Topic: Ripple: A Distributed Exchange for Bitcoin - page 15. (Read 66657 times)

Hi Joel

Could you please help me understand ripple

My questions are in this thread, here
https://bitcointalksearch.org/topic/m.1834158


TIA

Correct. If we, for example, change our server to do something nefarious, it would just be ignored just like someone changing their Bitcoin client to send transactions spending Bitcoins they don't have.

One could imagine a scenario where those keys get into the hands of some group bent on destroying the Ripple network. It's hard to imagine what they could do though other than release them in large quantities and cause the price of XRP to drop. But that wouldn't harm Ripple's use as a payment network. In fact, it would just make it cheaper to use.

And, of course, the community of Ripple server operators could react with any technical changes needed. In principle, they can make any change whose advantages outweigh the disadvantages in the eyes of the majority of major validators.

They way in which I understand it is that as soon as the server source code will be made open source, the whole network will completely become decentralized because everybody can run their own Ripple servers/nodes/gateways. OpenCoin does not have to do anything after that to keep Ripple running.
I dont know what you mean by 'central bank keys' though, I guess you are referring to the vast amounts of Ripples (XRP) that they will still be holding? Their stated pla is to distribute those over the course of the next years (decades?) to help solve the initial distribution problem. If at any point in time regulators would threaten to seize those assets (XRP), they could simply move them to an unknown address or distribute them all at once in a faucet like fashion as they have done in this forum already.

OP is right. Ripple has the best chance at becoming the decentralized exchange that everyone is crying for. For MtGox to function, it has to manage both assets (USD and BTC), as well as actually creating the exchange. Don't like MtGox's AML policy? Worried about their handling of BTC? Don't like the lag? What are your other options? MtGox is the jack of all trades, master of none.

But ripple brings together gateways. A gateway may only have one function, such as exchanging BTC for BTC ious. Another gateway may be exchanging USD for USD ious. The gateways don't have to worry about matching orders or creating an exchange. All they have to do is give specie for ious. With each part of an exchange broken down, now you can select where you do business with USD (and USD alone). Another gateway may have BTC security policies that you find attractive.

I'm just getting familiar with ripple.

Does anyone know if it worths to trade BTCs on it?

I have to read up some more, I'm not sure how to top the wallet up with funds (GBP, EUR).

I would be interested in other's opinion.

Cheers

(I'm assuming you mean these two offers both take the same crossing offer and therefore conflict. If they just place two offers, then both offers will just get placed, no problem. If they 'fully conflict' that means one transaction makes the other impossible. If they 'partially conflict' that means it matters which you execute first.)

First, it's important to see that all that is required is for the servers to agree on the ordering of these two transactions. So long as all servers agree which is "first", then there's no problem. The second transaction will fail or execute in some way that doesn't conflict with the first (if it can). And it doesn't matter which one they agree is first really, no outcome is "right" or "wrong". They must, however, agree or ledgers diverge.

Ripple servers have an "open ledger" and a "last closed ledger". Servers process live transactions against their open ledger and snapshot the open ledger when the ledger closes. So these two servers will have different and conflicting open ledgers when the ledger closes. (If they fully conflict, one will have one and one will have the other. If they partially conflict, one will have one first and one will have the other first with different results.)

When the ledger closes, each server proposes the set of transactions it thinks should go in the next ledger. An avalanche algorithm is then used to reach a consensus on what set of transactions should go in that ledger. See the wiki for details on how a consensus on the candidate transaction set is reached. There are three possibilities:

1) If they don't fully conflict, then the servers actually agree that both transactions should go in the ledger. In this case, a deterministic algorithm decides which one is applied first. All servers will agree on the new last closed ledger.

2) One transaction winds up in the consensus set and the other doesn't. In this case, all servers will agree on the new last closed ledger and will apply the other transaction. If it was only partially conflicted, they'll likely agree to apply the other transaction in the next consensus round.

3) Neither transaction winds up in the consensus set. In this case, a deterministic algorithm decides which one is applied first to each server's new open ledger. The other transaction will also get in if the first transaction only partially conflicts it. The servers will agree on the new last closed ledger with neither transaction and will all agree on one or both transactions in the next round.

