Any such risks would have been equally obvious, or non-obvious, to the issuer and to the investors. It's not like some decision on the part of the issuer exposes the investors to this risk in a way that was hidden.

In retrospect, I agree with you. Anyone listing an asset on a stock exchange like GLBSE should have included a section in the offer explaining the risks associated with using an anonymous exchange. And they should also have included a section explaining what would happen if the exchange failed or delisted the asset.

It's hard to blame any specific person for not predicting this because, as far as I know, nobody predicted it. In retrospect, though, it's obvious that it's a significant risk that should have been addressed ahead of time.

It looks like the crypto-currency community is going to have to learn every single painful lesson that the regular financial world has learned. Probably more than once.

Nobody knows how to do this. If you do, please explain it to me. For example, what happens if someone claims they held the asset but the asset issuer claims they weren't on the list? This could be resolved if GLBSE was actively participating or had made the lists cryptographically verifiable. But so far as I know, there is zero participation from Nefario/GLBSE.

These are mostly very minor or aren't a case of an asset issuer having to trust an exchange.

I don't see how this is trust extended to the exchange. You mean if the exchange does something wrong, the issuer might get sued? I would consider that a pretty unrealistic fear. We'll find out which of us is right.

True. As it happens, giga isn't anonymous, but that applies to some issuers.

That doesn't really hurt the issuer.

I don't see how that's trust extended to the exchange.

Yes, I agree that asset owners have to trust the exchange. The question was whether asset owners had to trust the exchange.

Sure, if asset issuers also hold assets, then in their capacity as asset holders, they have to trust the exchange. But one can issue an asset without holding any assets on that exchange. As asset issuers, in that role, very little trust is extended to the exchange operator. (This is not the way it would normally be, it's unique to the Bitcoin universe.)

I hope this is either some kind of language barrier issue or some bizarre case where they're actually having a technical problem removing the hold. Otherwise, there's no excuse for telling a customer they have to wait for days while you chase down a technical person to fix a non-technical problem. Whatever problem he is having, why isn't there a person on call during regular business hours who can fix it and who the support staff can easily contact?

He didn't do that, his investors did. If GLBSE had run off with their money, would you have argued that's somehow his fault?

That's kind of a circular argument. If this is his mess to clean up, then he extended trust. And if he extended trust, it's his mess to clean up.

In retrospect, we all should have thought of it, but I don't think anyone considered a sudden, complete GLBSE shutdown likely. So I don't think you can argue trust was extended on that basis. If someone gives you free concert tickets and while you're gone they have a friend rob your home, you trusted them to know when you would be away from home in a sense. But you didn't really decide to trust them if you never seriously considered the possibility that the concert tickets were a trick to rob your home.

Any contract so far as I know. I don't know of any asset listed on GLBSE that obligated the issuer to cover the costs and risks associated with preserving asset ownership across a sudden GLBSE failure. As for the risks, there are many. The specifics depend on the contract. The most common risk is the risk of paying false claims. And the costs are obvious -- figuring out which claims are legitimate is not free. No asset issuer, to my knowledge, ever agreed to bear those costs or arranged a way to pass them on to the asset holders.

I don't quite follow you. In what sense did you extend trust to GLBSE? In what way could GLBSE have done something that would have hurt you?

Now, the people who bought assets from you had to trust that GLBSE would actually deliver their money to you. Otherwise, they wouldn't get any dividends. And they had to trust that GLBSE would actually distribute the dividends to the asset holders. And, of course, they had to trust that GLBSE would remain in operation and not somehow lose or destroy their ownership interest. They had to trust that GLBSE would remain operational to ensure there continues to be a market that preserves the value of their assets. So they extend all kinds of trust.

But what was your exposure exactly? Maybe I'm missing something. I guess there's risk to your reputation and risk of lost profits.

He never necessarily extended any trust to Nefario at all. Asset issuers don't have to extend any trust at all to the exchange. You don't have to trust someone to take someone else's money -- the person giving them the money has to trust them to do the right thing with it.

Surely you understand that there's a difference between claiming you have no other option and actually having no other option. We know of no reason Nefario couldn't have done an orderly shutdown of GLBSE. And his prior acts with Goat show that he was acting rashly and in complete disregard for his obligations to others. Giga actually has no other option -- he never extended any trust to Nefaro/GLBSE nor did he ever agree to extend any. The contract doesn't obligate him to accept these costs and risks.

There are a variety of possible reasons. One would be if that's not what the market wants. There's no point in offering a product nobody's buying. Another is that it's risky for the lender. If they make unwise loans, they can lose everything. And even if they do everything right, a run can bankrupt them.

On the other hand, if the market demands a fractional reserve (say because people insist on being paid interest and aren't willing to pay storage fees), then a bank or asset holder will have not choice but to do fractional reserve lending.

I agree that a lender is not unhealthy just because they have a fractional reserve. A lender is unhealthy if their assets (including loans) don't significantly exceed their obligations (including deposits). A lender is also unhealthy if the value of its assets are overstated, for example, if many of the loans are shaky.

That remains to be seen. I think it's very dangerous to assume any particular model will win out. You're trying to do the equivalent of predicting how people will use the Internet 20 years from now, and Ripple makes other models possible.

For one thing, with interest rates as low as they are now, it's not clear a fractional reserve actually makes sense. Of course, interest rates probably won't be this low forever.

