Highlights from today's ILP presentations?

[celticwarrior72]

I missed the first few presentations form today's ILP talk.  Anyone want to fill me in on the highlights?

[Guest]

I've got some notes I'll post tonight. Hopefully the recording will be available soon, too.

[Guest]

I'm not going to be able to get around to it tonight but definitely will tomorrow morning.

[David2142]

Just now, celticwarrior72 said:

I missed the first few presentations form today's ILP talk.  Anyone want to fill me in on the highlights?

you can start of by -- if you didnt see the slides yet...

 

http://www.slideshare.net/Interledger/ilp-webinar-102015-55230856

[David2142]

* sry thats old 

[David2142]

im still searchin the web looking for a video

 

[Guest]

On 2/25/2016 at 11:32 AM, celticwarrior72 said:

I missed the first few presentations form today's ILP talk.  Anyone want to fill me in on the highlights?

First off, it was nice to meet @Haydentiff, catch up with @MundoXRP and meet Meg, Ripple's social media person and one of their writers. Meg, if you're reading this, please stop by XRPChat once in a while - we'd love to have you join in on some of the conversations here.

This presentation was recorded, so presumably it'll be available for viewing sometime soon. I took notes on whatever struck me as important, so here they are, along with some impressions of what was shown and discussed. (Take this all with a grain of salt. If you notice something that's wrong, please let me know and I'll change it. Also, I didn't take down who said what for the first part of the talk and I'm not going to try to guess.)

The first presentation was from Adrian, Stefan and Evan (of Ripple @justmoon @Evan Schwartz), who gave overviews and demos of ILP using Five Bells and answered a round of questions. There were some questions and comments around pathfinding in the Five Bells demo. In the demo, the pathfinding is "naive," in that it takes the lowest cost route. This works well for a small-ish graph (tens?) of ledgers/connectors, but there was talk of other pathfinding algorithms as the network of ILP-enabled ledgers and connectors grows, avoiding known paths with low success rates, and pathfinding services. More talk about pathfinding later on.

About connectors:

- One of the presenters commented that, as payments work now, the sender and receiver take the risk, which they have no real knowledge about, but with ILP, one of the connectors along the path takes that risk and, given their position and knowledge of the system, can price in that risk.

- While a connector is modeled as a separate entity (third party between ledgers) in ILP, it will probably be the case that some ledgers will run their own connectors, and also there exists a potential opportunity for running metaconnectors that use white-label services to serve as hired connectors.

- What about connector metainfo? Will there be a connector "DNS?" Maybe connectors just register with their ledger and state to which other ledgers they connect. This way, ledgers know where liquidity can flow.

About AML/KYC: The presenters commented that there must be tools available so that ledgers can comply with their local laws, which requires some layers above ILP. Ripple is also working on schemes for prepayment info.

About atomic mode ILP: Everything should be considered universal mode. Certain portions of the ILP graph that trust each other, often process transactions together, and choose to use notaries in atomic mode can be modeled as one system. Regarding net payments, everything is considered settled by default, but two ledgers may choose to operate a netting connector. But because of settled-by-default, this is the ledgers'/connector's risk and transactions between the sender and receiver are settled.

About privacy in ILP: Receipt privacy is important to the recipient, as you don't want a signed receipt public until funds are actually received.

Stefan gave a demo of ILP using two UIs for different Five Bells ledgers with a connector (red/blue/connie).ilpdemo.org. This showed your typical web UI and how an ILP-routed payment might work for a user. In short, very easy and seamless. You can play with these ledgers and see JSON-formatted sender and receiver info for your payment. There is still a need to define higher levels to relay info about to whom and for what the payment was made. There was also a comment that was made about specing out what browsers need for ILP payment, and it sounded like some of that has been going on at the W3C meeting this week.

Evan then gave a "content micropayments" demo. For me, this was the most exciting demo of the day. Though micropayments may never add up to the huge amounts we see for international, b2b and consumer payments, I think it's very exciting and will change how the web works and how creators are paid for their media, code, and other intellectual properties. Using an HTTP/ILP Chrome extension and some ledgers, Evan showed how an ILP-enabled micropayment removes a paywall. He also showed a demo of how shares in a piece of media content can be split between the creator, the producer, the production company and other parties, and when that piece of content is viewed, how all of those parties are paid via ILP. There was a bit of discussion about how to control micropayments (daily limits at specific sites and such). It seems to me that there will be lots of opportunities to build protocol and services layers on top of ILP for micropayments and that content micropayments will be very exciting for content creators, web businesses and consumers in the near future.

There was also a comment comparing potential layers on top of ILP to the Payment Services Protocol module that Ripple has built on top of RCL for bank-bank compliance efforts. Finally, as I think has been said before, Codius was a major motivating use case for ILP, as there were no good options for (micro)payments to service providers that run smart contracts.

OK, back to work. I'll post more about the other presentations later today or tomorrow.

