Credentials Community Group

Harrison_Tang: Welcome everyone uh to this week's w3c shishi G meeting so this week we're excited uh to have the author of the paper uh Nemo paper neuron from MIT here to actually talk about her paper so the paper mitigating barriers to public social interactions with uh neuromuscul uh actually gather quite a bit of conversations about a month ago so there was uh quite a bit of uh uh emails uh on 2 or 3 email threats so very excited uh to actually have the opportunity uh to invite her to talk about her paper and then the hopefully uh have further discussions uh around the topic of me me me.

Harrison_Tang: Before we start just want to quickly.

Harrison_Tang: Uh do a quick reminder on the code of ethics and professional conduct just want to make sure that we continue to have respectful uh and constructive conversations.

Harrison_Tang: A quick note on the intellectual property anyone can participate in these calls however all successive contributions to any ccg work items must be the member of the ccg with full IPR agreement signed so if you have any questions in regards to getting a w3c account or the w3c community contributor license agreement please feel free to reach out to any of the cultures.

Harrison_Tang: Uh these meetings are being automatically recorded and transcribed uh so we will publish the transcription the audio and video recording in the next uh 1 or 2 days.

Harrison_Tang: We use GG chat to cue the speaker so if you have any questions just type in Q Plus to add yourself to the queue for a q minus to remove uh you can type in Q question mark to see who is in the queue and that will be moderating the queue.

Harrison_Tang: All right uh this is time for the introductions and reintroductions so if you are new to the community or you.

Harrison_Tang: Being active and want to re-engage uh feel free to just uh unmute and uh introduce yourself.

David_Karger: Hi there I'm David Karger since you've invited um I'm Nouran's uh advisor and I'm joining the call because I'm very interested in marigny and uh looking forward to this conversation.

Harrison_Tang: Welcome welcome David.

Harrison_Tang: Anyone else and you want to leave you I'll ask you to introduce yourself later before your presentation.

Harrison_Tang: Anyone else for the introductions were reintroductions.

Harrison_Tang: All right uh announcements and reminders any new announcements in regards to events or different papers.

Harrison_Tang: Thank you thanks Clea.

<manu_sporny> Example update from VCWG: https://lists.w3.org/Archives/Public/public-credentials/2024Jul/0000.html

Harrison_Tang: Thank you Manu.

Harrison_Tang: Any other uh announced announcements we're going first.

Harrison_Tang: Any updates on the work items.

<manu_sporny> There are 13 specifications being moved through the standardization process at W3C, currently!

Harrison_Tang: By a quick note on the work item so uh we will start asking the people to provide work items updates starting in August so so I think uh world and I kind of mentioned that uh about a month ago but just want to a quick uh reminder and uh we'll has a reach out to different people in regards to a different work items uh so so that if uh.

Harrison_Tang: If the the leaders of those work item and initiatives are present uh they can provide an update if not uh if other people have uh know what's going on in those work items it would be good uh for them to kind of give a quick overview of what's going on in those work item and progress.

Harrison_Tang: Um any last calls for the work item updates announcements or introductions.

Harrison_Tang: All right uh let's get to the main agenda so as mentioned uh.

Harrison_Tang: A month ago uh you know we had 2 3 Threads around uh this concept of Marin neimy and uh a quick uh note marinin is means like a part of a hole so kind of like an engine is uh marinate of a car for example and by the way I did look it up in the dictionary otherwise I wouldn't have know this so uh.

Harrison_Tang: It is a pretty cool concept and uh uh meronem actually kind of allows right people in the public space to selectively uh disclose only relevant and verify aspects of their identity uh to uh carry out public discourse so uh we we had a great conversation around that and we're very fortunate to actually have the author of the paper um New Orleans from MIT uh to actually talk about uh her work so you are the force yours if you don't mind kind of introduce yourself a little bit and then uh.

Harrison_Tang: You know talk about your paper and uh we're very excited to have you here.

Nouran_Soliman: Oh of course thank you Harrison so my name is Nan.

Nouran_Soliman: I'm a PhD student at MIT um I work with David Karger who is also here in the call um and.

Nouran_Soliman: Um I got distracted by the.

Nouran_Soliman: Notifications I thought the voice is not clear um okay um yeah so I work with David on um problems around um.

Nouran_Soliman: Online spaces and how to make them more safer and more trustworthy.

