I just want to add that I am excited to see this conversation about different 
roots of trust. We live in a world where both trust models exist and will 
continue to grow for their respective reasons. We need strong centralized trust 
for government and corporate identity. We also need distributed trust to 
support ad-hoc communities. Both have software supply chains which need SBOM 
support, SBOMs that should be digitally signed.

For Windows components we are relying on COSE signed <product>.spdx.json SBOMs 
built at each step in our supply chain, the signature gets verified by each 
consumer in the pipeline. In our ecosystem the strong root of trust is 
important. However, if we needed an open source component, the one signed by a 
distributed authority that we trust would be far stronger than a package that 
was not protected by a signed SBOM.

Just my $0.02.
--
Joe Bussell
Windows Engineering System | Tool Benders

🙋‍♂️ My pronouns are he/him<https://pronoun.is/he> (why this 
matters<https://pronouns.org/what-and-why/>)
🗓️ Book a meeting with 
me<https://outlook.office.com/bookwithme/user/[email protected]/meetingtype/23WCp6gy8UKwSRsqIkcCmg2?anonymous>
👍 Inviting Feedback<https://forms.office.com/r/FxAwaVPjjb>
⌚ Timezone: (GMT-8) US Pacific

N.B.: I may send mail during times when others are not working. I do not expect 
engagement from you when you are not working.

From: [email protected] <[email protected]> On Behalf Of David 
Kemp via lists.spdx.org
Sent: Monday, February 6, 2023 10:24 AM
To: SPDX-list <[email protected]>
Subject: [EXTERNAL] [spdx-tech] Identities

Gary,

Yes.

The W3C Distributed Identifiers 
Architecture<https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Fwww.w3.org%2FTR%2Fdid-core%2F%23architecture-overview&data=05%7C01%7Cjoe.bussell%40microsoft.com%7C5ddc4e6d9c7b4eb988d608db086f5344%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638113046472799521%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=jiW8Q9UniF298ttqrjrNInnSmkYqpClJ2rZD1pZAPx4%3D&reserved=0>
 has an identity manager (DID controller) that assigns distributed identifiers 
(DIDs) to DID subjects.  Presumably it uses the word "controller" instead of 
"authority" to contrast with X.509 certificate authorities, but the meaningful 
distinction is between centralized vs. decentralized identity management, not a 
difference in meaning between authority vs controller vs manager vs. assigner.

Other points of interest:
* A Decentralized Identifier, or DID, is a URI.  -- The only thing that 
distinguishes any string or URI from an identifier is the fact that an 
identifier corresponds to an identity (DID document) managed by an 
authority/controller/manager/assigner.  A URI that looks like a DID is not a 
DID if it has no corresponding identity, just as a 9 digit number is not an SSN 
unless it is backed by an identity.
* The subject of a DID is, by definition, the entity identified by the DID. The 
DID subject might also be the DID controller.
* The controller of a DID is the entity (person, organization, or autonomous 
software) that has the capability to make changes to a DID document.  --  DID 
recognizes autonomous software (SoftwareAgent) as a peer of person and 
organization.

In DID, entity means a real thing (person, group, organization, thing, or 
concept) to which an identifier is assigned.
SPDX models real things like persons, packages, and relationships using data 
structures (Elements).  If "Entity" is used as an SPDX type, then it must model 
all of the real-world entities covered by SPDX, not just those related to 
identities.  I.e., "Entity" would be the name of SPDX Element, not a subclass 
of Element.  But it is easier to understand and keep straight "Element models 
an entity" than "the Entity class models a real-world entity".

Regards,
David

On Wed, Jan 11, 2023 at 9:48 PM Gary O'Neall 
<[email protected]<mailto:[email protected]>> wrote:
I had a similar though.  I was wondering if the definitions provided would 
support a BlockChain like approach which does not have a centralized 
“authority”.

Gary

From: [email protected]<mailto:[email protected]> 
<[email protected]<mailto:[email protected]>> On Behalf Of 
William Bartholomew (CELA) via 
lists.spdx.org<https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Flists.spdx.org%2F&data=05%7C01%7Cjoe.bussell%40microsoft.com%7C5ddc4e6d9c7b4eb988d608db086f5344%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638113046472799521%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=XL7hHeom8f20I%2FOMEhjLrL746zKM7LCaosJx2jUoQkQ%3D&reserved=0>
Sent: Wednesday, January 11, 2023 4:14 PM
To: SPDX-list <[email protected]<mailto:[email protected]>>; 
[email protected]<mailto:[email protected]>
Subject: Re: [spdx-tech] Identities

These all seem reasonable to me. My only comment is that there may not be a 
"formal" authority. For example, an identification scheme could use an 
algorithm to derive a globally unique identifier or use a convention to 
guarantee sufficient uniqueness. An authority may or may not associate an 
identifier with an identity.


Regards,

William Bartholomew (he/him) – Let’s 
chat<https://nam06.safelinks.protection.outlook.com/?url=https%3A%2F%2Faka.ms%2Fbook-willbar&data=05%7C01%7Cjoe.bussell%40microsoft.com%7C5ddc4e6d9c7b4eb988d608db086f5344%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638113046472799521%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=SUE%2BjZXl5kK%2FsHJ%2FeVqf%2BAizvq%2Bc47o9SBaefn2fXLc%3D&reserved=0>

Principal Security Strategist

Global Cybersecurity Policy – Microsoft



My working day may not be your working day. Please don’t feel obliged to reply 
to this e-mail outside of your normal working hours.