Edited February 28, 2016 by Guest
typos and a few short additions

[thinlyspread]

Fantastic summary, thanks @tomxcs. I am particularly interested in the pathfinding component. I think this area is going to be huge and have been saying so for years now and mostly people ignored me or thought I was a babbling lunatic. We're going to see huge specialisation, like we did with content search engines e.g. Google, into pathfinding across/between ledgers, taking into account many factors such as regulatory compliance, international laws, preferred trusted markets/companies, etc.

[Guest]

Part 2...

After the Ripple folks gave their ILP overviews and demos, three other people gave talks:

Zaki Manian, co-founder of SKUChain

SKUChain is focusing on blockchain-based solutions for trade and supply chain financing and Zaki talked about their "Brackets" computable contracts. I believe that SKUChain wants to "close the loop" and have payments flow in response to manufacturing and shipping info. It seems something like ILP is needed so that payments can be initiated, signed and flow from their Brackets contracts in response to that supply chain data. Very interesting - a complex and concrete application of ILP fitting into global trade.

Dimitri De Jonghe, Ascribe / BigChainDB

BigChainDB is a scalable, high-throughput database with blockchain functionality. Dimitri described and demoed an ILP-enabled version of BigChainDB, which he said did not take very much time at all (days I think?). I was very impressed with this - as long as your ledger has the right pieces for cryptographic escrow, is a basic implementation of ILP really that easy? If so, that's awesome and very exciting.

Jehan Tremback, Universal Payment Channels

Jehan would like some functionality of his UPC in ILP. UPC essentially works like Bitcoin lightning network: For networks that have a very high volume of transactions such that they would place too much of a load on a settlement provider's ledger, some kind of secure p2p accounting can be done, with net settlement done at a lower frequency. Basically, parties keep funds in escrow with a third party (bank, utility provider, etc.) and transact with each other directly - a fraction of a cent going one way, a few cents the other way, etc. - and then net settle with their third party escrow provider at a later time. Jehan's example is a mesh network of routers, but you could imagine this might also work for distributed PV electricity generation within a neighborhood or similar distributed/mesh networks. UPC - from what I understand, kind of a generalized lightning network protocol - seems like a great compliment to an interledger protocol, but I'm not sure what putting some UPC functionality in ILP means.

Stefan then talked about ILP's cryptographic conditions, focusing on what is in the receipt. One of the interesting things from this section was the construction of Boolean circuits for signatures. Right now, it's fairly easy to do either single signatures or m-of-n signatures. But for a lot of applications, we need to do something like x AND (y OR z) signatures and we can construct these signature Boolean circuits using stacked m-of-n signatures.

There was also discussion about how ILP decides on which cryptosystems to use and how to go about standardization. This standardization question was just about the only part of the workshop that seemed a bit contentious, with the obvious want/need to implement ILP pulling against warnings about potential problems with standardization committees.

The last talk (Evan, I think?) was a quick run-through of the Five Bells payments and ledger API for those wanting to experiment more with ILP. I had to get to work after this and left, but several stayed longer to break up into groups to talk more about pathfinding and receipts. Overall, it was a very informative and interesting workshop. Really glad I was able to attend in person and thanks again to Ripple for hosting.

Edited February 28, 2016 by Guest
typos and a few short additions

[Guest Haydentiff]

Great summary, @tomxcs! It was nice meeting you and @MundoXRPas well.

The most interesting part of the workshop for me was that Codius seemed relevant again. I was disappointed when it got shelved, but it sounds like that happened because they realized they needed ILP.  In order to move assets around, Codius contracts need to be able to talk to ledgers. If ledgers support cryptographic escrow, a form of escrow could provide a good interface.

The most exciting part of the day was that I got to meet David Schwartz.

 

[Morty]

59 minutes ago, Haydentiff said:

The most exciting part of the day was that I got to meet David Schwartz.

Not sure if small woman or big man. 

[Guest Haydentiff]

1 minute ago, Morty said:

Not sure if small woman or big man. 

If I stand up really straight, I'm almost 5 '4.  I think that's average.

(The camera angle was kind of weird.)

[Morty]

17 minutes ago, Haydentiff said:

If I stand up really straight, I'm almost 5 '4.  I think that's average.

(The camera angle was kind of weird.)

He's standing a bit closer to the camera too. 

[tulo]

On 2/28/2016 at 10:22 AM, Morty said:

Not sure if small woman or big man. 

Not sure if lucky girl or lucky man

[tulo]

On 2/27/2016 at 11:19 AM, rippledigital said:

Fantastic summary, thanks @tomxcs. I am particularly interested in the pathfinding component. I think this area is going to be huge and have been saying so for years now and mostly people ignored me or thought I was a babbling lunatic. We're going to see huge specialisation, like we did with content search engines e.g. Google, into pathfinding across/between ledgers, taking into account many factors such as regulatory compliance, international laws, preferred trusted markets/companies, etc.

Cannot agree more.

I always tough that if RCL/ILP will take place, pathfinding will be a big business (and hard work). Lots of research is needed to move from the state-of-the-art graph pathfinding to some fast and reliable algorithm for huge dynamic graphs.