[MINUTES] VCs for Education 2025-10-06

Here's a summary of the meeting:

*Meeting Summary*

The meeting of the W3C Task Force on Verifiable Credentials in Education
focused on self-issued credentials and their application, particularly in
the context of skill-based credentials. The discussion revolved around the
Links project, a platform designed to facilitate the creation, evidence
attachment, and recommendation of self-issued credentials.

*Topics Covered:*

   - *Self-Issued Credentials:* Definition, benefits, and technical
   challenges.
   - *Links Project:* Overview of the project, its serverless design,
   workflows, and the role of evidence and recommendations.
   - *Governance and Future:* The open-source nature of the project,
   funding, and potential for productionalization.
   - *Skill-Based Credentials:* The approach to defining and verifying
   skills within the platform, challenges, and potential solutions.
   - *AI Integration:* Discussion on the role of AI in the platform,
   including skill extraction and the potential for a metacognitive tutor.
   - *Multi-Skill Credentials:* A discussion on multi-skill credentials as
   a solution for aggregating multiple credentials.
   - *Identity Management and State Systems:* The role of state identity
   systems and their potential integration with the platform.
   - *Virtual Notary:* The concept of a virtual notary service and its
   importance in the credentials ecosystem.

*Key Points:*

   - Self-issued credentials provide increased accessibility, particularly
   for those without traditional credentials.
   - The Links project aims to make it easier for individuals to create and
   manage their own credentials, with a focus on skills.
   - Evidence and recommendations are critical for increasing the
   believability of self-issued credentials.
   - The project emphasizes a serverless design to avoid the handling of
   PII data.
   - Discussions on the challenges of skill taxonomy and the need for
   AI-driven solutions.
   - The potential integration with state-issued credentials and virtual
   notary services.
   - The importance of addressing the challenges of defining and verifying
   skills within the platform.
   - The need for a multi-skill credential.

Text:
https://meet.w3c-ccg.org/archives/w3c-ccg-vcs-for-education-2025-10-06.md

Video:
https://meet.w3c-ccg.org/archives/w3c-ccg-vcs-for-education-2025-10-06.mp4
*kte-hamg-bpj (2025-10-06 11:01 GMT-4) - Transcript* *Attendees*

Alex Higuera, Deb Everhart, Dmitri Zagidulin, Eric Shepherd, Hiroyuki Sano,
Ildiko Mazar, James Chartrand, JeffO - HumanOS, Karen Passmore, Kayode
Ezike, Naomi Szekeres, Nate Otto, Phillip Long, Phillip Long's
Presentation, Sharon Leu, Sheela Kiiskila, Ted Thibodeau Jr
*Transcript*

Phillip Long: We're there. Can you hear me? Okay, just making sure

Dmitri Zagidulin: Yeah, I can hear The Okay.

Eric Shepherd: Okay.

Phillip Long: We might make this an abbreviated meeting…

Phillip Long: if we don't have more people. Yeah.

Dmitri Zagidulin: It's always like that,…

Dmitri Zagidulin: It's a variable amount.

Dmitri Zagidulin: All welcome everyone. We're going to get started in about
a minute.

Eric Shepherd: To be fair,…

Eric Shepherd: at the last meeting I heard that it didn't start till five
after. So I went, great. I can do these three things before." So that's why
I was two minutes late…

Dmitri Zagidulin: I No worries.

Eric Shepherd: because I went, they're going to start five minutes late
anyway." That's the challenge of announcing you're going to be

Dmitri Zagidulin: No worries.

Dmitri Zagidulin: No, we usually try to start 3 minutes after more likely.
and I think All it's three minutes after. Let's get started. so, welcome
everyone to the weekly task force verifiable credentials in education.
these calls are recorded. we are under the W3C code of conduct and in order
to make substantive contributions you need to be a member of W3C and the
credentials community group. but aside from that these calls are open to
everyone. you don't need to be a member to just join the call.

Dmitri Zagidulin: Today before we jump into the main topic does anybody
want to do an introduction or reintroduction? Does anybody new to the call
or haven't introduced themselves in a while that you can just raise your
hand.

Karen Passmore: I'll introduce myself because I don't usually attend. So I …

Dmitri Zagidulin: Yeah. Hey Karen please do.

Phillip Long: Hello.

Dmitri Zagidulin: Yeah. Welcome.

Karen Passmore: So I'm Karen Passmore, CEO of Predictive UX. I work with
Phil and Dmitri quite a bit on the project that they're going to be
presenting today. so it's nice to meet everyone. Is that enough of a
background or do people go for

Dmitri Zagidulin: No, that's great. Thank you. yeah, and Karen does
excellent work in the fields of user interface and usability. she's been
doing working on a lot of projects in verifiable credential space on wallet
UI So we're excited to work with her on all of this.