Nouran_Soliman: Um and um yeah uh so generally I focus on online trust and um safety um and today I would like to talk about uh this project around me which I'm we managed to publish in um Chi which is a human computer interaction conference um this year so uh I'm going to share my screen so I'm going to turn off my video and I'll start sharing my screen.

Harrison_Tang: Thank you thank you and you are sorry to interrupt Joe do you have any comments or questions.

Harrison_Tang: No problem no problem thank you.

Nouran_Soliman: So can everybody see my screen.

<kaliya_identity_woman_> Just so you know the voice iant cominf throgh on the app on the phone.

Nouran_Soliman: Okay sounds good uh okay so um the the work I I want to uh talk about today is called mitigating barriers to public social interaction with miranis communication.

Nouran_Soliman: Um so um basically in this uh talk I'm gonna uh present Mighty which is a novel design Paradigm we designed based on our formative study and then I will talk about the evaluation of this uh in a system that we built named literature uh to incorporate the mirror nity framework.

Nouran_Soliman: So to start um if if we look at online public spaces we we will see like how we transformed uh public engagement by bringing people together uh in these cases where they can discuss an exchanged personalized information based on their specific needs and experiences.

Nouran_Soliman: so you.

Nouran_Soliman: Like so many different platforms like credit um Facebook um Goodreads and so on and so forth.

Nouran_Soliman: However there are a lot of social barriers that arise in these online communities um and and the are these these barriers are usually around self-image um like fearing to look stupid or um or burdening others or uh also uh so part of it is about self-image and the other part is about consumption of your Social Capital like as I said mention burdening others or uh having to reciprocate the obligation back and so on and so forth which usually discourages a lot of people from engagement despite the perceived benefits Behind These interactions.

Nouran_Soliman: Um and you for example you might have seen like similar experiences uh in online spaces around.

Nouran_Soliman: Um academic Twitter where like uh people post things um asking for papers or um kind of sharing uh their papers and so on and so forth.

Nouran_Soliman: So we were interested to look at Academia specifically in this project and um because of its interesting structure so if you take a closer look on Academia.

Nouran_Soliman: It's very hierarchical and it has a lot of rampant Notions of status and Prestige coming from specific institutions um which kind of creates uh a lot of.

Nouran_Soliman: Potent like potential professional risks and competition in that are reflected in these online spaces.

Nouran_Soliman: And although these Dynamics are uh prominent and in Academia it there is a growing number of academics that are drawn to social networking uh platforms like academic Twitter for example uh for the visibility this offers uh to the research uh as well as the the authors so.

Nouran_Soliman: We see a lot of people like tweeting uh about their papers um.

Nouran_Soliman: And so on and so forth uh and and previous work has also shown that there is uh increased visibility and uh also uh like when using these uh platforms and also it helps uh researchers kind of.

Nouran_Soliman: Construct uh a professional identity online.

Nouran_Soliman: But despite these benefits uh of of the public scholarly engagement similar barriers arise so for example Rising Juniors often fear the impact of their public engagement on their career progression due to like um.

Nouran_Soliman: The high stakes uh.

Nouran_Soliman: The high sticks that are attributed with such engagement because it's career related and it's going to reflect on them forever so in in the professional Community which usually makes them deter from engagement and therefore these communities become dominated by senior voices and also um becomes imbalanced in terms of like gender or other minority groups.

Nouran_Soliman: So in this work we particularly focus on online communities for academics as our test bed but the first question we looked into is in a general context what are the previous solutions for addressing these social barriers.

Nouran_Soliman: So 1 obvious solution is anonymity um so anonymizing 1's identity introduces social inhibition that helps users overcome a lot of these social pressures online however it still motivates toxic behaviors um and um it also de which which as a result the motivates engagement um and also obscures a lot of the social context around the interaction.

Nouran_Soliman: There is another solution um which is student limiting so you use a persistent synonym to engage uh such as spaces like credit um although this kind of provides a degree of privacy and a sense of accountability um.

Nouran_Soliman: There's still a problem where it's very hard to skip negative reputations or hide information from previous interactions which basically leads to self-censoring and or like in some occasions the creation of throwaway accounts where people like use the account for 1 for a 1 time thing and then they just like throw throw it away.

Nouran_Soliman: So the question we were asking is how do we balance between The credibility and context of full identity disclosure and the privacy and protection of anonymity and this is where Mirren nity comes.