________________________________
From: [email protected]<mailto:[email protected]> 
<[email protected]<mailto:[email protected]>> on behalf of David 
Kemp via 
lists.spdx.org<https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Flists.spdx.org%2F&data=05%7C01%7Cjoe.bussell%40microsoft.com%7C5ddc4e6d9c7b4eb988d608db086f5344%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638113046472799521%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=XL7hHeom8f20I%2FOMEhjLrL746zKM7LCaosJx2jUoQkQ%3D&reserved=0>
 <[email protected]<mailto:[email protected]>>
Sent: Wednesday, January 11, 2023 2:59 PM
To: SPDX-list <[email protected]<mailto:[email protected]>>
Subject: [EXTERNAL] [spdx-tech] Identities

At the tech meeting we decided to accept the current identity model and move 
forward without blocking the 3.0 release.  The discussion covered many ideas on 
which no decisions were documented, and I wonder if we can reach agreement on 
these points while the discussion is still fresh, without allowing any No 
Decisions to become blockers.

1) An Identifier is different from an Identity.

Discussion: Identifiers have the property of being associated with zero, one, 
or multiple identities over time.  Note: at any specific time an identifier 
should be associated with at most one identity.

2) Every Identity MUST have an authority.

Discussion: The authority associates identifiers with identities. If there is 
no authority, there can be no identity to which an identifier refers.
* The Social Security Administration is the authority that maintains records of 
peoples' identities. Every 9 digit number is an identifier, but only some of 
those identifiers are associated with an identity: 000-00-0000 and 123-45-6789 
are "SSN identifiers" but they (probably) have never been assigned to an 
identity by the authority.
* 
"hotmail.com<https://nam06.safelinks.protection.outlook.com/?url=http%3A%2F%2Fhotmail.com%2F&data=05%7C01%7Cjoe.bussell%40microsoft.com%7C5ddc4e6d9c7b4eb988d608db086f5344%7C72f988bf86f141af91ab2d7cd011db47%7C1%7C0%7C638113046472799521%7CUnknown%7CTWFpbGZsb3d8eyJWIjoiMC4wLjAwMDAiLCJQIjoiV2luMzIiLCJBTiI6Ik1haWwiLCJXVCI6Mn0%3D%7C3000%7C%7C%7C&sdata=Pgz6BXFWML%2FueedMdxSNK5hxvKXlR0SuPx0YXr1f3RU%3D&reserved=0>"
 is the authority that maintains hotmail identities.  The identifier 
"[email protected]<mailto:[email protected]>" is (probably) not an identity 
because of minimum length restrictions on the local portion.  The authority 
assigns identifiers to identities, ensuring uniqueness.  The identifier  
"[email protected]<mailto:[email protected]>" has probably been 
assigned to several identities over time. The authority determines if it is 
currently assigned to any identity.
* Without assistance from the authority it is impossible for SPDX to 
distinguish the identities to which an identifier is assigned.  If 
"[email protected]<mailto:[email protected]>" is an active identity 
in 2021 and 2022, it is impossible to know if they are the same identity or two 
different identities unless some other information (such as SSN or a 
hypothetical hotmail UID) is included in those identities. SpdxId is not part 
of the identity - many Identity Elements can be created for the same identity.

3) Authorities determine what subject types they support

Discussion: SSA will not assign identities to anyone other than natural people 
- it is fraud to attempt to create fake accounts.  Hotmail doesn't do any 
identity proofing - anyone or anything can get a hotmail account on request, so 
the distinction between person and organization doesn't exist for that 
authority. Squatters have claimed many obvious hotmail organization 
identifiers, but at the moment 
"[email protected]<mailto:[email protected]>" is available.

4) Some authorities create identities and assign identifiers to processes

Discussion: A process identity type is not a PID running on an operating 
system, it is a subject type accepted by an identity management authority.  
Hotmail has already created 
"[email protected]<mailto:[email protected]>" and 
"[email protected]<mailto:[email protected]>" identities, and 
"[email protected]<mailto:[email protected]>" is currently available to 
be claimed.  As above, hotmail does not do any identity proofing or declaration 
of identity types.  But the U.S. Government does explicitly manage non-person 
entity identities for corporations, devices and processes in addition to person 
identities.

It is neither esoteric nor difficult to accommodate process identities in the 
logical model; the same standard of acceptance should apply to 
Principals/Actors/Agents that are processes as to those that are persons or 
organizations.

Dave



-=-=-=-=-=-=-=-=-=-=-=-
Links: You receive all messages sent to this group.
View/Reply Online (#4953): https://lists.spdx.org/g/Spdx-tech/message/4953
Mute This Topic: https://lists.spdx.org/mt/96211555/21656
Group Owner: [email protected]
Unsubscribe: https://lists.spdx.org/g/Spdx-tech/unsub [[email protected]]
-=-=-=-=-=-=-=-=-=-=-=-


Reply via email to