Karen Passmore: Things to be true.

Dmitri Zagidulin: Anyone else? All right. community announcements. We have
IW coming up. I think we have Epic in Paris coming up.

Dmitri Zagidulin: We have TAC which is the W3C yearly in-person conference.
This one happening in Japan in November,…

Dmitri Zagidulin: early November. there's a big ITF meeting in Montreal I
think a week or two before TPAC. those are all the announcements I can
think of. let me know if I've missed anything.

Phillip Long: Isn't TPAC in Tokyo or…

Phillip Long: rather Osaka?

Dmitri Zagidulin: Yes.

Dmitri Zagidulin: Somewhere in Japan. I Yeah,…

Phillip Long: Yeah, it's a. Okay, right.

Dmitri Zagidulin: the ITF is in Montreal.

Dmitri Zagidulin: Yeah. yeah,…

Phillip Long: I and…

Phillip Long: I should mention that there is for those interested
potentially relevant to this conversation a meeting taking place in Utah
what is it next Thursday associated with recommendations for government
issued state credentials which Timothy Ruff is sponsoring and so if you're
interested in that you should contact Timothy Ruff.
00:05:00

Dmitri Zagidulin: that's going to be an interesting conversation. And thank
you, Kobe specifically near Tokyo is where TAC is going to be. All right.
in that case, let's move on to our main topic. So today we want to talk
about self-issued credentials a little bit. so we have Phil Long joining us
to talk about this who's been the sort of instigator and PI on several
selfissue credentials projects. before we launch into that, I want to say a
couple words on sort of introducing the context like what is a verifiable
credential or sorry what is a self-issue verifiable credential in the first
place. So let's take a quick look at that.

Dmitri Zagidulin: As always, the slides are going to be here in chat for
those who want to follow along and we'll hopefully remember to email them
after the call actually.

Phillip Long: Oops.

Dmitri Zagidulin: All So, everybody can see slide deck. All right. So
credentials what is a self-issue verifiable credential? So the very narrow
technical definition of it is it's a verifiable credential where the issuer
is the same as the credential subject. So that's the sort of technical
definition. What does selfissued Means the person who issued the credential
is the person about whom the credential is a person organization.

Dmitri Zagidulin: It also doesn't quite capture the flavor fully in the
sense that we also mean any sort of informal credential whether by a
natural person or an organization or institution where it's an unknown
issuer where the issuer is not likely going to be in any of the known
issuer registries because they don't know about those registries because
they're new to the industry. for meaning it's anybody who issues a
credential and then when a verifier receives it, it's going to be from an
unknown issuer. Right? So Bill's going to talk about this in more detail.

Dmitri Zagidulin: The reason we're interested in self-issue credentials is
of course dramatically increase accessibility. Not everybody has access to
traditional institutional credentials. and besides we want to explore the
very rich ecosystems and use cases where people can make claims about
themselves or even claims about other entities and then those claims be
progressively verified where they can accumulate third party corroboration
and so on.

Dmitri Zagidulin: And of course the challenge is you're probably asking
yourself so wait but then what does a verifier do with this credential
right so it's selfisssued which means you issue a credential for any value
of you the critical questions that a verifier has to answer is who's you in
question it's signed by an opaque identifier but who does that identifier
belong to typically that question is solved with looking up in an issuer
registry. But especially for natural persons, unlike higher education
institutions, they're unlikely to be in an issue registry. And in some
jurisdictions, there's strong legislations where they're not allowed to be
for personal privacy reasons in an issue registry. so who is the issuer?

Dmitri Zagidulin: And then once you identify that why should they be
believed right it's somebody's claim it's somebody's words and then of
course the technical challenges is as always how will our software handle
these how will these self-issue credentials will they be displayed Any
differently than institutional credentials? Will they just be flagged as
this is coming from an unknown issuer? and so on. Phil's going to talk
about this more in a bit detail. The current work that we've done that
we've experimented with in the realm of self-issue credentials is of course
adding evidence attaching and linking evidence to the credential greatly
helps its believability and then its claims.
00:10:00

Dmitri Zagidulin: And then of course recommendations and endorsements by
third parties including by somebody like notary public helps both with
identity verification and…

Phillip Long: All right.

Dmitri Zagidulin: in some cases where it's a subject matter expert helps
with the veracity of the claims and some topics for future sessions.
there's a lot more interesting work that needs to be done in the space
including adding user interface support for wallets and verifiers for this.
we have some interesting mechanisms coming in from the decentralized
identifier space such as did webvh on which we should do a presentation
because it's very promising did method sort of the successor to did web

Dmitri Zagidulin: but among other features it has a SL who is directory
where an issuer can accumulate trust marks or recommendations or identity
verification credentials that can be autodiscocovered by the verifier. this
is in contrast with u handing over the credential and any sort of trust
marks alongside a verifiable presentation versus handing over just the
credential and verify being able to autodiscocover the trust marks just
from the issuers's identifier itself. And then of course we have
tantalizing developments in the area of hey what if we give each verifiable
credential an inbox.

