HXWG last week - review of harms model

Drummond Reed commented this paper much needed

Nicky Hickman to x-check w/drummond key questions/ issues & drivers for this paper

Judith Fleenor asked what we were working on

• Phil Wolff Whitepaper + blog post - aim to look at what SSI does for systemic harms associated e.g. w/over vs under identification and aim is to provide tools for the community to prevent harm and consider in their work
• Focus on product, management and engineers who want to prevent harm ie for the ToIP community
• Not general public
• Judith Fleenor reminded us that there is the generic

HXWG update on Weds 20th - Phil Wolff Phil to do Harms TF slide

• sankarshan commented that there is an IIW special topic meeting Tuesday 9th August "Digital Identity Across Asia" Pyrou & sankarshan to do session on Harms

HXWG update at all-hands; Question regarding terms that we understand and assume sankarshan to add link to document

Expert Series - review and comment

Task Force BGBU / SSI Harms

USA / EU / Africa  Every Alternate Monday

9:00 Pacific / 12:00 Eastern / 16:00 UTC

Next call is Monday 1st August, then a break until Monday 12th September

Asia & Australasia Every Alternate Thursday

18:00 AEDT / 13:00 Indochina / 11.30 IST / 6:00 UTC

Next call is Thursday 4th August, then a break until Thursday 16th September

Harms White Paper Draft  Phil Wolff 

• use of 4/4 matrix not useful - just use headlines like ' Digital ID comes with known harms'  - remove 'known-knowns'
• See points in Philip Sheldrake https://www.linkedin.com/pulse/digital-identity-human-matters-philip-sheldrake/
  • Pyrou commented that in region WB pushing digital identity ID4D, developmental drive in region to use digital identity as a development aid - will 'solve poverty issues' - loose argument - no strong alternatives to WB model.  SSI can be a strong alternative to centralized models, but no strong activists influencing policy.
  • sankarshan many digital identifiers are framed as 'public goods' but don't have a sustainable economic model, problem creating s/w marketplaces without durability.  Most digital ID schemes are state-funded, reinforces sustainability, also in India based on facial recognition & biometrics; problem of changing embedded systems, processes and AI which reinforce these structures of power.  Worrying - the entire premise has not been challenged, no strong and viable alternative.  No consultative process.
  • Pyrou - no ability to say 'no' (Participation Principle is not observed); - embedding digital ID as a social norm, there are still Indian communities - tribal group recognised as 1st People, see commons in forest vs property - government appropriates the forests to access sustainability funding - digital recording of the - transfer of capital = transfer of power.  
• Qualitative differences between SSI & other ID tech?
• No calibration or measurement of harm from implementation of SSI
• Our definition of harm might be different from other definitions of harm e.g. variance from self-sovereignty
• Are here unintended consequences of doing things as we normally do them for IAM e.g. WEIRD models of identity in west vs other parts of the world
• risks of cultural imperialism through tech

Judith Fleenor commented that we needed to ensure that the white paper that enabled members to 'get stuff done', other ideas about the socio-cultural issues are a 'thing to think about'

Technology is not agnostic, it is political.  

Because of corruption - central authorities / governments are not 

Man is moral and SSI is a kind of moral technology - we are not neutral the status quo is insufficient - this is a better way. 

sankarshan Still need the basis even if this is socratic then build on the steps, still needs to be rights-based and addresses concerns

Dangers of 'othering the problem' - being more direct 

comms of the paper would be a series of blog posts a) harms embedded and scaling of harms; b) concept of legal identity vs digital identity c) other ways of using SSI to express human identity

Phil Wolff why should I care; b) characterising the harms; c) theory of how to frame harms d) turning concern to action, e) budget e.g. for harms assessment 

Judith Fleenor considered edge communities can create new harms to current power base (ie those that are currently benefiting) - zero sum game

There is now a terms wiki and we should use a # for SSI Harms  =  #harmtf  (see conventions)

Here is the document to start working with for ingestion Working Document HXWG Terms

Hannes Hahkio - based in Finland work for small consulting company called Hi Gear, Public sector companies are a focus of ours especially Citizen ID & SSI.  Act as consultants for gov & public sector.  Promoting SSI.  Aiming to understand landscape and then join some TF's and WGs.

Thomas Robin, works with Sezoo, based in Australia, specialising in digital trust & SSI models.  Where are the weaknesses to ensure we don't repeat the mistakes.