Nouran_Soliman: So basically uh the fundamental idea behind marigny here is that users can reveal specific aspects about who they are with every interaction.

Nouran_Soliman: Um and it comes from um marrow um which in Greek means partial and Nim is named so it's kind of partial anonymity.

Nouran_Soliman: So for example if I am publicly asking the academic Community about paper recommendations let's say on on a space like Twitter and I'm concerned about doing this from my personal account with my name attached so.

Nouran_Soliman: The idea there is that okay I can do this by hiding my identity but instead of saying that I'm an anonymous poster I can reveal some information about myself so for example my seniority level saying that I'm a junior researcher for example or my affiliation.

Nouran_Soliman: Saying that I have worked before or still working.

Nouran_Soliman: University uh or maybe a combination of all of those or my um network of co-authors uh or other uh signals um in my background so this kind of gives the audience a better sense of the poster is identity making the interaction feel more realistic and less random and and while providing uh context around the interaction.

Nouran_Soliman: So uh the first step we did is we ran a formative study with 20 Scholars uh in order to better understand the nature and challenges of online scholarly Communication in general and also in help seeking context such as asking for or providing research paper recommendations.

Nouran_Soliman: And uh participants confirmed uh their perception uh of these professional interactions as high stakes.

Nouran_Soliman: A lot of the time stops them uh from engaging with the community and asking for help.

Nouran_Soliman: Um most participants in fact expressed Their Fear to be judged as ignorant or stupid or needy even though they know the value of um asking for paper recommendations in in these spaces and how they consider um papers defined there as like Collective wisdom.

Nouran_Soliman: There was also a a clear preference for um knowing some signal about uh who the interacting with instead of complete anonymity.

Nouran_Soliman: Like a lot of participants stated that they are unlikely to engage with random questions from Anonymous uh people.

Nouran_Soliman: So in general the results emphasize uh the existing social barrier in these uh spaces and the potential for meronem as an effective solution.

Nouran_Soliman: So we used insights from our study to support the design of our mirin emiti framework uh and then we built a system named literature on top of Twitter and mastered them that incorporates this model.

Nouran_Soliman: so next.

<manu_sporny> That name is awesome "LiTweeture"

Nouran_Soliman: I'm going to present um our meronem framework and literature by walking you through a user scenario about uh a researcher who asks for paper recommendations.

Nouran_Soliman: Um and we were particularly interested in this use case uh of asking about research paper recommendations because in our formative study we found that it has a high value of our Scholars and also um there are um a lot of Social Challenges around the help seeking part of this um interaction.

Nouran_Soliman: So this is Alice this uh Alice is a junior student who works in Ai and Alice wants to explore the literature for a new project.

Nouran_Soliman: But then Alice remember is coming across high quality recommendations on Twitter and decides to ask AI community on Twitter.

Nouran_Soliman: However she doesn't want to look stupid or ignorant and she feels anxious about doing that publicly.

Nouran_Soliman: So Alice decides to sign up to the Tweeter which supports miranis interactions on Twitter and on mastadon.

Nouran_Soliman: So basically on sign up Alice has to link her claimed author profile on semantics scholar and her Twitter account to litscher in order to verify her identity so the system then uses uh Alice's publication history and Twitter network uh to compute verified identity description uh descriptors about Alice that can be used to compose Mirren when posting her questions.

Nouran_Soliman: But then since Alice is a junior she doesn't have much publication history so she adds Mark as her endorser uh through the Tweeter which means she can leverage marks reputation to present her identity mirrorless by using Mark's identity signals along with hers.

Nouran_Soliman: So for example she can present herself as someone who is endorsed by Mark.

Nouran_Soliman: So Mark then get uh Alice's request he then signs up uh to literature again with a claimed author profile uh so the process of claiming an author profile on semantic scholar you uh sign up there um it's maybe a couple of days sometimes it's quicker where they have like a rigorous verification process that I think is done by humans involved um and then uh basically once your once your profile is is claimed you're able to sign up to the Tweeter um and then also you Pro so so Mark has to do that as well and then the also he also provides his Twitter account.

Nouran_Soliman: Um and then he accepts Alice's request indicating advisor as the relationship um and then Alice now can compose her question select um and mix and match identity signals about herself as or her endorser uh to present her her identity.

