We believe we fixed the bug that was causing paths not to be found or inadequate paths to be selected. Please let me know if you're still experiencing the problem.

Yes, if you trusted Bitstamp. The payment would still have taken place, but it would have used Bitstamp IOUs and cost a tiny bit more.

Hmm, it sounds like that last send should have worked. When TTBit sent .15 BTC to nelisky, that was likely TTBit/BTC that got sent, since nelisky takes those. So when nelisky tries to send .1 BTC to Bitstamp, nelisky must use TTBit IOUs through TTBit to Bitstamp. Did TTBit hold a little over .1 bistamp BTC? Perhaps TTBit just didn't have enough Bistamp/BTC IOUs.

It could also be that our pathfinding found an incorrect or inferior path. We have some pathfinding issues that we're still working on. If you PM me the actual amounts and Ripple addresses, I can actually manually ask the system to compute the various paths and parameters and see if it's doing the right thing.

There's currently a 50 XRP reserve per trust line. You just have to have the XRP in your account and they can only be used for transaction fees until the trust line is zeroed. Then you get them back.

The XRPs allow you to pay transaction fees. The idea is simply that they're not available in infinite supply so someone who wishes to clog the network with meaningless transactions (say moving IOUs back and forth between accounts they control) won't be able to congest the network forever. In periods of transaction overload, the Ripple network raises the required transaction fee, essentially auctioning off the available transaction slots. If an attacker wants to keep your transaction locked out, he must put more XRP behind each of his clogging spam transactions than you are willing to put behind your one important transaction.

Transaction fees are destroyed. XRP are used for transaction fees because they're the only currency Ripple can manage at the network level without requiring a central authority to hold or destroy them and with no risk that an attacker can put himself on both sides of the transaction fee payment.

Currently, the normal transaction fee in Ripple is set to 100,000 transactions to the XRP. (Just like in Bitcoin though, individual nodes can refuse to relay a transaction if they think the fee is too low or they are locally overloaded.)

Actually, the comparison between these risks is not as simple as you make it seem, especially because you have to take into account not just the probability of the risk occurring but also the magnitude of the damage if it does. But all I was doing was disputing the claim that Bitcoin is zero risk.

I certainly agree that if you want to pay Bitcoins to someone who wants to accept Bitcoins, there isn't much of a reason to use Ripple to do it.

Streamlining and simplifying is critical to mass adoption, I agree. Bitcoin has had similar problems in the past, made great strides towards improving, and I think it has made a large difference.

It's not zero risk. There's a risk that a 51% attack will devalue the currency. There's a risk that a speculative bubble will collapse and devalue the currency. There's a risk that Bitcoin will collapse and leave you holding something that's worthless. These aren't high risks, of course, and it may make perfect sense to take them. But it's not zero risk.

No, that's a misunderstanding of how Ripple works. You can only lose money if someone you chose to trust betrays that trust. Other people defaulting cannot harm you.

Payment paths only exist for the instant a payment is made. After that, it's just balances between people who have chosen to trust each other.

To use a check analogy, say you pay your landlord with a $400 check. He deposits that check in his bank. Now your bank owes you $400 less and his bank owes him $400 more. So he considers you to have paid him $400. But if his bank collapses, that doesn't harm you. He chose to trust his bank and his bank accepted the check. It's only if your bank collapses that you may be out money.

You gave him a $400 check that was acceptable to a bank he chose to trust. That ends your end of the bargain. Him getting the $400 from the bank that he chose to trust is his problem. And, of course, my bank paying his bank is his bank's problem, not mine or his. If his bank agreed to accept checks from my bank, then if that goes wrong that's their problem.

You can never wind up being owed money by anyone you didn't choose to trust.

That's how we're pitching it because we think that's the use case that appeals to the most people in the short term and solves expensive, real-world problems that people actually have today. Long term, people will use it however it best serves them and that's very hard to predict.

I guess I don't understand why this particular issue is so confusing. Maybe I've done a bad job of communicating.

We're pitching Ripple as a payment system where most people extend "trust" only to a gateway. People using the system this way can make and receive payments and they don't ever owe anyone anything and their friends don't ever owe them anything. We believe this is the best use for Ripple for the most people in the short term. Ripple can provide numerous advantages over conventional payment systems such as irreversible payments, high-speed confirmations, cross-currency payments, low fees, and so on.

Yes, Ripple can also be used other ways such as to extend social/community credit to friends. But we're not expecting people to actually do that to drive the system. Personally, I think long term that may become Ripple's killer application and may change the way people think about money.

Consensus. See this StackExchange answer: http://bitcoin.stackexchange.com/a/7563/85

I agree that the case for using Ripple to pay in Bitcoins is not very strong. Ripple makes more sense as a way to make fiat currencies act more like Bitcoins. However, there are a few reasons:

1) Ripple has faster transactions.

