Ripple competition

[JoelKatz]

LOL, what human trust do I need to build? Just validating reasonable transactions from other gateways automatically and being trusted by my bots will make me a gateway that has a say in the "supermajority network consensus" after some time.

If people do stupid things, then bad things will happen. The trick is to figure out how not to do stupid things and how to make it easy to avoid doing them. I agree that automatically granting trust is really dumb, and that's why nobody should ever do that.

There is a very big difference between partitioning between you, a user, and the rest of the network, and  the partitioning of the rest of the network, (especially the major miners), to maintain your connection is way easier than trying to make sure the whole network is well connected, in Bitcoin, only the former case matters, in Ripple, the latter case matters too, or even more, that's why I doubt the claim of "reliable instant transaction" very strongly.

It's certainly harder for you personally to keep the whole network connected, but you don't have to. Everyone else is trying to do that too. Validators will be chosen for reliable connectivity. It seems very unlikely that having most of the network working almost all of the time will be difficult. It is true that Ripple requires this.

Besides, even if my Bitcoin node is completely isolated, it still costs much more to convince me, if I demand to see a sufficient number of confirmations, as long as I remember how many zeros I should expect to see in each newly mined block, a significant hashrate drop by more than 50% is also very detectable, and should make any sensible merchant suspicious.

I agree. And the idea that you could get cut off from 51% of the hashing power for long enough to get six confirmations with what's left is *very* unlikely. All it takes it one working link between the two sides, which would only be missing in the event of some major global disaster.

Ripple actually doesn't seem to fare much better on the detection of the former case either: if you don't keep record of every validator which has a long history of honest validations from time to time(kindly enlighten me if you have some better algorithms), you can't detect if the partitioning takes place because the supermajority you are seeing could just be "phantom validators" operated by a botnet, while with Bitcoin, all you need to know is a number: the hashrate.

You don't care about validators you don't know. And those validators don't care about validators they don't know. This is the number one scenario everyone knows they have to protect against, so the entire scheme by which validators are chosen will be designed to address exactly this.
https://ripple.com/forum/viewtopic.php?f=1&t=3881&p=19423#p19420

With Bitcoin, you can only statistically infer the hashing power you are seeing over time. You can't directly tell how much of the hashing power you are connected to.

[1714707381]

1714707381

[1714707381]

1714707381

[1714707381]

1714707381

[1714707381]

1714707381

[1714707381]

1714707381

[oakpacific]

[

Ripple actually doesn't seem to fare much better on the detection of the former case either: if you don't keep record of every validator which has a long history of honest validations from time to time(kindly enlighten me if you have some better algorithms), you can't detect if the partitioning takes place because the supermajority you are seeing could just be "phantom validators" operated by a botnet, while with Bitcoin, all you need to know is a number: the hashrate.

You don't care about validators you don't know. And those validators don't care about validators they don't know. This is the number one scenario everyone knows they have to protect against, so the entire scheme by which validators are chosen will be designed to address exactly this.
https://ripple.com/forum/viewtopic.php?f=1&t=3881&p=19423#p19420

With Bitcoin, you can only statistically infer the hashing power you are seeing over time. You can't directly tell how much of the hashing power you are connected to.

If I don't, what composes that supermajority that confirms that my trusted validators are not disconnected from the network at large?

[JoelKatz]

If I don't, what composes that supermajority that confirms that my trusted validators are not disconnected from the network at large?

Each of your trusted validators makes this determination based on their trust list and reports it to you in their validations. So you would need a supermajority of your validators to all see a majority of their validators and still not actually have a simple majority of weighted trust. That would require a very broken topology despite it being the very thing that the humans who choose the topology know that they have to prevent.

Most real world failures will be either nearly complete or nearly empty. Dealing with failures near the 50% mark is the hardest case. For example, Bitcoin has no solution to the case where you consider one or two confirmations sufficient and somehow 60% of the hashing power is cut off from you. One could be added, but it would basically require knowing which mining pools have the hashing power, just as Ripple requires some knowledge of which validators have the weighted trust.