Nouran_Soliman: So it looks something like that um so basically now Alice can uh has her question posted on her behalf so uh on on Twitter and mastered it through literature on a dedicated Tweeter page um with her miranis identity so uh at the top there is the question and then following is uh the identity and because of the Twitter character limit the identity is uh.

Nouran_Soliman: Like uh posted in in in a in another tweet in the same thread.

Nouran_Soliman: How will relevant experts come across the request.

Nouran_Soliman: So um Alice can actually add experts to her question to her request through the Tweeter who are people she thinks are best suited to answer her question so so Alice decides uh to add an AI expert Rita to her request um.

Nouran_Soliman: And on adding an expert on the Tweeter the system tries to figure out uh relationship signals between the expert and the asker so Rita and Alice and the expert and the endorser Rita and Marc.

Nouran_Soliman: Um Alice remembers that Rita collaborated before with her indoor third Mark so she decides to disclose this connection to Rita.

Nouran_Soliman: So basically Alice can um privately disclose these 3 that through email or direct messaging um so for example Alice can disclose to Rita that she is someone endorsed by Mark who some who who basically worked with Rita before.

Nouran_Soliman: That let's say she follows Rita on Twitter and vice versa.

Nouran_Soliman: So basically uh she can privately disclose more relationship signals uh specifically uh to each expert.

Nouran_Soliman: And then uh Rita uh receives uh the message uh she quickly recognizes the connection uh of the Oscar to her dear collaborate her Mark and that the Oscar is a junior student from the community who she follows on Twitter so she feels compelled answering the question on Twitter from her personal account.

Nouran_Soliman: It looks like uh to look something like that so now Rita are commented um.

Nouran_Soliman: on the.

Nouran_Soliman: But then now some experts uh will be able to see the question but how will more people from the community come across the post since like.

Nouran_Soliman: Um the questions get posted to the dedicated literature pages on Twitter and on masculine so um maybe rather than people do not follow the Tweeter uh so how do we increase the visibility of um.

Nouran_Soliman: Something of of these questions.

Nouran_Soliman: so for.

Nouran_Soliman: That we introduced the helper uh so Alice adds Dave her friend um as her helper so basically a helper in literature is someone who is motivated to help their colleagues uh that the poster trusts and basically the the helper receives a private request from the asker through email or direct messaging.

Nouran_Soliman: Identity of the Oscar disclosed so the idea here is that this is your friend uh their motivated to help you and um they kind of retweet your or reshare the content you posted uh to their networks in order to uh.

Nouran_Soliman: Basically increase the visibility and for that we kind of disclose the identity fully uh privately in order to basically provide this motivation of helping colleagues or friends uh in answering questions so kind of leveraging the idea of social capital in an academic context um and um um and that's why it's kind of like oh these are like the colleagues the close relationships.

Nouran_Soliman: Uh so in this scenario they've received a message saying um your friend Alice is asking you to sh to reshare her morinomiya the question is miranis but privately Dave knows that Alice asked this question.

Nouran_Soliman: Attendees feel motivated to support his friend by retweeting her question uh with a word of endorsement.

Nouran_Soliman: So um now.

Nouran_Soliman: more people.

Nouran_Soliman: People can see the question uh and through literature actually users who come across the post can also contribute to threads mirrorless.

Nouran_Soliman: In a similar manner manner as as described before so basically uh in this example the last tweet in the thread is an example of a user Who provided a paper recommendation through the Tweeter and chose to describe themselves as a junior student.

Nouran_Soliman: And uh of course for that uh responses have to be moderated uh by Alice before they get posted and broadcast it to both threads on Twitter and on master.

Nouran_Soliman: So now Alice feels happy and validated that her questions received engagement like and responses.

Nouran_Soliman: So how do we evaluate that uh so our evaluation aimed to explore how participants interacted with the key features of literature for seeking paper recommendations uh compared to the you their usual approaches we also wanted to gain insights into how the interacted with identities.

Nouran_Soliman: So we deployed that feature in a 1 month uh within subjects longitudinal field study so basically we asked participants to ask for people recommendations using the feature versus any other method uh of their choice uh so on signup participants were randomly assigned um to 1 condition um and then after 2 weeks uh we switched the condition and um basically the 2 conditions 1 is um ask a question and provide a paper recommendation using the Tweeter each week and the Baseline is basically just asking a question using regular methods that they would feel like using um.

