RE: [payment arch] Negotiation of payment terms

Hi Adrian,

I’d say that we’d come much closer to our goals if we made credentials the central concept instead of the symmetrical agent setup which I haven’t seen any benefit for yet. But I think I found a way to adapt:

We will need to state that credentials MAY be used through either interface of an agent.

Any comments on this approach?

Cheers,
                Jörg

P.S: One more response below

From: Adrian Hope-Bailie [mailto:adrian@hopebailie.com]
Sent: Dienstag, 12. Mai 2015 22:06
To: Heuer, Jörg
Cc: Dave Longley; Web Payments IG
Subject: Re: [payment arch] Negotiation of payment terms

I don't think the proposed architecture prevents that, in fact I'd say that is necessary for many implementation scenarios but I don't think it is essential.
What is stopping a payment provider (like PayPal or Venmo or Stripe or ....) including a payment agent in their app and building direct account access into this?
They would have to authenticate for security reasons – why not be consistent and require this to be done with credentials?
At this level of the design I think we are too abstract to say that credentials MUST be in any place or that we will always need credentials for every use case.
Is it possible that as an entity I have multiple payment agents and one of these aggregates the others?

On 12 May 2015 at 19:43, <Joerg.Heuer@telekom.de<mailto:Joerg.Heuer@telekom.de>> wrote:
If the agent holds credentials good for all my different funding sources and additional services, I’d see this as an aggregation on a level the user can understand and control (which would also make clear where the user interaction is orchestrated). Tokenization in payments and authentication tokens like OAuth among others are well-suited to implement relevant solutions on this basis.


From: Adrian Hope-Bailie [mailto:adrian@hopebailie.com<mailto:adrian@hopebailie.com>]
Sent: Dienstag, 12. Mai 2015 19:10
To: Heuer, Jörg
Cc: Dave Longley; Web Payments IG
Subject: Re: [payment arch] Negotiation of payment terms

I don't know what you mean by "the aggregation takes place via credentials". Can you elaborate?

On 12 May 2015 at 18:49, <Joerg.Heuer@telekom.de<mailto:Joerg.Heuer@telekom.de>> wrote:
Hello!

Actually, I reviewed the architecture document as I promised to introduce credentials back into the picture. As it shows, the current setup doesn’t really support this because the agents have access to the funds directly. Looking at the invoice-based initiation below, I come to the conclusion that credentials should play a crucial role in the architecture instead of a functional relationship as we have it now.

My main problem in the document was that we can’t differentiate between the individual funding source and the aggregation role of the agent. If the aggregation takes place via credentials, we’d get a good grip on the topics in real world payment scenarios. Basically the invoice constitutes a credential of some sort and it is responded to by either another (set of) credential(s) or meta data in a more complex communication. I think, this is a lot of benefit which we could create if we dare to rethink ‘agents’ again before we get stalled.

Jörg

[…]

A quick note:

If we design this such that the offer is digitally-signed and includes all necessary terms agreeable to the payee, then no interaction is necessary with the store's payment agent to initiate and process the payment. Instead, the payer's browser forwards the offer to the payer's payment agent, which will then simply communicate with the appropriate payment processor to perform the payment/hold and receive a digitally-signed receipt. Then the payer's payment agent passes the receipt to the store's payment agent, which checks the signature and the transaction will be considered complete (or, in the case of a hold, funds can then be captured).

I prefer the latter as it keeps the command interface very simple. There would simply be a method like startTransaction(URL paymentAgentUrl) which can be used for almost any scenario.

If we do the above, we could do: startTransaction(Offer offer) or startTransaction(URL offer).

What do you think? Does the command interface (user to payment agent) need to understand the higher level flows?

On 11 May 2015 at 14:48, Adrian Hope-Bailie <adrian@hopebailie.com<mailto:adrian@hopebailie.com>> wrote:
@Joerg: In many scenarios a "primed wallet" is ideal because of the speed and the lower number of back and forth messages required in the negotiation.
Personally, if you are online and the processes are automated by "primed" configuration (i.e. you have set up limits and preferences so that your payment agent can execute most negotiations without your input) I see no issue with many messages being passed back and forth.
A long negotiation is only a problem in the "brick and mortar" scenario and how big of a problem it is I consider up for debate. If we consider that the likelihood of what we develop being rolled out in the physical world in probably 5 years it's also very likely that the instruments and in-store technology will have changed dramatically.