In all cases, servers will agree on the new last closed ledger and any remaining conflict will be resolved by deterministic algorithms, possibly forcing one or both transactions to be delayed into the next consensus round.

If there are unconfirmed transactions left over, a new consensus round begins immediately. So this process will typically only take about 10 seconds.

The server never assumes that its open ledger is reliable in any way. It just uses it to screen transactions to decide if it thinks they might succeed and therefore should be forwarded and should be included in the server's initial proposal. It also uses it to help clients form chained transactions.

Misterbigg

You understand ripple best

So some questions if you have the time

1. Say the world has a fear of all fiat drowning in overprinting and inflation.  Does it make sense for the average joe to buy and hold ripple to save himself from hyperinflation ? Even With so much ripple in the hands of so few  ?

Or

2. Should the world  forget ripple as an investment or store of value and just move on to "...ripple as a credit card equivalent ... " ( no one believes that a credit card will save him later )

Or

....is it something else .... ?


Thanks

Yes I had the same thought....it seems like a downside for the currency if they are holding a bunch and hoping for appreciation.  In that case, they will probably do whatever they can to push the price higher..    Of course that could be a good thing for people speculating in ripples if they can successfully move the market higher.

That's a pretty easy question. 

Because nobody trusts "opencoin" at all now after what looks like such a clear commitment to their short-term-control wealth over any benefits the original ripple ideas may have had to offer them.  Imagine if Satoshi said in 2009 he would keep 11m coins for his company and open up the door to later coinbase creations, and that open source code would come "eventually"?           

It's not guaranteed. That's a forward looking statement.
[/quote]

Some projects like Bitcoin make claims based on what the software actually does at the time without giving you an honest assessment of whether it would actually scale or even be feasible to run for the typical user in the future[1].  Others like Ripple make claims based on what might be true in the future if the system does indeed scale to a larger userbase.

Then there are serious projects which tell you what they actually do and at least attempt to publicly verify how well they scale before evangelizing their software.

[1] This doesn't hold for bitcoin's claim of "anonymity", since it was never true at all, but I don't remember whether the official site ever claimed that or if it was just glib fanboys on the forums.

Well, even in the wiki server and validating node is used as synonym.

As the title here is about having a distributed exchange, I still wonder if ripple is really up for that task considering there might be a lot (!) of orders compared to transactions. Having only few validators in a market might make it possible to manipulate it, having a lot might be quite expensive memory wise... Also as I understand it I cannot just launch my own little validator at home, also existing validators need to trust mine. What's my incentive beyond being able to maybe cheat to privately run a validator for the mtgox order book and why should mtgox trust my validator?

Edit: if you sort by hash, you again create some kind of mining as there might be different ways to construct a transaction with lower hashes.

The way the Ripple consensus works, when the ledger is closed ALL nodes have come to an agreement on an identical set of transactions which will be applied. The timestamp on the transaction doesn't matter, only that they are both in the same set of transactions applied to the current ledger.

What you mean to ask is "what happens when two identical offers from different people are in the consensus set of transactions?" I don't exactly how Ripple does it, but there has to be some unique characteristic which can order the transactions in a way that cannot be manipulated. Sorting them by their hash would be one obvious way.

That's basically the same as the double spend problem.  https://ripple.com/wiki/Consensus_Graphic

You're assuming XRP has to be something more than a necessity to stop the network being flooded.  Screw XRP.  It's clearly a toy currency, but if you want to replace it as an anti-DOS device then come up with something better.  Bitcoin's distribution via mining was ingenious, but there's no mining in Ripple (ie there's no voting with computing power).

In your example they'd crash the price of XRP, but as I say screw XRP, it's only there out of necessity. 


Sorry, you're right that wasn't you at all.  Just because that's OpenCoin's business model doesn't have to make it their reality.  I do agree that XRP's are the most troublesome portion of Ripple, particularly when trying to sell it to the mindset of the typical bitcoiner, but I'm not sure it's a showstopper.  This isn't bitcoin and it isn't a libertarian's wet dream, it's something in between. 

So in case node 1 gets offer 1 and then offer 2 with identical timestamps and node 2 the other way round, how is this resolved? Do they ask a third node?