Nouran_Soliman: We then use the data from surveys so at at the end of each weekly tasks we asked uh participants to fill a reflection survey um and at the very end of the study we uh we interview participants to kind of gain more qualitative insights around um their experience so we used results from surveys interviews also action logs and task entries in our data analysis.

Nouran_Soliman: So I'm going to describe uh some of the results but you can like refer to the paper for more detailed discussion.

Nouran_Soliman: So first of all uh our results show that participants felt really comfortable reaching out to experts they would normally be intimidated by while wearing a little bit of a mask using the feature with evidence on contacting more experts they don't have an existing relationship with um unlike the Baseline condition so in the Baseline condition actually participants never reached out to someone they don't they haven't talked to before or the never had or they don't have a strong relationship with.

Nouran_Soliman: Um but in the literature condition that actually happened that there were more participants reaching out to uh people they've never talked to before or people they they learned uh they have they work on a similar line of work but uh they don't have a relationship with.

Nouran_Soliman: Um and in a lot of cases this actually yielded more novel paper recommendations.

Nouran_Soliman: Um the other big result is uh participants also strategically revealed certain aspects about their identity in order to balance between credibility and also uh Express the relevant experience while maintaining their anonymity when asking questions so um you can see the quotes on the screen where uh 1 participant said oh this provides a good balance between anonymity while also showing people that me and my indoor sir have the relevant experience in the field and that we are part of the community.

Nouran_Soliman: Um also uh a number of participants really like the idea of having an endorser where they can actually.

Nouran_Soliman: Kind of show that.

Nouran_Soliman: Oh we were really uh part of this community we were credible we're really serious about our questions.

Nouran_Soliman: Um my name is also provided context to participants when answering a question so 1 participant actually mentioned how they tried to tailor the response by providing Chi papers to the asker when the asker uh mentioned that they published before at Chi so uh basically kind of giving context on.

Nouran_Soliman: What kind of Oscar or why the Oscar is asking this question in which context and this was also interestingly reflected as well when people were providing answers so for example when.

Nouran_Soliman: Uh someone was providing uh paper recommendation and the felt like very confident about this recommendation they would uh most of the time signal uh their publication history and prestigious conferences or maybe their affiliation something that kind of indicates expertise versus like if they know of something relevant by but they don't think it's the best.

Nouran_Soliman: Uh recommendation they usually uh try to signal that by just sharing very basic uh signals things like Oh I'm a junior student uh things like that so also kind of like signaling the experience and uh and this was also reflected in how we perceive the recommendations as well like if it's from a junior student it's usually perceived differently um um.

Nouran_Soliman: Um different from like if if a senior student or someone who has knowledge in the field uh provided that recommendation.

Nouran_Soliman: So although our results um have shown great potential for meronem there are still risks to be taken into consideration um such as calibrating the social pressure um so the idea there is.

Nouran_Soliman: We don't want to literature to make it too easy to ask questions and burden others um so it's really important to think of ways to.

Nouran_Soliman: kind of.

Nouran_Soliman: Make it hard enough uh to mitigate the social uh pressure and yet um.

Nouran_Soliman: Kind of um do the.

Nouran_Soliman: sum amount.

Nouran_Soliman: I've heard uh in order to um ask questions.

Nouran_Soliman: Um the other risk is uh the possibility of the anonymizing individuals if they identifiers are very unique um.

Nouran_Soliman: For that we we actually provided um.

Nouran_Soliman: ERS with the flexibility to mix and match uh whatever identity you would like to reveal um participants were generally aware of uh this idea some were really uh more cautious than others uh more cautious participants tended to reveal very minimal signals uh but yet still relevant uh while other participants at the end of of the other spectrum is basically felt well I don't mind if actually someone.

Nouran_Soliman: Try is to do the work and the anonymized it just I just don't want my name to be in there.

Nouran_Soliman: um so there is that um.

Nouran_Soliman: Variability and also some participants felt more comfortable re uh uh revealing more signals about themselves the the longer they use literature so it's kind of like a problem of getting adjusted and trusting the system.

Nouran_Soliman: Third um concern is the idea of the rich getting richer effect where.

Nouran_Soliman: The most connected people would would receive most of the attention um so if I say that my endorser is amazing and I have all this experience maybe this gets my questions more answered um and for that we uh we think that it's really important to try to also um create more inclusive signals so signals that highlight if someone belongs to a minority group or maybe uh more signals about uh people or gender or and so on and so forth and kind of.