Update from sankarshan “Open Loop is a global program that connects policymakers and technology companies to help develop effective and evidence-based policies around AI and other emerging technologies.”

https://openloop.org/wp-content/uploads/2021/01/AI_Impact_Assessment_A_Policy_Prototyping_Experiment.pdf

Consider the ID of things (algorithms) and new TF in ToIP related to AI.

HXWG Meeting updates

sankarshan  Sovrin Ecosystem governance Framework https://sovrin.org/, now 2 governance framework with Layer 1 utility GF and L4 ecosystem GF.  The problem of ecosystems, lack of balance between big and small players.  Fairness, transparency and accountability plus collective ownership - require balance of these forces.  Prevents innovation and creates collusion, commercially unbalanced solutions and inappropriate evolution.  Sovrin is a lived reality.  

We would like to request you to go through the SEGF available from https://sovrin.org/library/sovrin-governance-framework/ (see section "Proposed Revisions/new documents") Additionally, there is a document which contains a list of suggested additional reading - we think that this compilation would be helpful in understanding the context

Also see this white paper from EFWG Trinh Nguyen

Feedback from grassroots research ref identity, personal data and harms

Major regional meeting last week with 10 countries across Asia with reps from 15 indigenous groups, main progress was to look at a framework for 

What is identity to you as an indigenous person

Ability to identify themselves with land & territory which was consistent across all groups, doesn't begin to define how they interact 

Identity linked to environmental ecosystem, e.g. different languages inside forest and outside forest, how they defined their connection to territory that were linked to landscapes that were precious to them many under threat, deep spiritual connection, so when you remove natural resources they loose their community and their ability to communicate.

Difficulty in interacting with society at large mainly because of lack of recognition - main barrier is driven by western centric way identity is laid out.   they are defined by those that colonised them, not applicable in Asia as colonised by other non-western nations, so distinct hard lines that don't necessarily reflect indigenous people and identity.  Their identity doesn't reflect their connection with nation, negotiating access to services e.g. if transboundary territories, e.g. China / India - how do they negotiate across these boundaries.  Even if they are able to self-assert, how is that recognized across those national boundaries.  EG in vietnam it is illegal to identify as indigenous.   Discrimination is then based on the assertion of indigenous - especially if mixed race children of ethnic mother, no option for dual nationality so children are appropriated to national identity and lose their culture and ancestral knowledge.  How do you prevent these harms if you can identify.

EG caste system in India, 6 ethnic groups from India, if indigenous not part of caste system are below / beyond untouchable.  They are not even recognized as a person.

HOW TO!?  Self-determination, no discrimination, being recognized, traverse admin national boundaries and engage w/society at large. 

Cambodian communities most fragmented, colonised by Khemir and gov has systematically tried to expunge indigenous identity & culture, Exceptionally fractured and deteriorated by lose generational knowledge, and loose language and culture.

Jo Spencer - ecosystem must be well defined scope & common objectives - having defined and being able to self-determine, link between national & cultural definition ask IS THIS USEFUL.  With SSI this is possible, but whether it is worth doing is the question.

Pyrou - in Cambodia high % of people have smart phones but not indigenous people, don't even have power & connectivity despite progress on SDGs and resilience in face of covid - tech itself is alien to them.  How do you address this & scale?

Hannes - Differences with meaning and words even if same language - same in Europe, different slang different language.  How do you convey meaning.  All these elements need to be represented to convey meaning, that way everyone can ensure they convey meaning .  Tied deeply to HX, how do you represent yourself and understand eachother, how do you resolve disputes?

Pyrou - we struggled with different languages - much gets lost in translation.  I am hopefully we can decolonise the tech ?  Many elders don't necessarily want tech in their community, so another barrier especially thinking about life and spirit, computers don't have that.

There is now a terms wiki and we should use a # for SSI Harms  =  #harmtf  (see conventions)

Dr Karen Elliot (Newcastle University) has offered to 'academify' the paper and contribute with some of her RA's once we have shell / draft outline with abstract / exec summ

Drummond Reed  highlighted a paper on decentralized identity systems - anti SSI - as supporting gov & tech systems - should be carefully scrutinized and shut down.  The paper is not unusual and made me think of this TF - will be a big advantage if ToIP is ahead of these kinds of key questions. Important paper will be supported from Avast. 