How about this outcome: we hoard billions upon billions of XRP. If you fork our implementation and remove the government mandated deanonymization, we will spend all our XRP and crash the price of that alternate chain. It cost us nothing. If instead you remove our initial endowment from the alternate chain, you will make everybody's wallets worthless because everybody has received coins that were initially ours. So there's no possibility for mutiny and you are our bitches.


One the long term XRP is the least common denominator and people will hold it, injecting value, especially if it's stable or slowly appreciating. That's the whole business model, sell people XRP and make them seem like a good investment. They can buy USD IOUs only when they need to cash out.
That wasn't me asking.

In Ripple I believe all orders are limit orders. But you do bring up a good question. Ripple can't support enormous trade volume with millisecond latency. Is this a problem? I don't know.


There is no "server." All validating nodes process the order book using a deterministic algorithm. Since they all have the same ledger and set of proposed transactions, the application of outstanding orders is deterministic. All nodes come to the same conclusion regarding orders, since they began with the same state and use the same rules.

A few seconds might be great for transfers, for trades however... just look at mtgox. Also who is the one running the order book? The server?

With domestic transfers I meant that ripple gateways would probably help in you getting a domestic transfer to your account instead of international wires. Instead of paying wire fees, you'd pay hopefully fewer fees to convert to IOUs that a domestic gateway accepts. Then withdraw there.

I know that ripple doesn't use proof of work for consensus, still there are quite some bottlenecks that I can think of. I really wanna know which parts handle which systems in the whole ripple idea.

Absolutely.  This is where I think the Ripple website doesn't do them justice.  They're well aware of that.  This is a more unified and open system of what's going on at the moment.  And while it's far from completely open due to XRPs it's still considerably more open that what's in place.

Again, completely agree, but the argument is that that's the state we're in at the moment anyway.  For one thing there's no reason that the entities have to be Bitinstant, they can be PayPal or BoA and have whatever regulation they want.  If there are regulated portions they don't have to trust unregulated portions.

Well, I certainly agree that Opencoin can only be trusted to care about Opencoin.  However, the exchange doesn't have to involve XRP (apart from the transaction fee, i.e. one side doesn't have to be denominated in XRP).  To be fair, you're saying you've seen it but you flat out asked where it was a second ago, which is the only reason it was brought up.

I'd really need to do some research and maths.  As I understand it all the fees and figures are set by consensus of the set of nodes that end up most trusted.   So, if there was later a movement against opencoin the fees and reserves could be adjusted by consensus so that whatever had already been distributed was sufficient to run an economy.  That said OpenCoin's control of the XRP supply is a clear avenue of attack against the system.  Also, apart from XRPs all the debts in Ripple represent actual debts and can be removed from the Ripple system if Ripple becomes infeasible.  If ripple is actually no better than what's in place at the moment then there would be no massive problem doing that.

There are only a few outcomes, and they can all be enumerated:

Q: Can OpenCoin hoard all the XRPs?
A: Yes, but if XRPs become too scarce then nodes will use the consensus algorithm to lower the XRP fees and reserve requirements.

Q: What if OpenCoin sells all it's XRPs at once?
A: There is little incentive to do that, since the price would crash and OpenCoin would make less money. Even so, the result would be similar to recent Bitcoin crashes. Mass panic, and then the price recovering. But OpenCoin could only do this once. After that, the market will set the price. The core Ripple functionality of sending fiat-denominated IOUs securely to anywhere on the planet would be largely unaffected.

Q: How should OpenCoin sell the XRPs?
A: To maximize profit from XRPs sales, OpenCoin needs to carefully maintain the price of XRP. They don't want to sell too much XRP at once since that would lower the price and reduce their profits. But they also don't want too sell too little. OpenCoin will likely sell just enough XRPs so that their scarcity does not slow down adoption of the system.

Once the give-aways are all over, the only way for a new user to create a Ripple wallet will be to either buy the reserve XRPs directly from a gateway, or to have the gateway subsidize the reserve requirement as part of the marketing costs of acquiring a new customer. Should OpenCoin sell too few XRPs, the cost for gateways to acquire new customers will exceed the expected return. At this point, there will be strong pressure for all participating nodes to vote to reduce the reserve requirement. If this vote is successful it would instantly liberate some amount of XRPs that were previously sequestered as reserves, thus reducing the price of XRP and also OpenCoin's future profits.