DeepFakes Part 3 – What Parts of the Identity Stack & Verifiable Credentials for Digital Provenance?

Verifiable Credentials for Digital Provenance? + Deepfakes Part 3: What Part of the Identity Stack?

Thursday 13G

Convener: Sarah Allen & Kathryn Harrison

Notes-taker(s): Sarah Allen

Tags for the session - technology discussed/ideas considered:

Discussion notes, key understandings, outstanding questions, observations, and, if appropriate to this discussion: action items, next steps:

Two Digital Provenance Use Cases presented:

Scenario (A)

In order to refinance a loan, there is a complex network of actors who review multiple inputs to assess the creditworthiness of the investment.

Actors

1. Homeowner- Owns the home and is submitting a request for refinancing.
2. Bank- Organization which owns the loan
3. Loan processor Pete- 3rd party organization working on behalf of bank to assess if house meets code requirements to get loan

Objects

1. Loan application
2. House photos- Pete takes photos to validate that the house meet the criteria for the loan
3. Loan evaluation- qualitative commentary and photo evidence

Flow

Homeowner-->Loan Application-->Bank--> Loan Processor Pete-->Photos taken by Pete-->Loan application

How do you know that the content in the photos is authentic?

Example-- to pass the loan criteria need a carbon monoxide detector but homeowner doesn't have one. But Pete is incented to get the loans approved and finished so he pull a detector out of his pocket, plugs it in and takes a picture.

Or he photoshops it?

Top of Form

Bottom of Form

Scenario (B)

Can we connect digital records such that the provenance of knowledge can be traced in order to streamline the research process? For a use case, we can look at a focused research questions that was answered in digital archives with multiple searches and expert research analysis (via Smithsonian Collections Blog)

Doris Cochran was a herpetologist who collected over 3,000 frog specimens from Brazil in her career. Her 1962-1963 travelogue documents Doris Cochran's National Science Foundation funded trip to visit museums in South America and collect frog specimens. In 2013, Smithsonian Institution created an online Transcription Center where volunteers could transcribe historic documents. One volunteer questioned whether the catalog meta-data was correct. The writer of the travelog mentioned a travel companion "Doris" -- was this book written by another scientist?

Working with archivists and researchers at the Smithsonian Institution Archives, we discovered the original NSF research proposal which mentions Mrs. Sydney Blake.

1. Archivist Annie -- takes papers from Doris Cohran's office, catalogs them for the physical archives and creates digital records (Finding Aid)
2. Digitizer Don -- creates digital image from the field notebook, giving it an independent digital record (ID: SIA RU007151) with reference to the collection of papers.
3. Researcher Ric -- documents NSF grant proposal with reference to Mrs. Sydney Blake
4. Researcher Ellen -- determines that Mrs. Sydney Blake is the "Doris" mentioned in the field notebook.

The group decided to focus on Scenario A because participants believe it to be simpler.

Actors

• Home owner "Wendall" (excluded to simplify scenario)
• Loan processor "Pete"
• Bank

Pete is issued a credential that he is qualified to inspect the home.

When Pete inspects the home, he uses his credential to issue a claim.

What credentials do they need to perform the role? What is the chain of events?

Verifiable Credentials issues to humans / real-world entities

1. Bank asserts that Pete is authorized to take photos for the loan documents
   • issuer: bank
   • subject: Pete
   • holder: Pete's Photography Co. (could be Pete)
   • claim:
   • Possible option: Bank provides Pete a digital camera >
2. Bank asserts that Pete is authorized to take photos for the loan documents
   • issuer: bank
   • subject: Pete
   • holder: Pete's Photography Co. (could be Pete)
   • claim:

Possible option: Bank provides Pete a digital camera >

Transactions

1. Pete takes photo and generates a report. (This report is a verifiable credential.)
   1. issuer: Pete
   2. subject: <house or photo (could be two subjects)>
   3. holder: Pete's Photography Co. (could be Pete)
   4. claim:
2. Pete provides the report to the bank. The bank can verify the report (VC) to ensure that it was created by the entity that was authorized by the bank.

Open questions

• ontology for the issuer/subject/holder fields need to be determined
• what happens in real world / real time? what are the digital parallels?
• what are the implications of this? does it reduce / increase trust in informal systems?

https://github.com/ultrasaurus/digital-provenance