On 11 May 2015 at 14:02, Adrian Hope-Bailie <adrian@hopebailie.com<mailto:adrian@hopebailie.com>> wrote:
I think there are two topics at hand here, linking value networks and the offer negotiation. They are somewhat related but I think worth discussing in depth. I think the offer negotiation has particular relevance to the architecture document that we're working on.
I will respond with a different subject to split the thread.

On 11 May 2015 at 13:52, <Joerg.Heuer@telekom.de<mailto:Joerg.Heuer@telekom.de>> wrote:
Hello Adrian,

Your points are taken. I like the idea to call a third party in case of no match. We need to see how these things would be represented. I would think it’s going to be another ‘instrument’ in both participants portfolio in the end (which would be fine, because I believe it’s important to boil everything down to the same pattern for transparency and so…)

Privacy aspects are in the very core of our aspirations. I think we came out with a pretty cool schema to solve similar situations:

Dude has been to the rug shop a few times before, so he knows that this loyalty cards would be accepted. Also, his decision to  buy the rug today has been influenced by a coupon campaign put out by the shop last weekend and he downloaded a 5% coupon into his wallet.
After having made his pick, dude ‘primes’ his wallet with a payment card (he thinks is applicable) plus the loyalty card and the coupon.
When approaching the payment terminal all these items are made available for the payment terminal  - but only those, not any of the rest of the content in the wallet.

We got good feedback on the privacy – but especially for supermarket scenarios with all those coupons, people liked the idea to make everything ‘ready for action’ while standing in the line – as long as the more ‘interactive’ variant to communicate with the salesperson was still possible. Also, this selection process could be supported by the payers wallet application using pre-defined rules, contexts, down to location or beacons. Important was only that the selection made was presented to the user first, before the actual transaction was started.

The same mechanism came to my mind in case the user doesn’t know what the payment terminal will support, but wanting to be really quick. Just pick a VISA, a MasterCard and a direct debit mechanism and let the merchant decide. Wouldn’t be my ideal, but if you are in a hurry… I would leave it as an option, because it fits the overall interaction patterns. But I see potential for problems, though. Our experience says: if it fits the overall pattern, make it possible and see when and where players in the ecosystem make use of it. Loyalty and coupons will be okay with it. If Generation of a token costs money, your wallet will have ways to keep multiple ones being generated without need.

Cheers,
                Jörg

From: Adrian Hope-Bailie [mailto:adrian@hopebailie.com<mailto:adrian@hopebailie.com>]
Sent: Montag, 11. Mai 2015 13:25

To: Dan Schutzer
Cc: Heuer, Jörg; Joseph Potvin; public-webpayments-ig@w3.org<mailto:public-webpayments-ig@w3.org>
Subject: Re: Linking Value Networks

I don't think that the payer should send the schemes or instruments they have available to the merchant. This seems like a privacy problem to me.
I like the idea that the offer includes payment options (and these could have different currencies and prices for different schemes or instruments).
I think that if the payer sees the offer and wants to pay some other way or doesn't have a supported instrument they can call out to some proxy to be the intermediary. I.e. as the payer I make requests to a payment provider with whom i have an account by sending them the details of the offer. They go out and find the cheapest way to pay and then come back to me with a request for authorisation to use the payment instrument of mine that they have selected.

I can choose, if I am an advanced user or have some clever software to do it for me, to perform the work of the payment provider myself.
The problem with the current scenario is that the payer and payee are always negotiating a way to take the payment off the Web. Finding a common scheme means finding a way to settle the value exchange on a value exchange network they are both a part of.
This should work like the Web. We shouldn't both have to be in the same value exchange network if we can find others to proxy between us. That is truly value exchange at Web scale as opposed to some new messaging on top of the existing status quot.
I started writing a stupid/humerous "script" of a real world payment to try and eke out the messaging requirements. I started with the real world and then started adding the blue text to describe the technical requirements. What's interesting is that the conversation (real-world) changes a lot when you start to worry about things like privacy and trust.
Let me know what you think and feel free to copy and re-use for a different scenario:
https://docs.google.com/document/d/1-eV4rB-mh4l8zYnU5q_ZzHcShTvQCIa_aRx46AB0Mrs/edit



On 11 May 2015 at 13:09, Dan Schutzer <cyberdan250@gmail.com<mailto:cyberdan250@gmail.com>> wrote:
Merchant might say I don't accept American Express or I will give a discount for cash for example.