Nouran_Soliman: Help Oscars provide more uh relevant details that could motivate the community to support these uh groups.

Nouran_Soliman: Uh we haven't implemented that but we think this is a really promising uh Direction which also raises um question of how do we verify these claims.

Nouran_Soliman: There are also interesting um directions in improving meronem in terms of combining meronem with pseudonymity so basically.

Nouran_Soliman: Someone can uh um combined a bunch of signals to create a Persona and then use this Persona to uh build reputation uh so basically you would have different mirror names and with each mirror name different reputations uh so we think this could be really interesting uh to try.

Nouran_Soliman: And then um finally you um something that I'm uh currently actually trying to pursue is broadening the application of Marin imiti to other use cases uh outside of Academia uh actually inside and outside but um I'm very interested to think about ways to apply meronem in a.

Nouran_Soliman: More General conversation that is sensitive uh in a professional organization or workspace such as the MIT campus um and how do we uh incorporate that uh what kind of signals do we use there and how do we verify these signals and also on the other side if if it's a if if this is going to be a sensitive conversation about contentious topics such as workplace harassment or um mental health or other like uh conflicts in the world how do we uh protect the receivers as well so there is a very um Central moderation piece here as well that we need to think about.

Nouran_Soliman: So I guess that's kind of the end of my talk um I'm going to leave this up here um with like the takeaways from the talk and thank you for listening and I'd love to receive questions or feedback.

Harrison_Tang: Great uh Greg you're in the queue.

GregB: Sorry that was an intentional.

Harrison_Tang: Got it no problem man wait.

Nouran_Soliman: Um I guess this is really wonderful uh kind of shipping something like that out in the wild and collecting data or trying to see how people use it uh I'm not working on that with anybody um I'd be very interested to work on that um I don't know if David also has other thoughts around it but that would be my personal take on that um and in terms of the research being ready for this.

Nouran_Soliman: I think so personally because we kind of try to test that in a realistic environment.

Nouran_Soliman: So um 1 month study where we got people to try to use it and we talked extensively about the idea we had them try it and it seems really promising so.

Nouran_Soliman: Um that would be something I'd be really uh interested to to collaborate on.

David_Karger: Uh yeah I'll just throw in a little bit on this question of of protocols I think it's quite interesting um and.

David_Karger: I mean you asked about you know meronem is a concept and it's interesting to wonder about the the maturity of a concept right I'm sure we'll figure out a lot of other things about meronem uh as we go forward but that doesn't really preclude thinking about how to introduce it already right so if I think about activity Pub um which is a a protocol we're using in some other projects we're working on.

David_Karger: You know Marini just raises the general idea of allowing people to uh make activity Pub posts that don't specifically identify the author by uh by by a a a clear identity uh but instead provide a a meronem uh within the post and I think I think.

David_Karger: you know.

David_Karger: Know exactly what the Marin be um and how it would be verified are very interesting open questions but we don't have to wait for a complete maturity of the meronem concept to be able to think about that basic question of just how to introduce meronem into activity P posts uh which I think would be really natural and and conceivably very useful.

Harrison_Tang: By the way a clarification question so for the literature uh literature is that a working thing like that's launched or just a prototype.

Nouran_Soliman: Um it was working uh until um Twitter API changed and right now we took it down so uh I might do some um engineering and just.

Nouran_Soliman: Back working uh with masteron but right now it's down.

Harrison_Tang: Got it thank you.

David_Karger: Yeah we we we unfortunately we launched this I think 1 week before they made the sudden change to the Twitter API so that you can't really use Twitter for anything anymore.

David_Karger: But I think we could it's it's only a matter of engineering to get it working with Mastadon again.

Harrison_Tang: Got it thank you.

Harrison_Tang: Phil your next in the queue.

PL/T3: There am I on now um.

PL/T3: First of all a great talk and very very nice to read the paper accompanying it um I'm just curious your perceptions of the just the degree to which um the various decisions that someone needs to make with respect to disclosing their identity.

PL/T3: Introduces friction to the point of of of uh interfering with the potential functional.

PL/T3: Come that you're seeking I can see it in the context of this particular use case with particularly high stakes from the academics point of view in terms of a junior researcher or student even.

PL/T3: To or wanting to get feedback from a senior person and you pointed out that you know those that information is going to be around forever so how they appear and and the way in which they comport themselves in this context carries a lot of weight.