Dmitri Zagidulin: it's an odd thought from the start in the sense that
we're used to people and organizations having email inboxes or social media
inboxes rather than documents or credentials. But aside from that, in a
world where, even digital objects where we're having inboxes is essentially
free. By giving verifiable credentials their own inboxes, we can now have
some really interesting features. We can have the inboxes accumulate
updates and corrections to the verifiable credential.

Dmitri Zagidulin: And then for things like self-issue credentials, the
inboxes can accumulate further evidence and further recommendations, not
just after issuance, but after they've also been handed over to some
person. but yep, that's future work. And I'll hand it over to Phil.

Dmitri Zagidulin: But first, are there any questions from folks about what
a selfish credential is? All right, Phil. take it away.

Phillip Long: All right,…

Phillip Long: let's see whether or not this works. this is your screen, I
guess, isn't it? You haven't released it. you're on mute.

Phillip Long: Am I on mute?

Dmitri Zagidulin: I think you're sharing it.

Dmitri Zagidulin: I think you're sharing it.

Phillip Long: My goodness.

Dmitri Zagidulin:

Phillip Long: Okay, let me just start again then. I apologize. I am a bit
under the weather in that I had my that's the correct thing.

Phillip Long: in that I had my combination of COVID booster and double flu
shot the other day and spent most of yesterday in bed as a consequence and
so I'm still not 100%. so in any event this is the credentials that we are
presenting and hopefully I have the right deck. let me go to my full screen
so I can see them and see whether I can make it move. And it doesn't look
presenting here.
00:15:00

Phillip Long: No, I am not able to. All right, let me try that again.

Dmitri Zagidulin: So we can see your screen.

Phillip Long: You in All But I'll probably let me just go over to here.
Hang on. All right. So we've been talking about this as self- asserted
credentials and Dimmitri mentioned the importance of evidence and just to
clarify mentioned in chat we're also starting with skillbased credentials.
Demetri's already sort of described what it is that it's a self-issued
credential that has the issuer and the subject the same.

Phillip Long: that will lead to a concern later that we might want to have
wallets have a tag that comes up that indicates self-issued as opposed to
just we don't know what this is. the audience is essentially everyone who
doesn't already have an LE or may be unlikely to ever have an L. 40% of the
US workforce are uncredentialed, meaning they have neither a completed
degree of any sort or a cert certificate or license of any sort. And it's
anticipated that that's unlikely to change particularly dramatically in the
short term at least.

Phillip Long: And so for those people to be in a world where applications
for jobs are looking for credentials they are yet again u on the sideline.
So one of the things that's essential for this approach is that the app
itself is serverless. That to say it's entirely browserbased. The
importance of that was we did not want per any information that might be
PII related sitting in a server that someone was managing. and that was
important particularly for the Chamber of Commerce who funded this work.
because they did not want to have any oci responsibility for PII misused or
discovered and other sorts of things. And so that's the design.

Phillip Long: it's about a credential that has a narrative with credential
links rendered in a browser shared view via SMS email or potentially if you
have a credential wallet QR code and it can be combined with other systems
that involve assembling things like a resume whether that's smart résumés
or the p we did with respect

Phillip Long: to open which is a resume author demonstration tool and most
significantly we are working towards and we'll talk about this a little bit
later and there will be presumably a separate session on storage a web data
store where the credentials can be stored in addition to the design of this
particular credential at the moment which is that the credentials are
stored in Google Drive so most of you have seen this before I'll go through
it quickly there's a selfisssued workflow and there's a recommener workflow
and various var and variations of that expected to recommend the issuer
workflow is very straightforward. The same pattern of steps occurs on in
each case.

Phillip Long: You start off with describing the skill that you're going to
desri talk about, how long it took you to acquire it, a description of it,
and then to provide evidence either from things that you can embed in the
credential or things that you want to link to that support the credential
and the individual in the UI is encouraged to do that simply to add
credence to the claim being Once the claim is made, you can ask for a
recommendation. You can see down in the lower portion of the slide which is
the fourth over of five that ask for recommendations prominently there. And
then finally you can initiate that and des and acquire or rather acrue
these credentials in your Google drive.

Phillip Long: to complete a credential looks like. Yep.

Dmitri Zagidulin: Wait, before one sec.