We did design the system to detect these failure cases and become unavailable if reliable operation is impossible. Long before poor topology could cause false confirmations, it would cause poor service, prompting humans to address it.

[smoothie]

Ripple carries counter-party risk by form of trusting another person or in fact trusting the actual creators of ripple and their gigantic stash of XRPs.

No thanks.

May as well play in the stock or bond market.

[JoelKatz]

Ripple carries counter-party risk by form of trusting another person or in fact trusting the actual creators of ripple and their gigantic stash of XRPs.

No thanks.

May as well play in the stock or bond market.

If you mean holding XRP as a store of value, we're not suggesting that anybody do that. What we are suggesting is that people adopt Ripple as a payment system. All we can do with our stash of XRPs is drop the value, which just makes the payment system cheaper.

If you mean the counterparty risk of using a gateway, I agree. If you consider this risk unacceptable for all current gateways, you probably shouldn't use Ripple, at least until there are gateways whose risk level you consider acceptable.

This is the price Ripple pays for permitting transactions with arbitrary assets.

[oakpacific]

If I don't, what composes that supermajority that confirms that my trusted validators are not disconnected from the network at large?

Each of your trusted validators makes this determination based on their trust list and reports it to you in their validations. So you would need a supermajority of your validators to all see a majority of their validators and still not actually have a simple majority of weighted trust. That would require a very broken topology despite it being the very thing that the humans who choose the topology know that they have to prevent.

Most real world failures will be either nearly complete or nearly empty. Dealing with failures near the 50% mark is the hardest case. For example, Bitcoin has no solution to the case where you consider one or two confirmations sufficient and somehow 60% of the hashing power is cut off from you. One could be added, but it would basically require knowing which mining pools have the hashing power, just as Ripple requires some knowledge of which validators have the weighted trust.

We did design the system to detect these failure cases and become unavailable if reliable operation is impossible. Long before poor topology could cause false confirmations, it would cause poor service, prompting humans to address it.

But if I don't get to check the signatures all the way down to the last node in the graph, how can I prevent, let's say, a validator cheats on me saying:"Oh, sorry, I was tricked by some nodes on my trust list as well, so I relayed to you this false information they relayed me about the network being good, I have removed them, don't worry.", so he succssfully double-spends me without getting caught? He may well be innocent as well, this is a unverifiable claim.

[JoelKatz]

But if I don't get to check the signatures all the way down to the last node in the graph, how can I prevent, let's say, a validator cheats on me saying:"Oh, sorry, I was tricked by some nodes on my trust list as well, so I relayed to you this false information they relayed me, I have removed them, don't worry.", so he succssfully double-spends me without getting caught?

A validator can't relay false information. Every message is signed by its originator and validations and proposals are currently flooded. (We have plans to optimize this in the future.)

You are correct that bad things can happen if the majority colludes against you. But that's true of pretty much every system. In Bitcoin, you can be screwed if the majority of hashing power colludes against you. In a Democracy, you can be screwed if the majority of voters decide to pass laws that affect you. The difference is that Ripple lets you choose who you will trust. Yes, someone can build trust and betray you once, but then they have to start all over again.

Interestingly, in the case you gave, they still can't really perform a double spend. All the honest participants will agree with you, and that's all any system can do. Even cash can't protect you against that. Say you hold a dollar bill with a particular serial number. Any group can get together and agree not to accept that bill, by serial number. You can still spend that dollar with anyone not in the dishonest group, but you cannot force the group to accept a dollar bill they choose not to accept.

We all know that the main thing we have to avoid is the case where a majority of trust will be controlled by one entity. So the entire system will be designed to make that as difficult as possible. (See my link upthread.)

[oakpacific]

But if I don't get to check the signatures all the way down to the last node in the graph, how can I prevent, let's say, a validator cheats on me saying:"Oh, sorry, I was tricked by some nodes on my trust list as well, so I relayed to you this false information they relayed me, I have removed them, don't worry.", so he succssfully double-spends me without getting caught?

A validator can't relay false information. Every message is signed by its originator and validations and proposals are currently flooded. (We have plans to optimize this in the future.)