PL/T3: I'm just wondering if you have thoughts about it in terms of diminishing the ways in which the friction to make selections over identity disclosure could be reduced so that it is more likely to happen in places where the motivations perhaps are slightly diminished thanks.

Nouran_Soliman: Um I I'm not sure I understood the question right but um.

Nouran_Soliman: The question like.

Nouran_Soliman: About the friction between revealing.

Nouran_Soliman: A lot about oneself that it basically interferes with the idea of.

Nouran_Soliman: Having things Forever on the Internet or.

PL/T3: The the question really is um the motivation for making a selection about what you're going to disclose take some thought and and and consideration and and it makes sense that someone will go to the effort to do that really carefully when the stakes are High um it's not clear to me that it would be as likely to happen um when you're interested in that a feedback in a particular set of outcomes.

PL/T3: Um the stakes may not be so dramatic and the question is is the effort it takes to make those decisions about the the balance of what's in the marinich um disclosure component um Ma you know what do you have thoughts that you might have to make that effort minimal and yet useful.

Nouran_Soliman: Uh okay that's a good question um.

Nouran_Soliman: In terms of effort I think.

Nouran_Soliman: In this specific study for example.

Nouran_Soliman: It was uh computed set of.

Nouran_Soliman: Signals uh that is exactly the same every time people want to interact so basically these just.

Nouran_Soliman: Familiarize themselves with it um and um kind of.

Nouran_Soliman: It again and again so that kind of helps with this.

Nouran_Soliman: I guess like the like it's and what I've seen is that people.

Nouran_Soliman: tend to.

Nouran_Soliman: To a number of signals most of the time so it's like maybe 1 or 2 conferences uh 1 or.

Nouran_Soliman: To um like the basically built a strategy and try to apply that every time um so I guess it I didn't see a lot of uh effort in in like um composing the identity as much as actually inviting experts and having to think about who has relevant work and so on and so forth.

Nouran_Soliman: uh so.

Nouran_Soliman: I think in terms of.

<pl/t3> A deterministic meronym is computed for them so the judgment to use it is less than I thought.

Nouran_Soliman: Making it easier um I mean things we've been thinking about is um.

Nouran_Soliman: What if we actually provided.

Nouran_Soliman: Like allowed people to just write whatever they want like in a free form text like would that actually how would that look like and how do we verify that uh but assuming there is a way to verify these signals I suppose.

Nouran_Soliman: This could also.

Nouran_Soliman: Be aware of not overwhelming users with a lot of signals to pick from um.

Nouran_Soliman: Yeah I I'll have to think about that more uh but generally this is what comes to my mind uh relevant to your question.

PL/T3: Thank you very much.

David_Karger: Um I mentioned an interesting or orthogonal there's there's actually a lot of work on identity disclosure um but it's it's interestingly orthogonal what what's usually studied is um the effort involved for people deciding whether to disclose a particular facet of their identity um for example if they're a member of some marginalized Community um so in in in that work which has been studied the identity of the individual is known but the what they wrestle with is the costs and benefits of disclosing some particular facet of their identity um Marin imiti kind of takes that works in the opposite direction um you start by disclosing facets of your identity uh and hopefully your your actual identity is not disclosed at all.

David_Karger: And of course that's that's the big risk uh is if you disclose too many facets of your identity uh it may make you identifiable um and and destroy the the partial anonymity that you're trying to preserve.

David_Karger: And so 1 of the things that we've talked about is uh how our tools can help a user uh be confident that they are not disclosing too much um you can imagine for example that if somebody uh generates a a mirror onimus description of themselves um a platform can actually tell them how many other people also fit that mirrors description.

David_Karger: Um so that's an example of Systems Support to help people with this decision making around uh the disclosure but but certainly it is a very important topic for for for thinking about the costs and benefits um of these sorts of disclosures and the risks that you are facing in having your true identity disclosed.

<pl/t3> @David - that's great observation. facet disclosure by anonymity preserved versus of identity disclosure and revealing facets

<harrison_tang> i feel that meronymity in some way just changes the "trust anchor" from an individual (e.g. someone) to a group (e.g. some organization or group). so people just have one more tool in their tool belt

Harrison_Tang: Dimitry you're next.

David_Karger: Uh we would we would love to have that conversation make that presentation can can you reach out to neuron to sort of try and move that forward.