Dmitri Zagidulin: As you're Let's pause on the screen. I want to make sure
to give people a chance to ask questions and also to say a few words in
context. So what the links project was was essentially Phil and myself and
some other people in including Karen asking the question of okay, we're
interested in this notion of selfish credentials. to make this happen? What
do we need to do for an MVP for a demo?
00:20:00

Dmitri Zagidulin: How do we explore this concept of how do we make it easy
for people to issue self-issue credentials knowing that just filling out
and issuing a credential is probably not enough that we want to be able to
offer enhancements such as like we mentioned evidence and recommendations.
So that's what you at. You're looking at a free open source app that
explores the concept of self-issue credentials by hopefully making it
easier by making sort of a walkthrough wizard for people to create selfish
credentials,…

Dmitri Zagidulin: attach evidence and ask people that they know for
recommendations. So that's the context there.

Phillip Long: It's Yeah.

Dmitri Zagidulin: Does anybody have any questions so far? Go ahead.

Deb Everhart: Can you…

Phillip Long: And it's a Go ahead. Yeah.

Dmitri Zagidulin: You're Excellent question.

Deb Everhart: then while that was good context, Demetri, can you say a
little about the current state of these tools and governance is a big word,
but who's governing it?

Phillip Long: You want me to take that? Yeah.

Deb Everhart:

Deb Everhart: What's the future of it? How is it funded? A little more
context along those lines.

Dmitri Zagidulin: That's yeah, one second. I always have that question as
well whenever any software is presented. So I'm right there with you. So
I'll let Phil answer that in more detail. the link to the current
deployment is in the very short answer that I'll give by jumping in is to
say governance is fairly open. This is open source software so open issues
but know that the dev team is small and the funds are limited. But over to
you Phil.

Phillip Long: and the intention is that we will hope someone will take this
frankly and do a more productized version of it.

Deb Everhart: Yeah, it looks great.

Phillip Long: We think that Karen developed is actually quite elegant and
works and works reasonably well. It works equally well if on a PC or a
laptop. But the project included funding which is a little bit unusual at
least through the end of the year and into pro and I think into early next
year for any support on all the pages there's a link to asking for support
or help. So there is a team that is following this and tracking any
questions that might arise or bugs that you might encounter so that people
can get a response hopefully reasonably quickly.

Phillip Long: And we're keeping track of that to see where things are that
we need to actually either change in the way we have designed the workflow
or make language clear in the text etc. After that it's a question that's
open as to whether it remains out there or not. We have of the intention of
having this available from the GitHubs that the US Chamber of Commerce
Foundation maintains. we've had a little bit of challenge with the chamber
in order to find who's got the actual permissions these days to allow us to
put it there. but other than that that's the intention to distribute the
code.

Phillip Long: We think that this will become more interesting and towards
the end of this I'll talk more about it with the respect to the new GB3 as
it's sometimes being called bill which is the government's big beautiful
budget bill which has documentation requirements for SNAP continuation and
continues documentation requirements for unemployment benefits.

Phillip Long: which this could potentially ease the current relatively
ownorous task of assembling stay tuned to that. You had a question the
followup. Then

Deb Everhart: just a quick follow-up question. So as you describe that
which is very helpful my head I put it in the same general category as
credential issuing and wallets which are source publicly available…

Deb Everhart: but not stood up as a production system. Is that a fair
comparison?
00:25:00

Phillip Long: Yes, I think that's a particularly fair comparison and…

Deb Everhart: Do you think that's comparison?

Phillip Long: and we would I Demetri, go ahead.

Dmitri Zagidulin: Yes, absolutely.

Deb Everhart: Okay,…

Dmitri Zagidulin: So, to be fair, T3 Chamber of Commerce is hoping to
productionalize it. but again,…

Dmitri Zagidulin: more not as to have billions of people use it and more
for its me chamber of commerce members to continue the conversations in
this field.

Deb Everhart: thank you so much.

Phillip Long: Right. Right.

Phillip Long: Which I will mention later but there is a business version of
this we've created for small businesses so that they could use this for
employees to make assertions about their skills and the to their supervisor
and their supervisor would do the recommendation endorsing or not that
credential as a way of keeping track of that for the performance reviews
and such.

Phillip Long: And we actually built a performance review credential inside
that some same self-issued workflow. And at the request specifically of the
chamber of commerce, there is a volunteerism credential because the chamber
has programs in existence amongst the chamber members to support
volunteerism activity, but they have no means of documenting it at least
any structured or consistent way of doing it. And so in that context this
turns out that it meets a need that they've had longstanding. so I'll
continue so the completed credential looks like any other you might
typically expect. We suggest people put in a nice image at the beginning
that's related to the credential topic. but it's their creative activity to
do that.

