Ripple competition

[JoelKatz]

What are the incentives for an exchange or a bank to become a gateway ?

The primary one is transfer fees. Every time an asset they issue changes hands inside the Ripple network, the gateway can charge a fee. Bitstamp, for example, charges 20 basis points ($2 per $1000). The net effect is that a gateway's outstanding obligations shrink without them having to do anything. This applies to any transaction made possible by the liquidity the gateway provides, such as cross-currency payments.

Gateways can also charge for any services they provide. This includes deposit fees, withdrawal fees, or even a monthly account maintenance fee if they want. In the future, when interest rates return to more normal levels, a fractional reserve gateway could draw revenue from interest on funds they hold.

https://ripple.com/ripple-gateways.pdf

[1713566185]

1713566185

[1713566185]

1713566185

[1713566185]

1713566185

[raid_n]

I think I'm taking this a bit off topic but what exactly is then the point of having a "de-centralized" exchange ? Either Ripple scales well for arbitrary trusted gateways in which case it does add a valid service or all you are doing is running additional ripple software on existing exchanges at the benefit of those who hold XRP.
If you rely only on a hand full of gateways ripple adds virtually no advantage, on the contrary you now have to trust both the gateways and the software itself to work as expected

Hundreds if not thousands of gateways, to my understanding, is the goal.  

One of the advantages of the distributed exchange, is to allow payments from any currency, to any currency.  The network will find the best path by utilizing this distributed exchange.  
Gateways are also advantageous over exchanges, because their Ripple transactions are all listed in the public ledger.  So at anytime, you can see how many IOU's are outstanding and how much cash reserves the gateway should have.  Good luck getting that from an exchange or bank.  

XRP is advantageous because it can be stored with no counter-party risk and can move through the network, without trust lines.  It can also move into or out of any currency on the network and is known as a bridge currency.  

Let's say you're going to send me payment, which I require in EUR but you want to spend BTC.  It's tough to convert BTC - > EUR and then send it to me, without Ripple.  Takes a long time and it's expensive.  With Ripple you can for instance:

You -> BTC -> XRP -> EUR -> Me

That would be one possible path through the distributed exchange.  And it all happens in seconds, for a fraction of a penny.  

If you're a person, who in the future only plans to deal in Bitcoin transactions, Ripple is of no use to you.  If you're a person that plans to deal in multiple fiat and crypto-currencies, Ripple is an amazing payment network.  

More info here:  http://en.wikipedia.org/wiki/Ripple_(monetary_system)

It would be dishonest if I didn't admit that I have not looked at the actual consensus code in the protocol, but I do know my way around distributed consensus.
Their website gleams with technicalities on IOU's etc but when it comes to how they are reaching consensus the air gets mighty thin.

This is not trivial, in fact it is the essence if Ripple can and will succeed. The way I understand it you have a set of trusted nodes which need to form a majority.

Depending on synchrony assumptions and message creation times I am very much inclined to believe that Ripple will need more than three times (3f+1) as many honest and correct nodes to avoid byzantine attacks.
Now what about network partitioning in the wake of failures? If you have a system with known participants all trying to agree on something you will need a total majority to ensure consensus, else you'll end up with two network partitions both possibly containing different versions of your lovely distributed ledger. With Bitcoin the average block creation time is around 10 minutes which should give enough time even if a major outage occurs to somehow route information to the other partition. When you are doing transactions in seconds time even a short failure of an important connection (undersea cable is cut etc) may cause a network partition. Now what do you do? let the minority partition wait until it is re-joined?

I am very sceptical about ripple because a) It is complex and complex systems are more prone to contain errors and b) the technical problems are very very far from trivial
Solving this on a medium scale system (several  exchange nodes) may work (look at Group Communication or atomic broadcast algorithms, they solve consensus in such instances)
But thousands of nodes? I am VERY doubtful

It would be great to see this solved and perfectly running as the software would then be usable for MANY other applications that require fast distributed consensus


[edit] ok I found the Consensus entry on the wiki, Complicated (who would have thought) Does anyone know if there is a whitepaper / formal proof of correctness? [/edit]

[mmeijeri]

http://ripplescam.org/

'Nuff Said

Eh, that's TradeFortress' anti-Ripple propaganda site. Enough said.