2) Ripple allows you to pay in dollars even if the recipient wants Bitcoins.

3) Ripple allows you to receive dollars even if you are paid in Bitcoins.

4) Ripple offers the ability to use community credit to borrow or lend Bitcoins.

As I said, the case is not very strong.

One nice thing though -- every merchant who accepts Ripple will be another merchant you can pay in Bitcoins. Once Bitcoin inbound gateways are set up, you'll be able to take a merchant's Ripple payment information, get a Bitcoin amount and account to pay from the gateway, and then the gateway will deliver the Ripple payment once you make the Bitcoin payment.

I agree, it's very unclear. When I say we're not promoting Ripple as a lending platform, I mean by the common sense definition of lending. Technically, it is lending. For example, when you have $500 in the bank, technically you have loaned the bank $500. But we don't think of it as lending.

For you to have the Ripple equivalent of $500 in the bank, a gateway must announce that it owes you $500. But that's not a lending platform.

You are 100% right. We do need to do this. Explaining bits and pieces of understanding that people are missing is no substitute for laying the entire system out in a clear and organized way.

Yes, you missed the whole point! Ripple is a *payment* system that allows you to pay people in fiat currencies like dollars in much the same way you can currently pay people with Bitcoins.

We're not promoting Ripple as a lending platform.

You can lend under any terms you want. Ripple doesn't compel any particular terms. The actual agreements are outside the scope of the system.

Gateways typically lend against 100% collateral, which makes it pointless to charge interest. If I hold $100 of your money and loan you $100, there's no reason for me to charge you interest because that interest equals the interest on the $100 I hold. That's the best way to build a payment system.

But Ripple fully supports other terms. You could charge someone interest instead of holding cash as collateral. You could charge an issue fee and make it settle on demand. You could allow the balance to float in either direction for mutual paying convenience and not settle. Whatever.

Ultimately, we don't know what model will win out or whether different lending models will co-exist for different circumstances.

Your argument is essentially that 21 million is not necessarily the upper limit because if in some distant future Bitcoins were horribly broken somehow, we might create more of them to fix the problem. Sure, that's true. But it's also totally irrelevant. So long as Bitcoins aren't fundamentally broken in some way, 21 million is the limit. And if Bitcoins are fundamentally broken in some distant future, they won't really be anything like what we now know as bitcoins and nobody plans with a time horizon that far anyway. (But it's a plus that should that ever happen, we can fix it.)

Not really. There's 100 billion XRP and the transaction fee is currently 100,000 transactions to the XRP. In some far, far future, should XRP become so scarce it limits the ability of people to perform transaction, the system could be modified by consensus to keep transaction prices in transactions per drip rather than drips per transaction. Alternatively, the divisibility could be increased or all balances multiplied by 100. It's about as likely as running out of Bitcoins.

I had the form for that on my wall for several years. Lots of people refused to believe that it was real.

We tried to find a way to transfer XRP to validators as an incentive to validate but couldn't come up with any scheme that didn't either require a central authority or that could easily be gamed by someone putting themselves on both sides. So we opted to destroy transaction fees.

Communism is when private property is abolished in favor of property ownership by the community as a whole.
http://en.wikipedia.org/wiki/Communism

Maybe I do and maybe I don't. If I don't, Ripple doesn't force me to. If I do, Ripple lets me. You don't ever have to issue IOUs to use Ripple to make and receive payments. You need never owe anybody anything unless you find it to your advantage to do so.

Again, you need never owe anybody anything unless you find it to your advantage to do so. Read the very post you're replying to. Nobody but a gateway ever owes anybody anything in the typical payment paths.

You have to choose between the certainty of a small loss and the very low probability of a moderate loss. Today, every business that accepts credit cards chooses the former. If you do the numbers, the second makes more sense.

As I said, we're not relying on community/social credit. We're promoting Ripple as a payment system based on gateways. The social credit system will always be there should people be ready to use it. I agree that it's difficult to use safely right now, hard to understand, and so on. Much better tools will be needed, and even then it may never catch on. I genuinely believe that one day it will be Ripple's killer app and will transform the way people think about money and finance. But we're not relying on it.

In any event, among people using the same currency, the path will almost always be either:

Payer -> Gateway -> Recipient

or

Payer -> Payer's Gateway -> Middle -> Recipient's Gateway -> Recipient

Where the "middle" will typically be someone who holds IOUs from the recipient's gateway and accepts IOUs from the payer's gateway. Both the payer and the recipient are only necessarily trusting one gateway. You won't have to do more than that to use the system to make and accept payments.

The number of hops has no effect on the number of people who owe you money. There's just no connection at all.