Phillip Long: and we thought that the innovation about this was just
basically with a really nice selfisssuer. And we very quickly realized that
we were wrong. And so we recognized that trust is not algorithmic or
cryptographic and a proof method. Trust is about humans making decisions
about individuals and relationships. And so we quickly implemented the
corresponding recommener workflow in order to have a connection between a
selfissue credential and somebody who is making a recommendation. And in
the recommener workflow the emphasis is on who the recommener That is in
terms of why are they in a position to make a judgment about this. in their
workflow the recommener is expected to include evidence or

Phillip Long: We strongly encourage evidence about their background in the
area that the recommendation or skill that's being claimed is pertaining to
and why they're in a position to make that judgment. so the recommended in
this case has how do you know the person proof of your qualifications a
comment directly about the individual and their skill claim and then
additional evidence that the individual who issued the recommen the self
asserted claim about their skill might have simply because the person
either knows them or has collaborated with them or has some other
third-party piece of corroboration for the skill that's being asserted by
that self asserted credentier.

Phillip Long: or you're asked when you're doing the recommendation and when
you're sending the recommendation to do so without simply a user interface
that invokes a mail and brings up automatically your mailer with perhaps
some pre-populated text or something like that built in. And the reason we
did this is that one of the key populations for this is intended to be
people that likely don't have their own computer. And if they're going to
do this, they're either going to have to borrow one or they're likely to
more often than not go to a public location like a library which has openly
available computers so that they can take advantage of a public machine.

Phillip Long: And in any public machine, whether it's a lab in a school in
a Goodwill office or in a library, there is not going to be a mailer built
in for the public. And so as a consequence, we simply pres present some
text that we suggest that they put in the body of the message for them to
edit when they actually open their mailer. and give them opportunities with
a button to copy that and then "Please open the mailer of your choice to
make your recommendation u or to send a request for a recommendation." And
in that way, while it's a little bit more awkward, it allows an individual
who has no access to a machine of their own and configured accordingly with
their own mailer to be able to still use it. we think easily enough for
longer term.
00:30:00

Phillip Long: we're working moving right right now towards improving the
ability for us to be able to ingest credentials into the workflow for
recommendation from other thirdparty issuers so that the recommener
workflow can be used for an issuer of of some other established entity
whether that's a college issuer or the product that the college issuer is
using territorium or something like that etc.

Phillip Long: as I mentioned a minute ago, we have this link credits for
business It does have a quasi employment verification thing where when the
individual starts off to make a self- asserted credential and claims
they're with this position in this business, they actually have to confirm
that with a six-digit code that's sent to an email that they present
associated with the business.

Phillip Long: We know that this is extremely weak and you can easily game
this but it's a step in that direction until we either have what Demetri
mentioned before is an online place that has some who will look up
capability or something like that and also some other mechanism that we
have yet to develop for confirming that an individual is working for a
business. If a business has its own email system and its own domain, then
it works quite well. But, if they don't and they're just using, Gmail for
free, then it's pretty weak. Dimmitri mentioned did the replacement we
think of webv, which has a system for updating log records of credentials
that you've issued.

Phillip Long: And so that opens up the possibility of keeping track of the
things that you've issued and recommendations you've received. it also
opens up the potential for devisizing some visualizations like a trust
graph to be able to see those as a set and perhaps down the road further is
a virtual note notary or digital fiduciary service where when you're trying
to deal with the question that Demetri raised at the very beginning and
it's true of all credentials to be quite honest if your identifier is a did
who is the piece of human carbon based living thing that did is really
associated with. And at the present time, our mechanism for doing that is a
notary or something similar where someone is actually looking at you.

Phillip Long: has a set of corroborating pieces of evidence that they go
through, a step-by-step process by which they check whether that evidence
in fact meets their criteria for being legitimate and makes an assertion
that says I've seen the person with this particular opaque identifier has
demonstrated control of the private key that control that is necessary to
use it and has corroborated evidence to the point where I believe that this
person is in fact who they say they are and therefore that this ID can be
associated with them. in this particular case they would in fact be issuing
a recommendation credential that would be accompanying the various other
recommendations that the person received. Sorry. so I mentioned these
immediate things about interoperability and getting recommendations.

Phillip Long: That's my fault for being a little bit out of it when I put
this together this morning. I mentioned the trust graph and the didd web.
We're also looking at approaching how to leverage the wallet attached
storage which Demetri might want to comment about a little bit in the
fiduciary service and I mentioned the impending GB3 requirements. There is,
as you probably know, GB the requirement for the continuation of SNAP that
the person demonstrate 80 hours a month of, either work or other activities
that they define as necessary for the person to continue to receive those
benefits. The current legislation, it makes the documentation about that
process reasonably ownorous.

