The W3C Credentials Community Group

Meeting Transcriptions and Audio Recordings (2014-today)

Go Back


Verifiable Traceability Task Force

Transcript for 2022-08-02

Our Robot Overlords are scribing.
Khalid_Gobin: Yeah for sure my name is Khalid I am a software developer internet May vinet and it's nice to be working with all of you.
Mike Prorock: All all detours briefly Chris has actually gone this week so kudos to him for actually leaving like the front of his computer the but we Orient have been talking for a little while about doing like a walkthrough of how do what are best practices and really start to finish like json-ld for vocabularies because there.
Mike Prorock: Kind of some specific things there.
Mike Prorock: I think unless or he's going to be out of office to we should try to do that during the trace called next week and if so because that way I think most people will be out of vacation time and here and stuff like that I just wanted to make sure like we had as many people as possible around for that any quick feedback anyone from either the transmute may have Annette said anyone could be out next week or anything like that that we should think about before we schedule anything like that.
Mike Prorock: And I'm.
Mike Prorock: Silence is like I can send an email to the list that we're going to do that so.
<tallted> fwiw I expect to be here next week
Mike Prorock: Great all right I'll send an email to the list to that effect so cool that was the only detour I think I aside from Ori since you and I spent last week having our brains full of things anything you want to note for my ETF that's relevant to this group just while we're here as a group or anything.
Orie Steele: I think that there is continued interest in the house E & Cozy representations for verifiable credentials so if you're not currently capable of issuing credentials and those formats you can consider whether or not you want to issue credentials in those formats we have some tests that cover some of that and I see.
Orie Steele: Posie & Cozy formatted verifiable credentials will change so I also understand if you don't want to implement them just yet because you know they're going to change that's really all I have to say about that.
Mike Prorock: Yeah yeah I will note that that seemed to be a running theme that I was encountering in a lot of different context but especially on things related to supply chain and ITF so so you aware if you're working on stats issuing and verifying credentials You Know bethink Thee for sure thinking about the future right and especially binary type stuff is we look at some volumes of data so and I think.
Mike Prorock: That's probably the last.
Mike Prorock: Detour and less.
Mike Prorock: What else has anything.
Russell_Hofvendahl_(mesur.io): Hey I think that was me.
Russell_Hofvendahl_(mesur.io): No problem so this the biggest you last time was that I wasn't clear enough on this to provide a good explanation of why it was needed thank you Mike for elaborating on that the only other than that the only change is as requested how the value with multiple possible types handled.
Mike Prorock: And I queued up because I can talk to this a little bit because it's yeah it's a bit of a weird one and Ted and I think the initial read that I think both said and everyone else had and it was my initial read when presented with this by customer was that why right and what it actually boils down to is that we have a couple of customers and we're able to test you know throwing some.
Mike Prorock: Any kind of data their way just to see how they.
Mike Prorock: To adjust it they wanted a very sad way to say that here is this programmatic way to make a labeled you know the thing with some external reference number that they handle separately in their system right it's totally blind to anything going on with the Trade Center on stuff that then correlates that label in their system with some external value and and the.
Mike Prorock: Kind of.
Mike Prorock: They like the easy example to think of is where you have a customer that needs to reference for instance an external database identifier in some third party system that there is a pre-arranged you know rule set between the two parties the up I can send you this data and you know what to do with it in your system knows how to programmatically interpret something you got to credential of this type and you then have this business logic that steps in and says.
Mike Prorock: I've got a rule.
Mike Prorock: You know have in there and then based on that I'm going to go step down some other business logic somewhere and so rather than just saying oh yeah just make a claim and call it whatever you want that made people very very nervous right just exploiting the actual default behavior of VCS they actually wanted some names attached to it and also one of the type assigned to the like a defined type for this kind of icky bear effectively.
Mike Prorock: You know key value pair so that they could intercepted.
Mike Prorock: Handle it appropriately that may or may not have triggered some questions from folks in the group not sure so.
Mike Prorock: It's a loaded term yeah exactly.
<orie> This PR is a disaster
Mike Prorock: Yeah well in in in a totally alternate path and I lost a language that they request because they effectively we're going after third party you know a very programmatic way of referencing third-party literal claims you know like here's an idea to go look up this additional data that will never make its way over here right you know the you know I think the terminology itself.
Russell_Hofvendahl_(mesur.io): Terminology was it would be easy just to rename this third-party claim so it's less ambiguous.