Christine Martin agree - we have problem with conspiracy theories - being politicized e.g. in Ontario.

for discussion see this Miro Board

NEW (Google Slides): https://docs.google.com/presentation/d/1Y404nJpSOkJFK5pc2aYUmJtrXtCkwcx-eIMC9ZMC0DU/edit?usp=sharing

Phil has transferred to a G-DOC https://docs.google.com/document/d/151cqN0HY-ECmGwcS_SSBCeCHtszuwGyQebLRJq8sODA/edit?usp=sharing which contains the same material but in a doc format

Johari window = personal psychology version of Rumsfeld  https://www.communicationtheory.org/the-johari-window-model/

Kaliya Young Problem w/AI is it makes the blind spot big and run by computers - 

Darrell O'Donnell Identification & ID Systems

Nicky Hickman - differentiate between identification & identity - problem of exacerbation

Kaliya Young DISCO - fun SSI party approach - build your own avatar in the metaverse - new market niches in SSI - we need more diversity like this in SSI than just Gov use cases 

Participatory design approach - use of on the ground communities to prioritise our work and give us case studies / persona's, will help us with prioritization and be more practical for implementers.

Many implementations are social networks backed by identity implementations - could have same problems as other social networks - always malicious actors - so how do we strengthen n/w's already to reduce risks of vulnerability associated with bad actors

Eric Welton- discussions with Myanmar Responsible Business Coalition - tricky situation with hostile relationships in government - establishing a biometric identity - difficult choice as someone is going to do it.  Can we do it in a more responsible, less harmful way.  May be useful to have a session on this in this group.  Difficult ethical questions for the team.  Interesting presentation on how digital ID can be abused.

sankarshan- biometric topic seems to be settling down from aid organisations - have to use biometrics, main focus is now on mitigating harms.  Myanmar, Afghanistan and other examples.  Uganda for example, optimistic programme, under-enrollment.  (the harms of untrustworthy systems) - now starting w/genetic profiling!

"The genie is out of the bottle"

Jo Spencer Activity on NSW gov initiative - driving license initiative hacked, DNA & biometrics further reinforce the argument. 

• We need to be better, we need to be very aware that the information we're sharing is not as secure as we'd hoped.  Law of Minimal Disclosure for Constrained Use.  Need to be cryptographically secured.  Driving new projects.  Victor Dominello  (NSW gov)  Inclusion focus on services for citizens, also allied with indigenous people programme.  Now an important process.  All exclusion factors considered (digital, literacy, social, political, disability, financial)

sankarshanMinimisation is not suitable for some interactions, e.g. Medical Records - much more important for secure and trustworthy data payload /  exchange.  Also Financial Transactions not very well suited to VC's.

Jo Spencer use of verifiable presentations which present derived data.

sankarshan taxonomy & semantics - quality & classification of data also needed

VC's are not the solution for many forms of data.  

Biggest harm is derived data sets from small data sets.

Eric Welton  - Use cases - all non-cash transactions will be exposed to government, e.g. national security / tax collection is given as purpose coming in ??.  Remember problems of correlation highlighted by Daniel Hardman

Also use case of VCs to convey health information - presentation that key emergency information for medical purposes (e.g. in accident) - could use biometrics to unlock the data for emergency responders.  VC is a PDF - convenience / emergency service.  Similar to ICE contact, could have ICE Credential. Could be linked to IATA - Good Health Pass

sankarshan Must be freeform data.  Accept that this is a new attack surface.  

John Phillips Humanitech conference - could be some good material & input to this discussion.  Great discussion and material. I'm now seeing an obvious connection between this work, and the work of the Humanitech organisation here in Australia (founded by the Australian Red Cross) who have been thinking about how they might "ensure frontier technologies benefit people and society". I was at, and spoke briefly, at their 2022 conference (https://humanitechsummit.org/2022/) - they've been considering the potential harms of the mis-application of frontier technology for some time.

Review of notes and insights from the AGENCY talk at HXWG last week.  Here is the wiki page 

On fake news (and news) there's a good 20 mins conversation from Davos hosted by Polkadot -

Rumsfeld Structure to scope  "Reports that say that something hasn't happened are always interesting to me, because as we know, there are known knowns; there are things we know we know. We also know there are known unknowns; that is to say we know there are some things we do not know. But there are also unknown unknowns—the ones we don't know we don't know. And if one looks throughout the history of our country and other free countries, it is the latter category that tends to be the difficult ones.^{[1]" (Source: Wiki)}

Also for discussion see this Miro Board

• differentiate between data exchange and identity as legal identification
• See comments in Miro Board
• Agree importance of Unknown Unknowns and use of resilience / vulnerability context relationships

Continue on storyboard roundup from Slide 31. Sections: "Indirect Risk", "Approaches to responding to the risk of human harm", "Whose job is it?", "Thank you. What’s next?".