Phillip Long: Whether that's intentional or not, one can argue, but we
think that something like this could be useful as a collaborative process
to document this particularly if the individual is doing so in conjunction
with a workforce development agency in a state because at least for the
continuing documentation of UI insurance that's been delegated to the
states as a source for designing their own individual way of doing that and
therefore this could be the place where someone is issues a credential and
the workforce development board is the initial location for a
recommendation to confirm it. and of course there's AI and I'll get to your
question in just a second T. and we are AISM is getting at everything at
this point.
00:35:00

Phillip Long: we've talked thought about in terms of skills taxonomy
search. When the person makes a self- asserted statement about their skill,
should they actually be using that as a prompt to be pulling back
structured and taxonomic ontologies that have what others out there think
is perhaps a better description of it. should it be used in a recommener
search or when the person says, who should I ask?" finally, is it in fact
possible to simply talk through the things that you want to do and have the
system develop the credential for you? And lastly is this might be a
mechanism for facilitating personal growth and development as a
metacognitive tutor.

Phillip Long: And I will point out that we have actually a project and Erin
is on the call to build a multi-skll credential in response to the feedback
that we have gotten from hiring managers saying that this credentiing stuff
is great but I don't want to have to deal with 40 or 50 individual single
assertion credentials about skills from someone. I need and want to have
that packaged in a way that stays together when I get it. And so this is
one thing that we're working on doing right now.

Phillip Long: Then.

Deb Everhart: Yeah, go back to the AI slide. so I have a comment here and…

Phillip Long: I agree with you 100%.

Deb Everhart: then a question about the prior slide. So my comment here is
that I think these last two bullets are great together. rather than having
an AI generated credential authoring which in a lot of ways could be a race
to the bottom in terms of yeah but to have an AI interface that is a
metacognitive tutor that helps someone understand their own skills and then
make that very simple AI assisted authoring.

Deb Everhart: I love the opportunity for combining those two things,…

Phillip Long: The AI credentiing thing reminds me of AI resume building.

Deb Everhart: Which is already erased to the bottom,…

Phillip Long: a cover letter writing which essentially takes the person out
of the bloody thing to begin with…

Deb Everhart: right? Yeah.

Phillip Long: which everybody recognizes as a mistake but is being used.

Deb Everhart: And then Yeah.

Phillip Long: You want me to go back to the previous slide? Sorry.

Deb Everhart: had a question about the previous slide. identity management
is partly a different topic from what we're talking about here. feel free
to deflect or defer this question. But how much do we think in these new
ecosystems especially some of the new requirements of GB3 that we can rely
on state identity systems?

Deb Everhart: especially since I think 17 states now have state driver's
licenses and how diverse are those systems or is that something that we can
start to hook into as the identity management component of this work at
least in the US?

Phillip Long: And that's a great question.

Phillip Long: And the reality in that space it is a mixture of
implementations at this point in time. It's not at all clear the inter
interoperability that will emerge between them. some are doing what
California has done which is basically using the verifiable credential
model for the MDL. So they're not using the ISO standard intentionally.

Phillip Long: and whether that was because they recognized the issues
associated with server verification or not that is the approach that they
took in part because they wanted to make it compatible with other
credentials which they are interested in having in the credential wallet in
California. And as an example of that, they included off the get-go the
true age credential for age verification without disclosing the identity
and other related information that typically is on a driver's license that
you have to be able to look at if you're going to determine age. So I think
that's a great call out about the concerns there.
00:40:00

Phillip Long: The other point I would note is that when I referred to state
assistance in this, I was particularly referring it to corroboration of
evidence from pursuing work opportunities. and that seems that's nothing to
do necessarily with identity per se, although obviously the workforce board
has to know who that person is.

Phillip Long: and presumably might have other mechanisms in place to assure
that go to us first before you use your credential for doing evidence on
this kind of thing so we can deal with it. Absolutely.

Deb Everhart: Right. At the very least,…

Deb Everhart: they're going to have to prevent fraud in those processes
through some identity management. So,

Phillip Long: and right now there is no mechanism whatsoever to do that in
any meaningful way frankly because what they have to produce when they
provide that evidence is a photocopy of the job application they saw it and
maybe something that they can get signed from the office that they
interviewed with or whatever. It's essentially very weak currently and
that's an issue that perhaps we can improve on. Illico, thanks for your
question. Dad

Ildiko Mazar: Thank you, l. my question is more regarding the definition of
these skills. earlier you had a slide showing those beautiful screenshots
of how people can self assert their skills…

Ildiko Mazar: but what's the guarantee that they can actually capture I
think there was a mention of an advanced barista or something like who can
say what's an advanced level of something and whether I guess my main
question is that if this tool can link to wellescribed tax skill taxonomies
that actually give a proper description of what's a proficiency level one,…

Phillip Long: right? Right.

Ildiko Mazar: two, three, four, whatever.

Phillip Long: Both very good observations. I did to some extent trick you a
little bit in that the particular thing about advanced skills was
associated with a certification that said advanced skills on it and came
from a reputable agency in this case the Red Cross. And so to some extent
my saying it in text was simply corroborated by something I had as an
external certification.

