Very interesting that Ripple is migrating to a platform focused entirely on an old idea, atomic cross chain trading (ACCT)
It looks like they have written a 25 page white paper expounding upon the concept in very general terms.
It is clear why banks are interested in this: they each have their own internal ledgers and they want to keep it that way. This means the only solution that the banks will accept is one built around atomic trading.
Atomic trading depends upon multiple ACTIVE participants in every trade and is limited in speed by a need for various timeouts and exchanges of data.
At the end of the day picture two banks using this system. A customer of Bank A wishes to send funds to a customer of Bank B and we assume Bank A and B are not cooperating with each other. Their protocol assumes that Connector C has an account with Bank A and Bank B and that both Bank A and Bank B support time locked escrow protocol (ACCT). Connector C must run a server to facilitate the transfer of money from A to B and customers must know how to discover the potential paths between various banks.
Connector C is effectively a trust-free version of Shape Shift, Meta Exchange, and Block Trades. When user Sam of Bank A transfers $100 BankA.USD to user Alice of Bank B, Alice receives $100 BankB.USD. The connector is performing a shape-shift from BankA.USD to BankB.USD and as a result is exposed to any and all volatility between BankA.USD and BankB.USD and must also maintain sufficient liquidity. If there is a bank run from BankA to BankB the connector may run out of funds in its account on BankB while its balance in BankA grows.
As a result of this, connectors must charge fees and their fees will change based upon how their portfolios are balanced. Making a payment between two trustworthy solvent banks issuing notes for the same currency will have relatively low spreads. Providing a connector from Bitcoin or BitShares to a bank account would have much higher spreads. In fact the fees would probably be prohibitive when compared to Credit Cards.
Furthermore, even though there is no longer any need to trust the connector, the connector is theoretically exposed to the same legal risks and barriers to entry as ShapeShift, MetaExchange, and BlockTrades. Market forces will tend to push all trades through the same connectors to maximize volume and minimize spreads charged by the connectors.
The ultimate conclusion from all of this is that governments will regulate the connectors which will effectively keep some ledgers from ever gaining a connector.
Creating a version of ShapeShift or BlockTrades that is Trust Free does mean that there is less need to regulate the connectors from the perspective of potential theft, but regulations will still be required from the perspective of Anti-money-laundering.
Ultimately all they have accomplished is the elimination of the need to trust ShapeShift for a single minute. The amount of money lost in the market by companies like ShapeShift defaulting is pretty close to 0. Therefore the value of eliminating this particular risk is pretty close to 0.
The only way that this Interledger system would have any value at all is if the operators of Connectors were free from all regulations because they can perform their function in a trust-free manner. This would allow many people to set up all kinds of businesses like Shape Shift to connect everything to everything for very low fees. Eliminating the need for regulating these entities would have tremendous value.
Unfortunately I fear that Connectors will be heavily regulated. BankA or BankB could shutdown Connector C at any time by freezing their account. While no user funds would be lost, the market would be anything but free and the only connectors that would exist would be the same ones that could be trusted in the first place.