On Mon, May 11, 2015 at 7:04 AM, <Joerg.Heuer@telekom.de<mailto:Joerg.Heuer@telekom.de>> wrote:
Hi all,

Obviously, things can get pretty complex – not too surprisingly I might say.

In a more abstract manner I would then state that the ‘initiator’ of a payment transaction should also start to state the terms i.e. the supported instruments.

A) In the simplest case this means that the merchant’s  payment system (on behalf of the payee) declares what the total asked is, give a reference to the transaction itself (description, ID, ….) and the supported schemes. Dan, what would that mean with respect to the demand that merchants might want to decide? Could their statement be mandatory (‘I only take VISA and that’s it’) and, thus, fulfill the requirement or would the following be what you had in mind…

B) the payer states total, instrument(s) and a reference to the actual transaction and if the payee is okay with it, just confirms and does whatever the ‘reference’ implies (e.g. opening a turnstile or starting a song on the jukebox)

Both cases could provide for another round of ‘negotiation’; e.g. because the instruments or currencies couldn’t be matched or whatever. They could use the ‘reference’ (which we would likely ask to be ‘unique’ in some way) to continue communication, which might or might not imply other data or even protocols.
However, one thing should be taken into account: With the passing of a credential, the receiving side might already have received the money – or at least gotten the right to withdraw the given amount from the payer’s funding source. In the case of alternatives presented this seems like a no-go for real value tokens (‘I give you three different tokens for my three applicable cards and hope you only use one of them’). Also on the payer’s side this might already create costs (creation of tokens might cost something, amounts might be reserved or already be deducted). Thus, we might be forced to run protocols with either one full-value token right away, or a set of meta data about the instruments to be negotiated.

My goal would be to make the basic communication as simple as we can – and compatible to the current and the real world –  but leave options open for additions. As I see it, we were able to complete most transactions in one go and even with simple credit card-like data exchanges for a start.

Cheers,
                Jörg


From: Adrian Hope-Bailie [mailto:adrian@hopebailie.com<mailto:adrian@hopebailie.com>]
Sent: Freitag, 8. Mai 2015 22:25
To: Dan Schutzer
Cc: Heuer, Jörg; Joseph Potvin; public-webpayments-ig@w3.org<mailto:public-webpayments-ig@w3.org>

Subject: Re: Linking Value Networks

I agree with everything you have there Jeorg.
I think the offer of sale should list payment options up front (and different payment terms for each). This allows the payer to immediately respond with their choice of instrument. The payee's response could be some type of digitally signed quote confirming the terms which is the signal to the payer to complete the payment and send proof.
This flow allows the payer to seek alternative means of making the payment if they don't have an instrument that matches any of the available mechanisms.

On 8 May 2015 at 20:06, Dan Schutzer <cyberdan250@gmail.com<mailto:cyberdan250@gmail.com>> wrote:
There are situations where the payee or merchant gets to decide what forms of payment they are willing to accept. Would be good to allow for that option

On Fri, May 8, 2015 at 8:34 AM, <Joerg.Heuer@telekom.de<mailto:Joerg.Heuer@telekom.de>> wrote:
Hello!

It’s good we are gathering confidence for automated solutions being implementable. In the proximity world we already have to cope with such situations and should probably try to do something similar in the digital world in a first attempt:

•         First of all it’s a decision by the user. Just because I happen to have a certain payment instrument doesn’t mean it’s open for use in a given transaction at all. So we potentially have a subset of pre-configured (by the user) instruments for a given context.