Phillip Long: But that doesn't answer fully your question because the truth
is we kept it so that the person wrote themselves what the description of
the skill was and labeled it themselves. Part of that was because when we
initially looked into doing something with a skill track what we found in
initial work was that the skill taxonomy brought back so many op returns
that is you did a search on that and you came back with 30 to 50 skills
types that were related to this particular thing you said.

Phillip Long: and remembering the audience that we're trying to reach, we
thought, that's not gonna work. They're just going to look at that, glaze
over, and say the heck with this. and so we simply punted to be honest and
said, we're going to have to rely on the corroborative evidence that allows
this to be triangulated and see whether that's good enough for the time And
the other thing we knew had to happen which is I think being something that
Karen might be able to comment on and you can do a skill search but there
has to be something in the front that once you get that goes through and
does itself an analysis of that set and determines what the relative
confidence is that the thing the person described is as close to that
taxonomic description in the end in

Phillip Long: inventory and gives them the top three or four choices or
something like that. we didn't have frankly the money in the budget we had
to build that intermediary parsing process and come up with something that
would be useful. So, we left it for some other time and with luck maybe,
Karen's work can be a step in that direction. do you have any followup that
you'd like to pursue, Elico?

Ildiko Mazar: Yeah. the reason why I asked is we all know that there are
taxonomies like the CFR which is the taxonomy for foreign language
knowledge and it very well describes the levels and it's very narrow so
it's very easy but I wonder if this is intended for regular people if there
could be some use cases or a component that supports the building of a then
sharable and commonly usable I don't know takes economy builder or
something. I'm just getting ideas. I'm inspired by your work,…
00:45:00

Phillip Long: Absolutely. If you could shoot me as the reference that you
just described,…

Ildiko Mazar: but yeah, I wonder if this would be something that would be
helpful.

Phillip Long: that would be very helpful.

Phillip Long: Then thank Eric, you have a question. Mhm.

Eric Shepherd: So Phil,…

Eric Shepherd: I'd love to understand your point of view on the difference
between roles, tasks, and skills because this discussion has been about
skills. But when I look at the user interfaces, it's saying elder medical
carer, barrista, and landscaper. And those are roles. And if we think about
elder medical care,…

Phillip Long: Close. Peace.

Eric Shepherd: there are tasks that you might have to perform in order to
care for someone who's elderly. And then there are different skills. And
the challenge is that it doesn't enable us to spot the missing skills in
order to learn them, in order to move up. So, for instance, if we're a
barist peels me out with a barristister, but if there's a number of skills
that a barista might have, such as customer service skills that are
transferable and if they choose to pick another profession they want to
move into because they've just closed their local Starbucks, then what are
the skills that they might add to those skills that they've

Eric Shepherd: developed as a barista, performing tasks as a barista.

Eric Shepherd: So, it feels like we're adding a little bit of confusion
between the task, and the skill by calling roles task. But I'd be
interested to understand your point of view.

Phillip Long: Yeah. No,…

Phillip Long: that's a really good call out Eric and thank you for that. we
struggled with this frankly because one of the challenges we were trying to
address was what is the target audience of this how is the target audiences
of this going to deal with the nuanced distinctions that you were just
describing because I'm skeptical that the average individual has a good
clear understanding of the differentiation among these things and therefore

Phillip Long: if we actually tried to say, use something like this, but
realize that this may be composed of individual skills, etc., that we would
lose them and simply dissuade them from saying anything. but…

Eric Shepherd: Yeah, I understand.

Eric Shepherd: I understand and I could see the wisdom of your choices. one
way you had spoken earlier was this idea of putting in multiple skills with
one shot.

Phillip Long: but yes, please do go ahead. No, no, go ahead. Mhm. That's a
great idea.

Eric Shepherd: And so a way you might be able to back into this is to group
those skills as tasks. whether you call them tasks or you call them some
name that the average man in the street or average person in the street
would understand that's fine. But you could kind of back into it that way.

Phillip Long: And I think that is one of the notions that we are going to
try to implement at least explore and see if we can in a meaningful way in
the multiskll credential. but you're absolutely right. I mean the challenge
when we were doing the initial lookups and we got the huge response back
from the particular taxonomy services where we were exploring was
overwhelming and so that led us down let it be free text and let's see if
we can improve when we start to see how it's being used.