Mike Prorock: You know I would you know that or external value right or something like that right so you know or keep are credential write something like that key value pair and it's just a very verbose way of basically saying there's key value pair that we explicitly know is going to reference some external system that's the goal right so we can bike shed the terminology our we want but that's what they actually.
<orie> I feel like you don't undersand what JSON-LD is.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): I think this needs to be an issue before it becomes a PR I don't think that it's flushed out nearly enough to understanding either what these end users want or how the best way to handle that is first we really do have to be clear on what they need and how it is not addressed by what we've already got.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): And we can hammer on what's the best way to deliver that thing that we don't already have.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): Because it's been discussed a bit and apparently not fully understood also want to talk about how claims became credentials verifiable claims to The uninitiated Listener and also to an awful lot of initiated listeners suggest that the statement IE the claim.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): Itself is verifiable so that you can confirm the factual nature of what is being stated and that is not what's going on.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): Verifiable credentials what is verifiable is that the issuer said these things made these claims.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): There's nothing about their validity their accuracy anything like that it's only that Joe said that Frank is in the park that's it Frank may not be in the park.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): May not be a park at all Frank may not exist.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): But we know that Joe said that thing that's it.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): And that being the case.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): Not sure that this customer wish this end-user wish is.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): That's it for now.
Orie Steele: Yeah I mean I think we have some building blocks which were extending here and I think the challenge that I have with this pull request is it's Reinventing the building blocks that were using so from a data modeling perspective if we say you know json-ld and semantic data modeling that requirement we have a spec that defines sin.
Orie Steele: Tax for that and this.
Orie Steele: Using that existing syntax again to Define that existing syntax just it seems to miss the point of verifiable credentials or Json I'll be generally I think the if the customer's request is I want a way to have key value pairs that are unique or in some structured format good news is that json-ld handles that really really well and I don't know that you necessarily need.
Orie Steele: To do anything.
Orie Steele: Much more in order to accomplish that like the type you know the question is you're trying to define a key value pair are you trying to define a key value pair that's just a property or an rdf class what which one are you trying to Define I think that's that's where we should start because that's the language that Jason will be uses to Define key value pairs so I think.
Orie Steele: Should have a conversation on issues about it it's possible that maybe some basic education around json-ld would also help unblock the stuff a happy to do that yeah when I want to understand why it is that we're Reinventing json-ld in a vocabulary.
Mike Prorock: Yeah well and it's it's more as a way of an interface to people that cannot will not nor have no desire to work with json-ld want to treat this stuff just as Json and want a programmatic way to say hey here's this identifier I'm going to give you and here's a foreign key or some other value that you can use to go look that up in that thing that's it and it could literally just be a key value pair like if we.
Mike Prorock: Do your external reference or external okay said like pretty flexible on the naming here but it's going to be a market reality that we're going to see.
<tallted> didn't we have something like "correlation"?
Orie Steele: So I'm cute to respond to that we are contact we are.
Orie Steele: Working on this vocabulary this is a json-ld vocabulary if there's disagreement now around whether or not that's what we're working on together I think we should just pause and have a longer conversation around that because there is ways there are ways to make json-ld sore friendlier to folks but there has to be some sort of shared a green around the value of a vocabulary and in the context of the work item as it's been developed so far.
Orie Steele: Far what you just said Mike is super.
Orie Steele: I don't really know how to effectively support folks who want to use a json-ld vocabulary but don't want to know what json-ld is.
Ben_-_Transmute: Yeah you know for me my my kind of argument or question behind this is you know if the customer wants to have some key value lookup or some way to do that wouldn't you know if they just have some Json interface you know that they wanted program at they want to do you know could they do that outside of vocabulary you know is that just a script or program that they write that does this is this really something that would be committed.
Ben_-_Transmute: Baseball cap or could you say you know.
Ben_-_Transmute: The scripture hears it.
Ben_-_Transmute: Your face feel free to use it outside of this.
Mike Prorock: Maybe yes maybe no but I don't think so and for one very good reason right if you think about putting a traceable presentation over.
Mike Prorock: We expect that everyone will have all of their data for all systems related to all things related to a supply chain or regulatory bodies Etc and this vocab and described in this vocab cool then I think then it's then we don't have a problem and we can kill this PR right but I think in fact I know realistically there will be an onboard time and a step-by-step type process and I'll give you a really good example which is Organic certification.
Mike Prorock: In the agriculture sector.
Mike Prorock: There are folks that will want to say look I've done the bare minimum to issue a VC that literally like I might not even do it right like I'm doing this real bare minimum you know almost hard coating it to say this you know ID you know number identifying the organic search that you can come look up in our system has such name attached to it and hands that to it.
