ILP and notaries

ILP and notaries

Postby tulo » Thu Oct 15, 2015 3:00 pm

Can someone from RL explain a bit more about notaries and safety of ILP?
Basically notaries are exactly the same as validators in RCL.
So is there the equivalent of UNL in ILP?
Who guarantees that escrow act based on the decision of Notaries?
tulo
 
Posts: 829
Joined: Mon Jan 20, 2014 2:38 pm

Re: ILP and notaries

Postby JoelKatz » Thu Oct 15, 2015 4:56 pm

tulo wrote:Can someone from RL explain a bit more about notaries and safety of ILP?
Basically notaries are exactly the same as validators in RCL.
So is there the equivalent of UNL in ILP?
Who guarantees that escrow act based on the decision of Notaries?

In universal mode, it depends on the nature of the ledger the asset that is escrowed trades on. If that ledger is a blockchain like Bitcoin, then the Bitcoin transaction is constructed such that the transfer of the Bitcoins is controlled by the action of the notaries. If that ledger is the Ripple consensus ledger, then the funds are held in a suspended payment constructed so that the release is controlled by the action of the notaries. If that ledger is a private ledger, then the release of the escrow is controlled by that ledger's ILP implementation. It's a cryptographic check that can be performed entirely locally.

In atomic mode, you do have something more like a UNL. The ledgers agreeing on the transaction have to agree on a set of notaries that they trust. It is each ledger's responsibility to act on the decision of the notaries. Should a ledger fail to do this, the other ledgers will have proof that it failed to do what it promised to do.

The key that makes this all work is that a ledger failing to do what it promised to do can only harm the party expecting to receive a payment on that ledger. That party necessarily already trusts that ledger. One advantage of atomic mode is that if a ledger does fail to do what it promised to do, the other parties have cryptographic proof of the precise failure and the precise party responsible. The one thing universal mode can't do is provide proof that something did not happen, allowing for some cases where ledgers can point the finger at each other.
User avatar
JoelKatz
Ripple
Ripple
 
Posts: 1859
Joined: Sun Dec 23, 2012 3:45 pm
Location: Oakland, CA

Re: ILP and notaries

Postby tulo » Thu Oct 15, 2015 5:46 pm

JoelKatz wrote:
tulo wrote:Can someone from RL explain a bit more about notaries and safety of ILP?
Basically notaries are exactly the same as validators in RCL.
So is there the equivalent of UNL in ILP?
Who guarantees that escrow act based on the decision of Notaries?

In universal mode, it depends on the nature of the ledger the asset that is escrowed trades on. If that ledger is a blockchain like Bitcoin, then the Bitcoin transaction is constructed such that the transfer of the Bitcoins is controlled by the action of the notaries. If that ledger is the Ripple consensus ledger, then the funds are held in a suspended payment constructed so that the release is controlled by the action of the notaries. If that ledger is a private ledger, then the release of the escrow is controlled by that ledger's ILP implementation. It's a cryptographic check that can be performed entirely locally.

In atomic mode, you do have something more like a UNL. The ledgers agreeing on the transaction have to agree on a set of notaries that they trust. It is each ledger's responsibility to act on the decision of the notaries. Should a ledger fail to do this, the other ledgers will have proof that it failed to do what it promised to do.

The key that makes this all work is that a ledger failing to do what it promised to do can only harm the party expecting to receive a payment on that ledger. That party necessarily already trusts that ledger. One advantage of atomic mode is that if a ledger does fail to do what it promised to do, the other parties have cryptographic proof of the precise failure and the precise party responsible. The one thing universal mode can't do is provide proof that something did not happen, allowing for some cases where ledgers can point the finger at each other.


Ok thanks, exactly the answer I was looking for :D
So at the end 100% of the risk is on the shoulders of connectors that should trust 2 ledgers.
And what about universal and atomic mode? Why and how one should use one or another? It seems more interesting (and robust) the atomic mode but I need to finish reading the details of the whitepaper.
tulo
 
Posts: 829
Joined: Mon Jan 20, 2014 2:38 pm

Re: ILP and notaries

Postby JoelKatz » Thu Oct 15, 2015 6:51 pm

tulo wrote:So at the end 100% of the risk is on the shoulders of connectors that should trust 2 ledgers.

Strictly speaking, each person who has agreed to receive a payment on a particular ledger or hold funds on a particular ledger is taking the risk that that ledger will fail to credit them or fail to allow them to spend.

And what about universal and atomic mode? Why and how one should use one or another? It seems more interesting (and robust) the atomic mode but I need to finish reading the details of the whitepaper.

There are advantages and disadvantages to each mode. The biggest disadvantage to atomic mode is that it is more complex and requires agreement among the participants on the notaries. The disadvantages to universal mode are that it doesn't work as well with complex payments that have multiple paths and that it cannot provide proof that a ledger failed to act.

The biggest advantage of atomic mode, IMO, is that it provides better protection against technical problems such as loss of a ledger's connectivity in the middle of a transaction. If connectors assume risk due to these kinds of failures, they'll have to price them in to their rates.
User avatar
JoelKatz
Ripple
Ripple
 
Posts: 1859
Joined: Sun Dec 23, 2012 3:45 pm
Location: Oakland, CA


Return to General Discussion

Who is online

Users browsing this forum: No registered users and 7 guests