[JoelKatz]

Depending on synchrony assumptions and message creation times I am very much inclined to believe that Ripple will need more than three times (3f+1) as many honest and correct nodes to avoid byzantine attacks.

The only attack that would be possible is a denial of service attack though -- causing the network to not reach consensus. And since each node signs all of its proposals with the key by which it's known, it should be obvious which nodes to stop trusting. So to launch this attack, someone goes to all this trouble to build human trust, they maybe stop consensus for some period of time, but then they lose all that trust and have to start over. It seems unlikely anyone would bother.

Now what about network partitioning in the wake of failures? If you have a system with known participants all trying to agree on something you will need a total majority to ensure consensus, else you'll end up with two network partitions both possibly containing different versions of your lovely distributed ledger.

Correct, but everyone will know the network is partitioned because they won't see the validations and proposals they're expecting. A key Ripple concept is this -- if conditions make it such that reliable operation is impossible, don't pretend that everything is fine.

With Bitcoin the average block creation time is around 10 minutes which should give enough time even if a major outage occurs to somehow route information to the other partition. When you are doing transactions in seconds time even a short failure of an important connection (undersea cable is cut etc) may cause a network partition. Now what do you do? let the minority partition wait until it is re-joined?

Exactly. So you get quicker absolute confirmation unless conditions are such that this is not possible. In that case, you don't get quick confirmations, but at least you don't get confirmations that don't actually confirm anything. If Bitcoin has an extended split with mining power on both sides (as it did during the blockchain split), both sides will think they everything is perfectly fine and will see confirmations that appear perfectly normal in all respects.

I am very sceptical about ripple because a) It is complex and complex systems are more prone to contain errors and b) the technical problems are very very far from trivial

Agreed. Ripple is more complex than Bitcoin, has a larger attack surface, and has not had as long to prove itself.

Solving this on a medium scale system (several  exchange nodes) may work (look at Group Communication or atomic broadcast algorithms, they solve consensus in such instances)
But thousands of nodes? I am VERY doubtful

Actually, the more nodes, the easier. It does take a bit longer, the consensus time goes up with the log of the number of participants you need to encompass most of the weighted trust, but it's much more robust.

Unfortunately, I'm not sure how useful our consensus algorithm would be for other classes of problems. It might be, but it might not be. We get a lot of simplifications from the fact that we were able to structure the problem we needed to solve specifically to make it easy to solve. For example, all a consensus is really established on is which transactions are going to go in a particular set. If two transactions conflict, to avoid a double spend, you just need to agree which comes first (most of which you can do with a deterministic rule). We've removed a lot of the double spend problem by the design of the ledger and transaction mechanics.

For each transaction any node has seen during a ledger window, the network need only come to consensus on whether or not to try to apply it, and can implicitly agree not to apply it simply by failing to agree to apply it. Everything else is handled by deterministic rules known to all nodes.

This is surprisingly simple because:

1) If there is absolutely no conceivable reason why the transaction should not be included, every honest node will agree to include it.

2) If there is some reason why the transaction should not be included, there is no harm in not including it. (If it's valid, every honest node will see no conceivable reason not to include it in the next round anyway.)

And it doesn't matter if invalid or conflicting transactions get agreed on. The deterministic rules will ensure valid results regardless of which transactions are included or when.

[pbody]

Again a lot of people here seem to confuse "Ripple" (the distributed exchange) with "Ripples" (XRP, the native asset on Ripple).

When you "invest" in the latter, be aware that there are many times more than there are currently available to the public still waiting to be distributed, often for free or only minimal amount of work. This is very different to a lot of alternative currencies that are being distributed via PoW or PoS.