Mike Prorock: Third party who then.
Mike Prorock: In a traceable presentation right there trusted there's a there's an agreement of business logic agreement between these people and a contractual agreement between these orgs that's the kind of situation we're going to increasingly run into as we start interfacing this stuff with the rest of the world right and and there and that's the issue is how do we then make it very as simple as possible to just our Bertie you know all on you know when needed say yep I've got this idea it's such a.
Mike Prorock: Or such and such foreign key that exists in an external system but I know enough about verifiable credentials to say yeah I want to sign that piece of data right but that's it I might not even know anymore than that right and I understand that sure it's not ideal it's not right it's not perfect etcetera but I but I think it's also practical.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): It's so starting from the base point that what we're building in this work item in this call is a traceability vocabulary.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): Is there nothing in the existing vocab as we've built it that covers this is question one and question two is is this something that exists in the paper world that were theoretically Reinventing with this vocabulary or or not so much Reinventing as standardizing we're building an OE D in a sense of all the traceability forms that are being used and so we need the vocabulary to.
Mike Prorock: The paper yeah exactly yeah.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): Of the filament blankly and if this doesn't if this is me just as a fill in this blank label then we're already going outside our remit.
Mike Prorock: Yeah Ted there's actually a really common paper example you see which is a letter that says it you know issuer I so-and-so right as the issuer about this subject right this person say that they have an import permit for X with such-and-such number or a organic search for such number it's literally that so it's precisely the use case for verifiable credentials it's just the barest minimum true.
Mike Prorock: Aimed down version of that.
<orie> I suggest you review `@id` and `@type` :)
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): The network they should use is a barest version trimmed down verifiable credential or they should attach maybe some binary exhibit to this.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): But that's that's where we're going here is we're talking bare-bones minimum and they're not willing to educate themselves about the bare-bones minimum they're using if you give somebody a hammer and they're not going to take the time to realize they should not use it on their head that's not our problem.
Ben_-_Transmute: Yeah I could I could I think you know we might kind of want to move on to an after this so basically just a summary is the sounds like a like Evite this just seems like mapping a key value Jason pair to arraylist and verify the credentials for people who don't understand the json-ld I think that.
Ben_-_Transmute: that you know.
Ben_-_Transmute: The purpose of this project is that we're defining a vocabulary in which we have structured data that can be grouped and and haven't already of representation so this doesn't feel like a stop Gap is much as if the claims and you know in this context aren't going to be mapped to anything it sounds like you could almost use at vocab and just have you know terms of floating or something like that it seems it seems like there could be it seems like you know a go.
Ben_-_Transmute: cat could be abused.
<orie> `@vocab` + `@id` and `@type`.
Ben_-_Transmute: Or what do you think.
Orie Steele: Yeah I think it's possible to do I mean there are as possible to do this and have term definitions that are somewhat useful it's possible to do it with term definitions that are useless and it's possible to do it while saying this is arbitrary Json I'm signing over it and I have no idea what it is you can do all of those things.
Orie Steele: Which of.
Orie Steele: Things are we trying to do you know I suspect based on the pull request we're trying to sign over terms that are meant to be interpreted in a generic way and I think that's very much achievable but my concern is that it's pulling the vocabulary in a direction that maybe isn't fully thought through and then the question is like well how does this approach integrate with all of the work we've done up until now.
Orie Steele: Because I think.
Orie Steele: Could easily start like patching less and just undo all of the vocabulary and go back to like signing simple key value pairs in and out vocab context with arbitrary Json inside of it and it might I could see a sort of like going all the way back that direction I'm not in favor of doing that I think we're here to create semantic Clarity and that means some extra effort has to be applied and then you need a way of signaling you know.
Orie Steele: No if you can't meet those.
Orie Steele: Eat that semantic Clarity bar as an issuer and yet you still want to issue credentials you should have a way of signaling that and are we actually have that today it's what you get when you just sign arbitrary Json within a vocab and you don't Define your terms properly.
Mike Prorock: I defer I kicked myself out of queue to let person case he was oh yeah.
Mike Prorock: Yeah yeah I'm going to I'm going to make a fundamental Point here that I think is important which is we will never go map every piece of.
Mike Prorock: Jack related to all vocabularies because otherwise we're going to do exactly the same cluster that exists for all of the mess of vocabs that are already out there so one of the things that I think we need to be cognizant of is do we or do we not at a fundamental level want to provide some way to represent a relationship between in our case a traceable presentation with some third-party system in a consistent program at.
