest nodes from reaching consensus. The honest nodes would effectively stop operating, as every ledger confirmed by the malicious group would be marked as unreliable by the honest group.
    - maaku 13y ago
      Yes, it's in the original bitcointalk thread where Ripple.com was announced. A note about your example: it's not a majority of all nodes, it's a majority of your trusted node list. Which is different for every node.
- makomk 13y ago
  Nope, it's entirely correct. During the consensus process, each server only listens to servers that its operator has marked as trusted - there is no way for servers outside the trusted clique to affect in the consensus within the clique. This has a few consequences: - Currently, there's a small clique of servers run by OpenCoin Inc and their partner businesses which only trust each other and which everyone uses. The ledger closure process and transaction acceptance are effectively centrally controlled by that group; if your server closes the ledger itself rather than waiting for the official ledger from on high, it will be permanently stuck on its own fork of the network. - Because of the way the ledger closure process works, if you set up your server wrong and end up in a part of the trust network that's poorly connected to the official servers you will end up permanently stuck on your own fork of the transaction history. (In practice, since you can't disagree with the official OpenCoin servers and you can't influence their consensus, you should just set your trust list to the OpenCoin-approved list exactly and forget about peer-to-peer stuff.) - There's a mandatory transaction fee in a virtual currency, XRP, and unlike in Bitcoin the XRP used in that transaction fee is destroyed. The fee is set as part of the ledger consensus process, which is controlled by OpenCoin Inc. They also control the supply of XRP. In effect, there's a central organisation with the power to charge arbitrarily large fees for the use of Ripple.
alwillis 13y ago
Nope, that's not how it works; perhaps you should do some research then get back to us? https://ripple.com/wiki/Consensus https://ripple.com/wiki/Consensus is a good place to start; there's also http://bitcoin.stackexchange.com/questions/13330/what-is-the-exact-consensus-protocol-ripple-uses http://bitcoin.stackexchange.com/questions/13330/what-is-the....