NEW (Google Slides): https://docs.google.com/presentation/d/1Y404nJpSOkJFK5pc2aYUmJtrXtCkwcx-eIMC9ZMC0DU/edit?usp=sharing

OLD (pptx format):  https://docs.google.com/presentation/d/1KoWjJx8LMwqNHKhAs-gK1uLdDL6zniIR/edit?usp=sharing&ouid=102748924597224658467&rtpof=true&sd=true

• Phil: Moved up the section on Aisha's story to after the Intro and before "our motivation." Make it human and personal to start. Expanded Aisha's story. Tied BGBU harms to Aisha. 
• Discussed indirect risks / harms
  • Ref https://github.com/hyperledger/aries-rfcs/blob/main/concepts/0207-credential-fraud-threat-model/README.md

N/A HXWG - Expert sessions forthcoming in May (link to HXWG webinars)

Good podcast ref exclusion in Aadhar - Christine Martinto get Reference - 

Potential Women in Identity ref their exclusion research as expert for Q&A

https://docs.google.com/presentation/d/1KoWjJx8LMwqNHKhAs-gK1uLdDL6zniIR/edit?usp=sharing&ouid=102748924597224658467&rtpof=true&sd=true

• Reviewed to Slide 31`- notes in slides

Discussion

• Balance between individual rights or objectives vs collective rights e.g. covid vaccine example 
• 2 stories: - 1 related to electric vehicle standards / data in India, 1 related to surveillance at Grand Mosque (company using AI realising the world of Minority Report) - being proactive (predictive analytics)
• Covid has highlighted this argument - is there a special case in the pandemic vs 'normal' times.
• See https://www.technologyreview.com/supertopic/ai-colonialism-supertopic is a good series recommended by sankarshan
  
• Consider 'medical vector' as a means of asserting power 
• AI more deployed into affluent areas, if you don't fit in then you are suspect
• Facial recognition - Chinese dataset doesn't really work outside Asia.  (see Biometric State: The Global Politics of Identification and Surveillance in South Africa, 1850 to the Present” <-- highly recommended by sankarshan)
  
• “When Biometrics Fail: Gender, Race, and the Technology of Identity” recommended by sankarshan
  
• Aadhar reports that highlight the unintended consequences and outcomes that have arised.  Using Aadhar-like systems are more harmful than helpful because starting from a bad base.  
• No playbook that identifies the problems associated with legal identification systems. In a normal world systems are hardened through testing, but national systems are often on a wing and a prayer.  focus too much on MVP and think about things e.g. guardianship, different name formats, different genders other than M & F.

Problem w/SSI - in an effort to pursue

Updates

• Discussion re moving meeting to 11am PT = 14.00 ET = 18.00 UTC = 19.00 BST = 20.00 CEST - put in slack question to group to move.
• Links & References:  Please use this Responsible Tech Resources page to capture links to media or online resources that can help us with our work. Responsible Tech Resources  All are updated now
• SSI Harms now changed on wikis, slack & calendar invites etc to BGBU
• Our First Deliverable Storyboard, is open for business here.  Aim is to have key questions/ discussion points ready for IIW by end of month - no group members attending so use another mechanism to challenge our thinking later on in our process.
• Sankarshan working on voting and democracy work, Nicky attended APPG on blockchain & voting, and some interesting perspectives from Robert Herian on the dangers of blockchain & voting voting https://www.researchgate.net/publication/339353856_Blockchain_GDPR_and_fantasies_of_data_sovereignty
• Danger of using global south as a set of experiments - urgency and pragmatic approaches vs north america, problem not a strong representation from Latam.  Similar to clinical trials - now happening with financial services and digital products, identity, identifiers etc  These are specific harms.  EG health privatisation e.g. HealthID.  We can't look at SSI on its own, must be in connection with the businesses and services it enables.  Decentralization also leads to other things if left unchecked.  All 'sold' as discovery - but in effect it's service providers discovering you.  
• Consider 3 pillars of inclusion = connection, identity, value exchange
• Must at least ensure that folks are not worse off as a result of these developments.  Problem of data silo's even within decentralized systems.  Always going to have one player that gets more and more data and has undue influence within the market.  Data breeds data.
• Virtual IIW - let's make a proposal to Phil, Doc & Kaliya
• Indigenous group, example - document the concerns, position paper (problem definition), capture the needs at the margins.  Good example of common problems but highlighted because they are at the edge/periphery. Document real-world problems and impacts.
• problems of colonial past, current harms of centralized state systems.
• Structural biases that are embedded in our own frames of reference, e.g. UN.  eg identity & violence book.  Also aboriginal man who Stan Grant.  https://www.booktopia.com.au/on-identity-stan-grant/book/9780733644238.html
  Even naming is deeply politicised, internet meme 40 different ways that coders get names wrong.
• HARM!  not being able to use your own name - = identifier, can have many names - in West is a unique and fixed identifier.
• Anglo-Indian names have ' in them - Aadhar can't accommodate names with apostrophe - but PAN does, so Aaadhar overwrites the name in PAN then creates authentication problems.  'why should I have to change my name if the states' ID system is poorly coded'
• How can vulnerable people be cared for online?  
• CO-CREATION - Group needs representation from those we seek to support / help / protecting.  We understood representation from 'real people'
• Need to be engaged now - up front to own the process.  We are promoting a model of digital interaction which we think is a good thing, but we need to understand the existing digital access/ interaction and then understand from them as to how VC's - pre-mortem / pre-parade, if this were successful what would happen, if things went wrong what could cause that?
• Groups definition
  • understand challenges in working digitally with specific groups in the community
  • use exclusion factors?
  • power imbalances & poverty of human dignity at heart of all harms
  • Start with recognition
  • Goldilocks not too much identification, not too little, just enough for the purpose (Kim's 2nd Law) - Harm with issuance, and then when you have to use it.
• Evolving framework / discussion as with guardianship