•         The actual pick needs to be made on the payers side as we won’t try to send information about the content of a user’s ‘wallet’ to a merchant. (I hope we agree on this…)

•         If automation just supports a user who is present, conflict situations should be referred to him/ her to intervene and potentially bring other options to the table – e.g. choosing an instrument which costs him/her extra money..

•         If there is no user present, automation might require a second round but we need to make sure the matching really ends after that unless we want to create situations in which everybody waits for the other to move…

•         For future scenarios we could consider that the payee annotates the options offered in the first place to communicate rebates etc. to promote specific options. (Which could in turn feed into decision making in the payer’s side’s automation)

•         In how far automation with user present or user not present can get more complex, I wouldn’t want to predict, but I’m sure it’s important to always think such protocols in a ‘manned case’ as well as in the ‘unmanned case’ for a number of reasons (simplicity, trust in automation, backward-compatibility).

We have never implemented these more complex processes, but I know pretty well we could do so pretty easily along a real human interaction sequence. Adding automated negotiation e.g. could be an interesting option for the future but will be much harder to make people trust in. So let’s start simple, I’d say.

Cheers,
                Jörg

From: Joseph Potvin [mailto:jpotvin@opman.ca<mailto:jpotvin@opman.ca>]
Sent: Freitag, 8. Mai 2015 04:35
To: Dan Schutzer
Cc: Adrian Hope-Bailie; Web Payments IG
Subject: Re: Linking Value Networks

RE: "requires more than just a technical solution, it requires some business innovation"

The answer is:

OASIS UBL v2.1 Universal Business Language
https://www.oasis-open.org/committees/ubl/

http://ubl.xml.org/wiki/ubl-resources

...which is currently advancing as ISO 19845
http://www.iso.org/iso/catalogue_detail.htm?csnumber=66370

See also:
http://docs.oasis-open.org/ubl/UBL-Governance/v1.0/cn01/UBL-Governance-v1.0-cn01.html

...and for a couple of examples regarding its significance:
http://eur-lex.europa.eu/legal-content/EN/TXT/PDF/?uri=CELEX:32014D0771&from=EN

http://eeiplatform.com/13559/towards-single-standard-e-invoicing-eu-public-procurement-6-years-wow/


Joseph Potvin
Operations Manager | Gestionnaire des opérations
The Opman Company | La compagnie Opman
jpotvin@opman.ca<mailto:jpotvin@opman.ca>
Mobile: 819-593-5983<tel:819-593-5983>


On Thu, May 7, 2015 at 9:45 AM, Dan Schutzer <cyberdan250@gmail.com<mailto:cyberdan250@gmail.com>> wrote:
You make a good point - but to address this concern it requires more than just a technical solution, it requires some business innovation, but it can be addressed much in the same way that Square and PayPal can helped in areas where the payee is too small and not credit worthy enough to directly accept credit card payments.

On Thu, May 7, 2015 at 9:33 AM, Adrian Hope-Bailie <adrian@hopebailie.com<mailto:adrian@hopebailie.com>> wrote:
In working on the manifesto and the architecture document it occurred to me that we (or maybe it's just me) may be missing an essential feature in the payment agent model.
If our payment agents are expected to talk to one another to negotiate the terms of a payment, including the choice of payment scheme, then what do we do when there is no common scheme between the participants?
Does the payment agent give up and say: "Sorry Alice, you can't pay Bob he only accepts Visa, Bitcoin and ACH and you can only pay via MasterCard and XRP, transaction aborted"?
If so then it seems we aren't solving anything. Our vision for inter-connected value networks falls flat if our payment agents can only facilitate a payment within existing closed networks.
Would I be correct in saying we need to consider that in many scenarios there will be one or more intermediaries that "bridge" the two networks by being plugged into both? How do we fit these brokers/intermediaries into our architecture?
I think they are also payment agents of some sort but who do they interface with? The sender, receiver, both? And, how does the payment flow between Alice and Bob play out when this intermediary is required? At what point do their agents say, "Oh dear, we don't have a common payment scheme we can use, let's call Fred to act as a broker between your MasterCard and my Visa accounts".
I'd like to discuss this on the call today as I think we need to figure it out and put it in the document.
Adrian




--









--

Dave Longley

CTO

Digital Bazaar, Inc.

http://digitalbazaar.com