I still do not get why they have Ripples other than to just try to get some extra money. The whole thing seems like a big marketing scam. Eventually there may just be Ripple the exchange? They will succeed because of this whole Ripple garbage? That said I do own some because I blindly went in and bought the hype.

Maybe Im just not understanding it? Bitcoins make sense because of the mining. Maybe ripples are what dollars are to gold? Why not just stay with the dollar then?

Is there is a Ripple White Paper?

[JoelKatz]

I still do not get why they have Ripples other than to just try to get some extra money.

There are three main reasons:

1) XRP funds transactions and ledger entries as a means to control spam. For technical reasons, the only currently practical alternative to proof of work is a native crypto-currency built into the system. The problem with proof of work is that attackers will always have advantages over defenders -- attackers can optimize their hardware to do proof of work while legitimate users have to use whatever they have, such as a mobile device.

2) XRP serves as a "lubricant" to improve liquidity. When introducing a new currency into the system, the easiest way to make it liquid is to place offers to exchange it for a currency that's already liquid. Any currency but a crypto-currency in its native environment will have counter-party risk and is likely to be regional. That makes it less suitable for this "universal currency".

3) XRP makes Ripple Labs' business model work, funding the development of Ripple and making giveaways possible both to promote Ripple and to do good. The value produced by Ripple Labs -- the Ripple network, protocol, and software -- belongs to the world, and Ripple Labs neither owns not controls it.

Maybe Im just not understanding it? Bitcoins make sense because of the mining. Maybe ripples are what dollars are to gold? Why not just stay with the dollar then?

Because Bitcoin uses proof of work to secure the currency, you have a problem in the beginning. When Bitcoin was first introduced, Bitcoins were worth zero and transaction volume was zero. There was no way to use transaction fees to fund the proof of work needed to keep the currency secure. The block reward's technical function is to keep enough proof of work going on to make the currency usable (until transaction fees can take over) by providing a financial incentive. Every dollar that buys mining hardware or goes to an electric company to power mining hardware is one dollar of value the Bitcoin ecosystem created that leaves that ecosystem. The Bitcoin economy has to produce millions of dollars in value a day just to hold the value steady. Mining is Bitcoin's biggest downside, taking much of the value Bitcoin creates and giving it to chip manufacturers and electric companies.

And, unfortunately, it's economically impossible for mining to be a wealth-spreading mechanism to the world at large or to people in need. If mining is more profitable than other things people might do, then mining will increase -- thus so will the difficulty, until it's not more profitable than anything else people can do with the resources they have. It will likely always be a near-break-even proposition, profitable only for those who have especially good deals on hardware or electricity, sucking money out of the Bitcoin economy.

To be clear, I'm not saying Bitcoin is broken in any way. I'm just saying that there's no factual basis for seeing mining as a strength.

If we had created Ripple with mining, Ripple Labs could still have the same amount of XRP holdings today. The difference is that we would have had to raise a lot more money and pay it to electric companies and hardware manufacturers instead of being able to use it to build the team that's developing Ripple. It would mean that instead of being able to give away XRP to do good in the world, it would go to whoever was willing to waste the most electricity mining it, people who would likely barely break even. Bitcoin's early day miners who made large profits generally did no better than people who bought Bitcoins and certainly repeating Bitcoin's early mining success stories is nearly impossible. Ripple doesn't need to incentivize people to provide massive amounts of proof of work.

[pbody]

what does Ripple believe ripples will be worth in a year?

[jubalix]

What are the incentives for an exchange or a bank to become a gateway ?

The primary one is transfer fees. Every time an asset they issue changes hands inside the Ripple network, the gateway can charge a fee. Bitstamp, for example, charges 20 basis points ($2 per $1000). The net effect is that a gateway's outstanding obligations shrink without them having to do anything. This applies to any transaction made possible by the liquidity the gateway provides, such as cross-currency payments.