Phillip Long: The other thing that we recently had conversations with
Workday and one of the things Workday mentioned was that particularly for
what they were referring to as frontline workers the things that they were
really interested in weren't really frequently the highlighted items in a
Litecast skills repository or things like that. because what they were
really looking for was evidence that the person showed up every day, that
the person was interested in and willing to learn new things on the job.
and if they could provide evidence of that, they would find that very
helpful. even if the person could self-state that without prompting, they
would find that very helpful.
00:50:00

Phillip Long: but they were very concerned in the conversations that we had
with them recently that the taxonomies just didn't get at the kind of
underlying behavioral characteristics that frontline workers with the
expectation that there was mobility and transition with reasonable
frequency in those positions would in fact put in useful skill statements
of the sort the taxonomies that is meaningful. And so all they seem to be
suggesting to us, yeah, let it be what it is and we can follow up with the
conversation and the hiring process to try to deal when we see inklings in
kind of behavior that we're looking for. So I give that as a

Eric Shepherd: of course. Yeah. I mean just to comment on that the behavior
is very much dependent on the incentives and…

Eric Shepherd: environment that an employer provides. So there is the

Phillip Long: Absolutely. Yeah,…

Phillip Long: absolutely. did I miss someone?

Phillip Long: Demetri, I see you're up.

Dmitri Zagidulin: So yeah,…

Dmitri Zagidulin: I think I'm on the queue. So first of all, this is
fantastic discussion. I do want to say that the question of skills roles
versus tasks is an incredibly hard one, right? in a way that's what all of
us in VC Edu and in open badges, that's what we've been struggling with for
years and decades now. that's way beyond this simple project to solve but
of course we hope to make steady progress on it.

Dmitri Zagidulin: So this was more of any sort of self-issue credentials
and it just so happened that skills credentials was something you could
easily explain to people that's really why we chose them of you want to say
you can have latte art claim it and then create some pictures and add a
recommendation from your I don't know barista manager. so Deb asked a
question in chat about more about the virtual notary. So I don't think
there's write up at the moment. It's just a general need that's emerging.
we hope to do an open project involving it with Phil and Chamber of
Commerce.

Dmitri Zagidulin: and just like you said in chat, it's very much not even a
stop gap, a needed core piece of infrastructure in the credentials world
because even with institutional issuance it's incredibly helpful especially
when registering an issuer to a known issuer registry to involve somebody
like a notary public and

Dmitri Zagidulin: course it can be done virtually over video call. I think
a lot of us especially since the pandemic have had some experience doing
virtual notaries over video. The opportunities there are many. The
challenges are of course getting any sort of organization to bootstrap the
process to recommend an initial set of notaries…

Dmitri Zagidulin: who can then onboard other notaries and so on.
Absolutely. Absolutely.

Deb Everhart: could be a useful topic for another meeting…

Deb Everhart: since we don't have time today. I'm very interested in the
relationship between commercial entities that already do that and what we
would want in this type of work.

Dmitri Zagidulin: Anyone else? I think we're nearing time and we want to
give people a few minutes before being calls.

Phillip Long: Sharon. Karen, did you want to speak? Yes.

Karen Passmore: Hey. Thanks, I've been commenting in the chats. So, just to
recap what I was saying there is that the approach that we're taking for
the skills extraction project is to take that free form text and infer what
the user is describing as skills and apply some structure to it. That way
we have a good foundation for future AI projects, but also we're working
from a place where the skills are more hopefully interoperable with other
systems that might be ingesting that data. And this is just a first pass at
it. So, I'm sure it's not going to be perfect, but I love all the feedback
that we've gotten here. and I'm taking notes and making sure that I'll
follow up with Phil and Dimmitri offline to figure out if there's anything
that we need to adjust about our approach.

Karen Passmore: But I think it's a good first pass and I'm excited to see
us just be able to get to the point where today we just have free form
descriptions of skills and then tomorrow once this project is deployed
we'll have structured data and then that's going to enable recommenders to
specifically recommend a person for one skill instead of just recommending
them for an entire credential that they've claimed. Because a lot of times
the way that we're doing things now because they're typing in the
description of their skills, they're kind of embedding lots of different
things in there. And when a person is giving a recommendation, they may not
want to recommend them for all of that stuff. So this is kind of giving us
a way to extrapolate the skills and get more fine-tuned granular
credentials.
00:55:00

Karen Passmore: And I think that was the bigger reason why we embarked on a
skill extraction project. Yep.

Phillip Long: Thank you,…

Phillip Long: Sharon, did you want to make a last comment perhaps…

Phillip Long: because we're close to the end here about your relationship
in edu related skills? No. Appreciate that. Very direct. Dimmitri, over to
you. Thank you very much by the way.

Dmitri Zagidulin: All right.

Dmitri Zagidulin: Thanks everyone. thank you Phil for coming to talk about
this. Thank you everyone for fantastic questions. you early on were asking
exact important questions in the chat about we need better support from
wallets to be able to sign external credentials. we need to have more
deployments on nondig key methods. totally agree with you on all of that.

Phillip Long: Cheers.

Dmitri Zagidulin: Great topics for future discussion. All right, thanks
all. Cheers.

Deb Everhart: Thanks everyone.

Karen Passmore: Yeah. Thank you. Bye. Bye.
Meeting ended after 00:57:03 👋

*This editable transcript was computer generated and might contain errors.
People can also change the text after it was created.*