Mike Prorock: Akin to find way so this is why I said this is a terminology.
Mike Prorock: Right you know and or you know bike shedding whatever we can kill this and we can reopen up and identical thing or a very similar thing but basically do we want to provide in the classic example in supply chain as sap do we want to be able to reference and say this issuer who runs this closed up sap system makes the following Claim about a subject right you know and you so we're using issuer in.
Mike Prorock: Objects same way that they.
<tallted> have we defined this "traceable presentation"? or is that just a verifiable presentation using the traceability vocab?
<orie> but this isn't a traceable presentation!
<orie> its an endorsement use case
<orie> ....
Mike Prorock: I have some pick arbitrary x y z table and a foreign key for instance right and yes Ted we have defined traceable presentation it's in the vocab it's a way of attaching a correlation ID to credentials that are presented together at the same time and and so if yeah correct or this is not a traceable presentation it's an endorsement use case that that you're exactly right just following the chat.
Mike Prorock: So how do we want to represent that.
<ben_-_transmute> i think those gaps can be filled in with @vocab, i think our objective is to define richly defined vocabulary for specific document-types
Mike Prorock: That is a thing we need to provide an answer to assuming that we're not going to get immediate adoption or you know Etc of all things and have all database schemas Etc described but want to in a semantic way be able to represent that kind of foreign key relationship and I don't think yeah.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): This seems like this is an implementation guide this is a onboarding to the modern new way of doing things this is get away from your paper and get into the wire but I still don't have a solid use case that is not satisfiable given the things that we already have.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): I need.
TallTed_//_Ted_Thibodeau_(he/him)_(OpenLinkSw.com): I need to understand why how the things that we already have are not usable for this thing I think the correlation ID is right there and that correlation ID could be attached to a binary attachment it could be attached to arbitrary text can be attached to anything and that should deliver what is being asked for right now which is semantically empty.
Orie Steele: Mike can I take a stab at understanding your use case since we were clearly stuck on must this pull request so I'm a first-party issue of organic certificate for my farm so I am I'm a farmer and I'm signing and certificate it says this field here is organic.
Orie Steele: Start start me with the the an issuer first issuer making a claim about us.
Mike Prorock: The first issue are in this case would be a third-party regulatory body that comes in and perform something and assigns an ID of some kind right so in this case it will be in the use case we're talking about the actual organic certifier comes in and says Ah I've got this farm and their organically certified but I'm not.
Orie Steele: Okay so let me let me back up let's start with the subject what is the subject.
Mike Prorock: In that case that would be the farm.
Orie Steele: So the farm as an organization the entire thing okay.
Mike Prorock: Yes yeah we're some ID right reverence it did right.
Orie Steele: Right and in the first issue or making a claim about that organization is what what is the issue.
Mike Prorock: The first issue we're making the claim about that the issuer would be the organic certifying body.
Orie Steele: Okay so there's a legal entity that certifies Farms are the issuer they're claiming that a specific Farm is organic.
Mike Prorock: Yeah that's an example in organic is not a good one because and the reason I'm saying organic is not a good one as we can Define some basic credentials around this right but there are this is really like but it's that kind of a thing where yeah.
Orie Steele: Want to understand how far beyond the issuer makes claims about subject use case we have to go so if are we done and we if I have an issue.
Orie Steele: Okay so hold on.
Mike Prorock: Over because what they need to then say is not just issuer Organics it's not just organic search as the you know certifying body is the issue or saying that the subject has the existence of an organic thing right there at what they're actually doing is they're specifying that here's an IED for a type that I need you to go look up in my third party system organic sir tidy whatever and the chorus.
Mike Prorock: Clapping item.
Mike Prorock: +1 Nis
Ben_-_Transmute: For me okay I will just quickly give a short introduction to this poor past this is making a few lines change in regenerate that JS I realized that after the folder organization that we did were now that all the credentials are in one folder I can make the assumption that every single every single file in that folder is expected to have a proof and if it doesn't have a proof I can generate it or if the proof fails I can regenerate it and that's what these small line.
Ben_-_Transmute: organization is doing.
Ben_-_Transmute: Okay either ping me or pull pin issue.
<ben_-_transmute> merge out of call okay
Ben_-_Transmute: Also I think we should probably add result of jazz to the getting nor file so that doesn't get committed so I will.
Ben_-_Transmute: I think so it probably looks like you committed in the previous one and it's probably gonna it's probably gonna thrash for every for every poor request so we can we can fix that out a call and just say merge once files fixed.