You are correct that bad things can happen if the majority colludes against you. But that's true of pretty much every system. In Bitcoin, you can be screwed if the majority of hashing power colludes against you. In a Democracy, you can be screwed if the majority of voters decide to pass laws that affect you. The difference is that Ripple lets you choose who you will trust. Yes, someone can build trust and betray you once, but then they have to start all over again.

We all know that the main thing we have to avoid is the case where a majority of trust will be controlled by one entity. So the entire system will be designed to make that as difficult as possible. (See my link upthread.)

So that's what I was saying, I need to check signatures all the way down to possibly the last node in the reachable network to determine if the network is being partitioned, right?

[octaft]

Bitstamp has XRP available now, but holy crap are they charging a huge premium to buy it! BTC at 431, XRP (according to bitcoincharts) at about 43k per bitcoin, but it only offers me about 33k XRP for the USD equivalent of the current price of a bitcoin.

[JoelKatz]

So that's what I was saying, I need to check signatures all the way down to possibly the last node in the reachable network to determine if the network is being partitioned, right?

You can get as complex or as simple as you like. If you want to be really lazy, you can just pick a few entities that you know are very unlikely to collude against you and just check if they report that the network isn't partitioned in their validations. There's no point in ever checking the signature on a message whose public key isn't already known to you. So the work you have to do goes up with the number of public keys known to you.

With a small network, there aren't that many validators anyway. With a large network, statistical sampling is more reliable, especially since it's automatically two levels. (You are checking whether a supermajority of the validators you trust see a majority of the validators they trust.) We believe the most reliable topology is a hairball, so we're working on schemes that reliably produce them. You also want to avoid too many validators in the same jurisdiction or of the same type of entity (individual, commercial, non-profit, government).

[notme]

So that's what I was saying, I need to check signatures all the way down to possibly the last node in the reachable network to determine if the network is being partitioned, right?

You can get as complex or as simple as you like. If you want to be really lazy, you can just pick a few entities that you know are very unlikely to collude against you and just check if they report that the network isn't partitioned in their validations. There's no point in ever checking the signature on a message whose public key isn't already known to you. So the work you have to do goes up with the number of public keys known to you.

With a small network, there aren't that many validators anyway. With a large network, statistical sampling is more reliable, especially since it's automatically two levels. (You are checking whether a supermajority of the validators you trust see a majority of the validators they trust.) We believe the most reliable topology is a hairball, so we're working on schemes that reliably produce them. You also want to avoid too many validators in the same jurisdiction or of the same type of entity (individual, commercial, non-profit, government).

As a student of graph theory I am wondering if you could describe for me what you mean by a "hairball".  A google for "hairball topology" didn't yield anything except this: http://en.wikipedia.org/wiki/Hairy_ball_theorem and I don't think that's what you mean.

[MJGrae]

Let's get one thing straight here. If RippleLabs were to try and "Do anything they want" with the price of XRP as so many of you like to assert, I would be the happiest man on Earth. Why?

Because I would arbitrage the shit out of every single one of their trades and have the single biggest payday that I have or ever will have, in my life.

To top it off, the value of XRP probably wouldn't end up much different than it was originally.

In order to do "anything they want" with XRP, they would have to own a significant amount of every major currency with a strong market against XRP and simultaneously buy and sell these currencies with XRP in order to get the price to stick. But, they wouldn't want to do that because it would entail changing a significant portion of their XRP balance into another currency which in turn would make the XRP price manipulation less worth doing.

 Price manipulation would cost them far more than it would ever earn them and they know it. They aren't stupid people.

[bb113]

So that's what I was saying, I need to check signatures all the way down to possibly the last node in the reachable network to determine if the network is being partitioned, right?

You can get as complex or as simple as you like. If you want to be really lazy, you can just pick a few entities that you know are very unlikely to collude against you and just check if they report that the network isn't partitioned in their validations. There's no point in ever checking the signature on a message whose public key isn't already known to you. So the work you have to do goes up with the number of public keys known to you.

