I was asked to create a new Thread on this subject rather than derail the other thread. So here goes nothing.
Ripple at its conception was the idea that in a web of trust debt (IOU's) can be created and destroyed among two or more parties that do not know each other.
The maximum amount that these two parties would trust each other was derived from the chain of proximate trust among a string of people between the two unknown parties. The lowest trusted amount along the chain would be assumed to be the trust between the two unknown parties.
The people in between the unknown parties are called "Liquidity Providers".
Flow of IOU
Party A -> Party B -> ... Party Z
If the lowest trust in that chain is 50 cents, then it is assumed that Party A could get an IOU from Party Z worth 50 cents and vice versa.
The name "Ripple" came from the idea that debt would be created and passed along this web in a ripple-like fashion.
This would be extremely useful because potentially the IOU's could be traded on the market without even needing to redeem them at a "Gateway". (A gateway is essentially just another party that is generally trusted and has a lot of assets. Usually a Bank.)
This would replace all of the tedious vetting, oversight, and redemption mechanisms for debt that exist today.
Almost immediately upon the acquisition of Ripple by OpenCoin, the web of trust was almost entirely destroyed because what became immediately apparent is that a trust relationship among two end parties cannot be derived from the known trust relationships among consecutive parties no matter how intuitive it might seem.
Consider your best friend. You might trust this friend with $10,000 no questions asked. How much would you trust your friends friend, no questions asked? If you're prudent, the answer would be "nothing".
However, in the ripple system, if your friend trusts his friend for $10,000 then this unknown person could borrow $10,000 from
you!
Now if this person borrows from you and defaults, who's at fault? Would it be reasonable to sue your best friend because he's the one that extended trust out to this person that screwed you over? Do they share fault? This is frustrating enough, but in Ripple the defaulter could be a friend of a friend of a friend of a friend etc, and you could suddenly be out $10.00 .
I hope I've made it clear at this point that even one liquidity provider between a lender and a borrower is broken. It cannot be rationally done, and it cannot be fixed.
In order for this to work appropriately an algorithm to calculate the derived trust relationship between two unknown parties given only the trusted principle between consecutive parties in a chain would need to be formulated.
Given that trust is subjective, no such algorithm can ever be formulated.Therefore, the liquidity provider mechanism in Ripple needs to be removed entirely.
OpenCoin has not done this. Instead, no Gateways act as liquidity providers, and users are asked not to trust anyone but gateways. If a user is foolish enough to extend trust to even their best friend, they could be out real money in the aforementioned fashion.
At this point the only way OpenCoin's Ripple implementation works is if no one acts as a liquidity provider. No web of trust can exist. Therefore Ripple, as initially envisioned, has failed.
Now, even if the first impossible algorithm was formulated. There is a second fault in the Ripple system as original envisioned. The principle of a debt is not the only factor that determines the value of a debt. Initially it might seem intuitive that a $10 debt between two individuals is functionally equivalent, so that two lenders could trade their IOU's among each-other freely.
However, the value of debt in a marketplace is derived from the chance of default, the length of time the debt is held, the value of the principle, and the subjective value of a product held today versus the promise of a product at a future date.
Therefore, the value of a unit of USD IOU from Bank A and the value of a USD IOU from Bank B are different. They can not be equivalently exchanged based only on the principle.
This sounds crazy, but a quick thought experiment will prove to you that it is true.
Consider that your best friend owes you $10,000, and a your friend is owed $10,000, could you exchange the IOU's and wipe out the debt? No, because you still have to redeem the IOU's with your friends friend, which may not be good for it. You value the debt between your best friend differently than you value the debt with some person you don't know.
In summary Ripple, as was initially envisioned, failed for two reasons;
1.) No algorithm can be formulated to predict the trust relationship between two end parties given only the trust relationship among consecutive parties in the chain.
2.) The value if debt is subjectively determined. Therefore, no accurate calculation can be performed between IOU's, even if they are denominated the same underlying asset.
So that's how the original Ripple idea failed, but I will go a step further and describe why today's Ripple will not work.
Today's Ripple still holds onto the ghost of liquidity providers, but only among users (not gateways), and therefore exposes users to the risk of holding onto irredeemable IOU's in the event of a Gateway default (Or even a user default if they in-advisably trust someone other than a Gateway).
Not only are Ripple users not compensated for this risk, but Gateway will almost surely charge a premium to issue IOU's.
Think about that for a second. You go to a bank, give them money, and then receive a promise to get that same amount back from them less 1 or so percent at a future date. It's a CD with a negative return.
Not only is this perverse, but Gateways have an incentive to give out as many of these IOU's as they possibly can and then avoid redemption through any kind of contrivance that they -or their government- can dream up.
The people at OpenCoin say that this isn't a problem because the IOU's would devalue and the Gateway would not be able to sell their IOU's. This is true, but nonetheless there is a constant Moral Hazard built into the system that doesn't normally exist. (Fractional reserves are all but guaranteed.)
For the current system to work,
1.) IOU's would have to redeem for an amount higher than the principle. (I can't think of any way to do that besides introducing a Minimum-Held time to IOU's)
Edit: If users can't be Liquidity Providers, the above might not be necessary for some assets, given the utility gained by being able to exchange the IOU's.
2.) The final nail needs to be put in the coffin of "Liquidity Provider". (It should be removed entirely.)
While those problems are not unsolvable, they are tremendous tasks that are burdened by the cost of XRP, and it ensures only limited liquidity on the system.
Instant redemptions on this Ripple system are a fantasy. Edit: That's not really claimed anyway.
Now. If those problems are solved, Ripple will be an open source, trust-based, free to enter system for the trading of CD's denominated in any arbitrary asset (currencies/stores of value). Cool.
The great thing about that is any currency can be exchanged if the user base is big enough, without the need of Gateways to hold onto any IOU's of other Gateways.
For instance, you could trade a XBT IOU with a ripple user for a USD IOU, and then redeem the USD IOU at the USD Gateway, which doesn't care at all that you originally had a XBT IOU.
Edit: Expanded the explanation of the benefits of a working "quasi-ripple".
Excellent objections can be found here;
http://trilema.com/2013/ripple-the-definitive-discussion/http://ripplescam.org/