Harrison_Tang: I'll make the I'll make the introduction.

Harrison_Tang: You you're next.

<dmitri_zagidulin> will do!!

<pl/t3> @Nouran @David Dmitri and I are also working on a recommender service n to VCs and introducing meronymity into that would be teriffice. That's under LinkedClaims as a topic.

David_Karger: Uh that would be lovely uh I mean any any and all of the above you know obviously we're we're a very low Resource Group uh but you know we we have an idea that we think could be beneficial and so if there are things that we can do to advance that that idea towards people who can help turn it into a reality um you know that that's very much 1 1 of our goals is to is to improve online discussion generally and we think that marigny can be a part of that so if you think that there are people who would be interested to hear about it.

David_Karger: Different uh subgroups we'd be happy to talk to them.

Harrison_Tang: Great um great your next.

GregB: I wanted to say on the infrastructure part of things um and the verifiable credential group we've been working on.

GregB: Pseudonyms unlined disclosure and these are the kind of things that could be very beneficial to kind of building up the infrastructure it's it's a level down so you have a credential that's selectively disclose that's unlink we do worry about an anonymity set kind of things and.

GregB: some of.

GregB: Requirements go and feed the current work on certain types of signatures going on at the um ietf uh cryptographic group so if you've got inputs you're interested in some of these things we could try and help out on some of that infrastructure stuff uh we've got uh some specs that are getting pretty mature but and we're focusing on some of these things like pseudonyms at a low level.

GregB: Could be of help so being.

David_Karger: That's really great and Greg if I could fire a suggestion back um you know we did this work with the semantic scholar team.

David_Karger: And our our Marin are generated through uh through verification by semantic scholar.

David_Karger: And I wonder if they would be interested in uh sort of providing a more General um credential service of of the sort that you mentioned right um and if you want uh to be connected to the I mean you can see who our co-authors are who are at semantic scholar um but that might be an interesting place to begin that conversation about how to create these I I forgot what you called them but these these unlined that's it.

GregB: Unlivable selectively disclose credentials we have cryptographic standards that help us do that have.

David_Karger: Exactly so so semantic scholar might be a good source of those 4 academic.

Harrison_Tang: By the way Greg uh earlier you you gave a presentation on using DBS at the general knowledge proof for students like uh can BBS and general knowledge proof be used to implement narrow merrylands basically.

GregB: Well the thing that struck me uh 1 form of our cryptographic pseudonyms not to be confused with a a facing pseudonym is the fact that you could take credentials.

GregB: From get credentials from multiple places that are Anonymous but then be able to present them that they belong to me even though for they're from different issuers so the advisor says yes I certify this person is my adise then I've got another set from say man semantic scholar yes this person published these papers.

GregB: We have separate credentials but then we can go and present them into the meronem system saying look these all belong to me there really are me and such like that and do that in an unlink but they're all me and such like that so.

GregB: I there's some interesting possibilities and.

GregB: We can do it both ways.

David_Karger: I think there are 2 ways you want to think about that uh sometimes you want to say they're all me uh but sometimes you don't write like sometimes you want to keep a post entirely unlink and you know this is a post from somebody who has this meronem this collection of credentials um but it doesn't want to be associated with any other the value of a pseudonym is the way it links multiple posts together and you can assess that they're from the same author even if you don't know who that author is so sometimes you would like a Moran and otherwise at other times you want uh Anonymous Marini.

GregB: Yep yep yep.

David_Karger: But it sounds like your infrastructure could support both.

GregB: Yeah like I just I I gotta understand it enough so that I can get it all the way down to our crypto stuff but uh great conversation we can have more.

Harrison_Tang: Right well at time we're at time uh this is a great uh presentation thank you Nora uh if you don't mind can you actually share with uh me or the public group in regards to the link to your presentation.

Nouran_Soliman: Yeah for sure I'll mail it to you.

Harrison_Tang: Cool I'll follow up with you as well so thank you um yeah thanks a lot this is a great presentation uh you know this concludes this week's ccg meeting and uh thanks to neuron thanks uh everyone uh for for joining us today.

Nouran_Soliman: Thank you it was great to be here.

The W3C Credentials Community Group

Meeting Transcriptions and Audio Recordings (2014-today)

Go Back

W3C CCG Weekly Teleconference

Transcript for 2024-07-02