The original idea was posted here: https://bitcointalk.org/index.php?topic=10193.msg146250#msg146250
It has changed quite a bit since then though.

The important thing to keep in mind is that the number one priority of every honest system participant is that there be agreement on the database -- nothing is more important than that. So anyone trying to launch an attack that relies on some kind of ledger disagreement is going directly against the interests of every single honest participant. Worse, they either have to be willing to provably contradict themselves (in which case they would immediately lose any trust they had built) or be constrained to keep any commitments they made, which makes an attack awfully hard.

Under normal circumstances, no proof of work is used. Under unusual circumstances, proof of work may be used on a specific connection to prevent certain types of denial of service attacks.

Sorry I didn't notice this post.

There is no central authority or choke point. Anyone who wants to can run a server. The ledger is public.

No, no central server. No central authorization.

Anyone can run precisely the same software if they want. It will be open source.

The "you" is whoever is running a fractional reserve system.

Even if that's true, provided the bank has assets significantly greater than its obligations, it can still pay everyone back eventually with significant interest. People may be inconvenienced by not having the liquidity they expected, but they can't lose their savings. If a bank doesn't have assets that are significantly greater than its obligations, then it's undercapitalized, and the run just exposes that problem.

My point is simply that a run cannot cause depositors to lose their deposits if the bank was healthy prior to the run. Bank runs are a solved problem for everyone but banks -- a run can bankrupt the bank.

If there's a run on the bank, you may be unable to meet immediate demands for called in debt. But you can simply issue notes worth more than the called in amount. This will bankrupt *you*, but that's the risk you take when you implement a fractional reserve. Those who loaned you money won't suffer.

For example, say I have $10,000 in deposits, all of which I've loaned out. I have a $3,000 reserve. Now, I get $8,000 called in, which I can't pay. What I do is I issue $8,000 in interest-bearing notes at double the prevailing interest rates and use them to pay back my customers. My customers will be happy because they can sell these notes for more than the $8,000 they're asking for. As the loans get paid off, and using my reserve, I can pay off the notes. Nobody gets hurt but me.

Runs aren't really a problem. Of course, such systems can still collapse due to bad loans or insufficient capitalization. Generally, a bank has an equity crisis that ends with a run and people think it's the run that killed the bank. But that's just not true. A run can destroy a healthy bank but it can't hurt those who loaned the bank money.

This probably has nothing to do with Ripple though.

That was a link to an old Saturday Night Live fake commercial for "Shimmer", a product that's both a floor polish and a dessert topping.

I don't think I can answer that question. On another note, what do you think of our logo?

The web content on ripplepay.com does not relate to the new Ripple system. It's still describing Ryan's system.

In that case, the morality of the action would rest with whoever or whatever created the autonomous system. Since the autonomous system is autonomous, it is not a moral agent. But I can program a robot police officer to shoot the innocent or shoot in defense. The former would be moral on my part and the latter immoral. It matters not that the automaton then executes the operations amorally.

Any moral agent who followed Libertarian principles would be acting morally or immorally because they would be choosing to follow those principles. Any automaton who followed them would be acting amorally, however the creator of that automaton could be acting morally or immorally. There is no requirement that moral *principles* be incapable of reduction to algorithms. (And if there were, it would be almost impossible to come up with any moral principles at all. "Thou shalt not kill" couldn't be a moral principle.)

It's terrible if a human *blindly* follows someone else's code. Humans are moral agents and are responsible for the choices they make. They aren't automatons. But there's nothing inherently wrong with following a moral code if one has determined, to the best of one's ability, that that code is in fact moral and they are willing to change things if evidence points otherwise.

This is a ridiculous argument. You might as well argue that if I punch you in the face, that's not immoral because my fist is not capable of acting morally. Computers are just like our fists -- they do what we tell them to do. They are not moral agents, but when we command them morally, then *we* are acting morally and when we command them immorally, then *we* are acting immorally. You are erroneously looking at the actions of the agent rather than the actions of the master that commanded the agent.

That's the bad way to extend trust to strangers. There is a good way to extend trust to strangers though:

1) You see that these strangers have engaged in numerous real transactions such that the cost to create a fake set of transactions is more than $1.

2) You extend them about $1 of trust.

3) If you lose, you lose $1. No big deal.

4) As soon as this trust has made you more than $1, you increase your trust to $2.

5) Repeat, ratcheting up the trust such that at any time, if they default, your maximum loss is very, very small, way less than their cost to fake such an arrangement. Very quickly you'll reach the point where even the maximum loss they can impose on you is just a small dent in the benefits you've gotten.

However, again, if you don't find this convincing or this is not for you, don't extend trust to strangers. Unless the stranger is well-connected and liquid, there's no benefit anyway. And if they are, it's unlikely they'll abandon that just to rip off a few people for a few dollars. Personally, I believe that people will wind up extending significant trust to strangers and benefiting from it sufficiently that the occasional small loss will be well worth it. But not everyone agrees with me, and I recognize I may be wrong.

While there is where the name "Ripple" came from, this is just one of the things you can do with Ripple. I personally think that, at least long-term, it's going to be what makes Ripple great, but you don't actually have to extend credit limits to your friends to use Ripple. And short-term, I don't think that's how people will primarily use Ripple. I wish I could give more details, but I can quote our public blurb:


Notice it doesn't mention anything about extending credit to friends, or credit at all for that matter.