Russell_Hofvendahl_(mesur.io): I think we might have 512.
Russell_Hofvendahl_(mesur.io): Yeah sure this is pretty straightforward EDD one of the suggestions with EDD appear that was merged last week was to change SHP to shape which makes a lot of sense keep things consistent so this that's all this does.
Russell_Hofvendahl_(mesur.io): Yeah yeah this addresses all undefined terms in the nice Mama files that also where marriage last week.
Russell_Hofvendahl_(mesur.io): So this doesn't this is doesn't affect anything outside of the nice my files.
Russell_Hofvendahl_(mesur.io): Most of most of those a few issues where I just made blatant mistakes like with how with property names or maybe there was one case where there's a missing property there were also I think some cases where types needed to be defined better but yeah it was just kind of a cleanup thing.
Mike Prorock: Woman pulling it up the are details because I looked at this.
Mike Prorock: Yeah the are details were reasonably constrained prior and so there were some problems in there so that's what he fixed so he gave it the ability to return a raise and Cetera from are details.
Khalid_Gobin: Yeah but as for me.
Khalid_Gobin: Yeah so the idea of this was just tackle the issue of hosted Cosmic elections I was reading the Forum and I saw that the tasks were two further.
Khalid_Gobin: Like we did we discuss it in the chat where we were talking about the host Postman Collections and you said every moving away from that so I just added how to import the postman collections in the readme.
Khalid_Gobin: It is the other line.
Khalid_Gobin: Yeah I have one question regarding issue 100 which is the hosted Postman collections.
Khalid_Gobin: You okay yeah so you said that we need to add a list of collections to the Respec can you clarify what is the respect.
<ben_-_transmute> Respec document: https://w3id.org/traceability/interoperability
Khalid_Gobin: Yeah it does.
Ben_-_Transmute: She will put a tag on this if it's externally blocked or something too.
Ben_-_Transmute: I can make the label.
Orie Steele: It's what this is I guess it's just a comment also about the way that we're labeling or undefined terms because as you read them they kind of come up with these camelcase fragment based your eyes and there's a number of sort of unfortunate side effects that come from that like one it makes the term definition very unreadable and maybe that was your goal but that's sort of like throwing you're going.
Orie Steele: Throw an exception you might.
<ben_-_transmute> should we add "ready for PR"?
Orie Steele: Throwing nice exception right in the case of undefined terms it's not a thrown exception really if we're doing this at vocab mapping so I think we want to make it a little bit more readable a little bit friendlier to the folks that are going to produce tons of these I mean your first experience in json-ld as a Json developer is you're going to manufacture credentials with undefined terms that's what you're going to do and so we should be thinking about the.
Orie Steele: As of credentials and making them gentle but direct and so I think that changing the way we're handling at vocab there's options that we could there's things we can do like there that might be a little bit friendlier I think we should think about them I will leave some comments on the issue I think it's it's not done until we're done with undefined terms.
Orie Steele: Which will never be done with.
Orie Steele: It's other stuff to I'll leave a comment on the.
Ben_-_Transmute: Should Mark this is right for PR.
Orie Steele: I don't think it's ready for PR I think it needs to.
Mike Prorock: Orange defining new terms.
Ben_-_Transmute: Well know that way you can tell it was written by a true programmer.
Mike Prorock: Exam pant I mean it kind of makes sense it's an example expansion yeah.
Mike Prorock: Witnesses saying is that everyone except for obstinate me is well behaved.
Orie Steele: Archon I would just say I really appreciated having that conversation earlier Mike because it's this is what we're here to work on like we're doing a modeling issue and I think it's excellent and we should just refine it a bit more before we have such an open discussion on an issue but I think I like I like being able to also have these discussions and not just mindlessly cross issues with no commentary so I think.
Ben_-_Transmute: Yeah I I really I mean I would have been I would have been more than happy to spend the entire hour if we didn't have for request to get through that was very tempting.
Orie Steele: Yeah that I think is correct like we should.
Ted Thibodeau: +1 Vocal discussion is important and valuable
Orie Steele: +1
Mike Prorock: Well we totally my bad because I literally called Russell from like I did some weird place and was like Hey can you Johnson code down and I what I should have said was jot this example down in an issue is what I should have done instead I said hey just fire up in a PR and see what people say and that obviously was the wrong strategy.
<orie> but PRs do generate more conversation than issues :)
<orie> lol
Ben_-_Transmute: Have a great day.