With a small network, there aren't that many validators anyway. With a large network, statistical sampling is more reliable, especially since it's automatically two levels. (You are checking whether a supermajority of the validators you trust see a majority of the validators they trust.) We believe the most reliable topology is a hairball, so we're working on schemes that reliably produce them. You also want to avoid too many validators in the same jurisdiction or of the same type of entity (individual, commercial, non-profit, government).

As a student of graph theory I am wondering if you could describe for me what you mean by a "hairball".  A google for "hairball topology" didn't yield anything except this: http://en.wikipedia.org/wiki/Hairy_ball_theorem and I don't think that's what you mean.

Hairballs turn complex data into visualizations that are just as complex, or even more so. Hairballs can even seduce us to believe that they carry a high information value. But, just because they look complex does not mean that they can communicate complex information. Hairballs are the junk food of network visualization — they have very low nutritional value, leaving the user hungry.

http://hiveplot.net/


He probably just means supernodes not wanted. It will be interesting to see these schemes fighting against preferential attachment (if I understood rightly). I was under the impression it was accepted as being everywhere when it came to social networks.

[Sukrim]

Bitstamp has XRP available now, but holy crap are they charging a huge premium to buy it! BTC at 431, XRP (according to bitcoincharts) at about 43k per bitcoin, but it only offers me about 33k XRP for the USD equivalent of the current price of a bitcoin.

It is almost always cheaper to buy XRP on the Ripple internal market than through a broker like Bitstamp. You just need the initial amount (a few cents) from them, after that you can withdraw funds to Ripple and trade with your balance there.

[octaft]

Bitstamp has XRP available now, but holy crap are they charging a huge premium to buy it! BTC at 431, XRP (according to bitcoincharts) at about 43k per bitcoin, but it only offers me about 33k XRP for the USD equivalent of the current price of a bitcoin.

It is almost always cheaper to buy XRP on the Ripple internal market than through a broker like Bitstamp. You just need the initial amount (a few cents) from them, after that you can withdraw funds to Ripple and trade with your balance there.

So just get a couple dollars worth to get the required 100 XRP, and then I will be able to buy more on the actual site?

[Coinseeker]

So just get a couple dollars worth to get the required 100 XRP, and then I will be able to buy more on the actual site?

Correct. Just get enough to fund your account.

[BldSwtTrs]

So just get a couple dollars worth to get the required 100 XRP, and then I will be able to buy more on the actual site?

Correct. Just get enough to fund your account.

Sorry for being retarded but what should I do to get a couple of dollar on my Ripple wallet?

I seem i can only fund it via my Ripple address.

[Coinseeker]

Sorry for being retarded but what should I do to get a couple of dollar on my Ripple wallet?

I seem i can only fund it via my Ripple address.

You're not retarded nor do you need apologize.  It was new to us all once.  

I seem to remember I sent $1 or so worth of BTC to my Bitstamp account.  Bought XRP and funded my Ripple account.  Then I could move BTC or USD to and from my Ripple wallet and buy XRP from within the Ripple client at MUCH better rates.

[notme]

Let's get one thing straight here. If RippleLabs were to try and "Do anything they want" with the price of XRP as so many of you like to assert, I would be the happiest man on Earth. Why?

Because I would arbitrage the shit out of every single one of their trades and have the single biggest payday that I have or ever will have, in my life.

To top it off, the value of XRP probably wouldn't end up much different than it was originally.

In order to do "anything they want" with XRP, they would have to own a significant amount of every major currency with a strong market against XRP and simultaneously buy and sell these currencies with XRP in order to get the price to stick. But, they wouldn't want to do that because it would entail changing a significant portion of their XRP balance into another currency which in turn would make the XRP price manipulation less worth doing.

 Price manipulation would cost them far more than it would ever earn them and they know it. They aren't stupid people.

They have enough XRP to buy out every ask on every asset.  Not that I think they would, but they could.  If they did, XRP would be worthless.  How exactly are you going to "arbitrage the shit out of" that?  Are you going to sell all your assets at the new high price for the worthless XRP?

[wxg10]

The ripple guys should just go to an exchange and buy some Gold 2.0 with their premined 10^256 ripples