Gateways can also charge for any services they provide. This includes deposit fees, withdrawal fees, or even a monthly account maintenance fee if they want. In the future, when interest rates return to more normal levels, a fractional reserve gateway could draw revenue from interest on funds they hold.

https://ripple.com/ripple-gateways.pdf

fractional reserve gateways...that's scary....that you have said it lends credence that exchanges are fractional reserve as well....

[JoelKatz]

fractional reserve gateways...that's scary....that you have said it lends credence that exchanges are fractional reserve as well....

It certainly can be -- particularly if there's a lack of transparency. In the current economic climate, there's basically no choice but to keep a full reserve. And if you're holding Bitcoins, for the forseeable future there will be no choice.

But, for fiat currencies, that is really only because interest rates are so absurdly low right now. When economic conditions return to normal, it makes sense for gateways to store their fiat reserves in an ultra-low risk, interest-bearing form. It's not that difficult to structure things such that if the ultra-low risk investment somehow fails, the owners lose their capitalization but the customers do not lose their money. This should allow them to charge their customers less and also have more funds to invest in customer service, security, and features that make them more convenient to use.

I don't have a crystal ball though. My prediction of future gateway business models is basically just a guess.

[mmeijeri]

What are the incentives for an exchange or a bank to become a gateway ?

The primary one is transfer fees. Every time an asset they issue changes hands inside the Ripple network, the gateway can charge a fee. Bitstamp, for example, charges 20 basis points ($2 per $1000). The net effect is that a gateway's outstanding obligations shrink without them having to do anything. This applies to any transaction made possible by the liquidity the gateway provides, such as cross-currency payments.

Does this fee for cross-currency payments only apply when you are sending money to a recipient through the rippling system, or also when you explicitly place an order in the order book?

[Sukrim]

As soon as the moeny is moved, the transfer fee is charged, placing an order can move it (if it is immediately taken) or it can just stay and wait (if you try to sell 1 BTC for 5000 USD for example). In the latter case the transfer fee will apply after it actually moves, if you withdrwa that order in the meantime, nothing is charged (besides some tiny amounts of XRP, as with any transaction).

[mmeijeri]

Interesting. Is this fee just the "quality" setting of the rippling mechanism in action, or is it a separate setting?

[JoelKatz]

Interesting. Is this fee just the "quality" setting of the rippling mechanism in action, or is it a separate setting?

It is a separate setting. It's a per-account setting.

[mmeijeri]

It is a separate setting. It's a per-account setting.

Does that mean it only applies to IOU types where the restriction to to verified accounts has been enabled or can all IOUs apply this? I've been looking around on the wiki, but details on this and things like auto bridging are sketchy. I've been toying with the idea of trying to start a gateway that offers EUR IOUs since no one else seems to be doing that in a way that inspires my confidence.

[Entropy-uc]

Mods: Doesn't this thread belong with all the other scam coins?

[JoelKatz]

Does that mean it only applies to IOU types where the restriction to to verified accounts has been enabled or can all IOUs apply this?

All IOUs can have a transfer fee. The restrictions are that it's a per-account setting and that it must be a ratio of the amount transferred.

I've been looking around on the wiki, but details on this and things like auto bridging are sketchy. I've been toying with the idea of trying to start a gateway that offers EUR IOUs since no one else seems to be doing that in a way that inspires my confidence.

Autobridging isn't implemented yet.

[piramida]

It would mean that instead of being able to give away XRP to do good in the world

Unfortunately, it only sounds good. There's human nature, and when given a choice between good in the world and personal good, we always choose latter, being the greedy monkeys we are. But I'll be glad to be proven wrong, although I know that it's not going to happen.

[mckmuze]

If we have a currency crises, or liquidity crises to be more specific, Ripple will be the only place to go for any kind of liquidity. I personally dont think economic conditions will normalize. I think the new currencies and networks will take over as a last resort and skyrocket in value regardless of anyones opinion. I think everyone needs to quit overthinking our current economic situation and realize the traditional monetary systems are due for a schumpeter's gale. Watch as history repeats itself, but this time we have the internet. We are about to be thrust into an interesting world soon. Grab some popcorn enjoy the show.