• ACTION - Darrell to clone storyline format (from Drummond)
• ACTION - ALL to review background links (on Purpose page under Links & Files)
• ACTION - Persona Development (lapsed)
• ACTION - reach out to others to join or be an 'expert witness' - open for follow up after we have more clarity on scope and basic storyboard. 
• John Phillips to ask https://sigchi.org/ if they have any kind of framework for assessing interaction harms
  
• Nicky Hickman to get kinship structures / forms from anthropology notes https://docs.google.com/document/d/1WQCD9floOebL2716PHZBZ8a-JvMdwb-rdx02hkQdfns/edit 
• Eric Weltonto discuss domicile law and issues related to documentation and also on biometrics work next time.  Bound carefully with exam question.  Close during that session.

Scope & clarification from USA Call

• Suggested from USA call - that we de-scope to only focus on the new harms that could arise from use of SSI - do we agree?
  • e.g. Over-identification
  • How do we de-scope without removing before / after.  IE what's the cut-off function?
  • Specifically addressing the deployers of technology in ensuring that their deployments do not unwittingly increase threat surface/vectors to users (holders) causing harms.  The deliverable should describe the harms, how they arise, how they can be mitigated and recourse for victims, e.g. responses could include pooling risk or sharing harms, detect, intervene & recover. - Incidence response approach.  EG in detection how can you tell if things are going wrong, so you might monitor or surveille to ensure that harms not involving malicious actors for example are alerted and acted upon.
  • Support from USA group to use persona for storytelling, but interview real people on the ground to carry out research and understand harms.
• Recommendation from USA group that we do not exclusively focus on marginalized or vulnerable people, as we are all vulnerable at different times of our lives (e.g. as children, as frail elderly), however using extreme users or edge cases as persona can support robust design

Root Cause Analysis?

• Is part of the problem that many think SSI is specifically designed to address harms of Web 1.0 & 2.0?
• Is part of the problem that SSI is designed with a specific Western Educated Industrialised R Democratic perspective on the nature of human identity, the format of families, marriage, kinship and naming conventions?  An ego-centric (individual)  rather than socio-centric (dividual) view of identity.?  
• Must go deep to challenge the underpinnings of SSI - the problem of digital imperialism
  • One core under-pinning value is personal agency - this is based on principles of 'free will', this in turn is tied up with our ideas of free speech, free trade and is baked into our belief systems.  This is why it is so challenging.

Possible Frameworks:

• Creates or uses existing framework for analysing the dangers, risks, potential harms and threats (DRiPHT) to people with a particular focus on excluded, marginalized or vulnerable people. 

1. RIGHTS-BASED sankarshan's Digital Identifiers & Rights
2. HARMS -BASED Me2B Alliance Digital Harms Dictionary.  Uses DJ Solove's Privacy Taxonomy or Koops et al Typology to classify digital harms
   1. Action: Nicky to ask Lisa if we can have Digital Harms Dictionary in usable form to evaluate.
3.  ROLES - BASED: What are the unintended consequences of SSI or digital ID benefits on our chosen persona:  Children, Refugee, Indigenous People:   e.g. McKinsey.