Re: [VoT] How to express duplicate checks with VoT?

Eric Goodman <> Fri, 18 March 2016 18:32 UTC

Return-Path: <>
Received: from localhost (localhost []) by (Postfix) with ESMTP id 917F212D8B7 for <>; Fri, 18 Mar 2016 11:32:50 -0700 (PDT)
X-Virus-Scanned: amavisd-new at
X-Spam-Flag: NO
X-Spam-Score: -1.901
X-Spam-Status: No, score=-1.901 tagged_above=-999 required=5 tests=[BAYES_00=-1.9, DKIM_SIGNED=0.1, DKIM_VALID=-0.1, HTML_MESSAGE=0.001, RCVD_IN_DNSWL_NONE=-0.0001, SPF_HELO_PASS=-0.001, SPF_PASS=-0.001] autolearn=ham autolearn_force=no
Authentication-Results: (amavisd-new); dkim=pass (1024-bit key)
Received: from ([]) by localhost ( []) (amavisd-new, port 10024) with ESMTP id EzqHj6kmBWph for <>; Fri, 18 Mar 2016 11:32:46 -0700 (PDT)
Received: from ( [IPv6:2a01:111:f400:fc10::658]) (using TLSv1.2 with cipher ECDHE-RSA-AES256-SHA384 (256/256 bits)) (No client certificate requested) by (Postfix) with ESMTPS id 0DDFF12D780 for <>; Fri, 18 Mar 2016 11:32:45 -0700 (PDT)
DKIM-Signature: v=1; a=rsa-sha256; c=relaxed/relaxed;; s=selector1-ucop-edu; h=From:To:Date:Subject:Message-ID:Content-Type:MIME-Version; bh=vdCOG2MwtYj2GHUsuTgoReaFSl55dzIvXIyMA1WdvqY=; b=M49aZUanWKnwAhzR9zh6eMkt6Uf6SkAUVTiC523Ipc9zq0M1JcXozseORJ2QyopB27/911+h4bmQlpmss6NFeQFThPi3NkLvsP326VYWLMcZmEcPmQOgCkc7u60UtGWNec5o95sgnCTAtPJqzCN0ad/bhaPLnI06nSD/DDM+FiM=
Received: from ( by ( with Microsoft SMTP Server (TLS) id 15.1.434.16; Fri, 18 Mar 2016 18:32:25 +0000
Received: from ([]) by ([]) with mapi id 15.01.0434.021; Fri, 18 Mar 2016 18:32:25 +0000
From: Eric Goodman <>
To: "" <>
Thread-Topic: [VoT] How to express duplicate checks with VoT?
Thread-Index: AQHRgRjd7B/eYhgAUEWWABVUKcD48Z9fguaw
Date: Fri, 18 Mar 2016 18:32:25 +0000
Message-ID: <>
References: <> <> <> <> <>
In-Reply-To: <>
Accept-Language: en-US
Content-Language: en-US
authentication-results:; dkim=none (message not signed) header.d=none;; dmarc=none action=none;
x-originating-ip: []
x-ms-office365-filtering-correlation-id: ab4fadbc-8843-413c-a964-08d34f5ba732
x-microsoft-exchange-diagnostics: 1; BN3PR0601MB1954; 5:eztg7I9xgBGwAWpuadLfTG690cKk5BiaMx5+DE9KS65peQYPfmbEVvio2sUe1cBbfn4PKgiRdMVaFp2sYmv1W5GzFCobmxkVescp/BfcuKqShXq8FRSjiUxW4DR8xFSH6y+jvky7M/D/KZmlJDpCdw==; 24:Rwoy+UzazLIXF/c4uOrqV2kdwhRURGq6xKguIBuanr5fu7D74MhDSG7AKWDjEhIG6iMVNuy2fOW+dHMybsmj6izTLYEWSI0grYDWaIFB/A4=
x-microsoft-antispam: UriScan:;BCL:0;PCL:0;RULEID:;SRVR:BN3PR0601MB1954;
x-microsoft-antispam-prvs: <>
x-exchange-antispam-report-test: UriScan:;
x-exchange-antispam-report-cfa-test: BCL:0; PCL:0; RULEID:(601004)(2401047)(5005006)(8121501046)(3002001)(10201501046); SRVR:BN3PR0601MB1954; BCL:0; PCL:0; RULEID:; SRVR:BN3PR0601MB1954;
x-forefront-prvs: 088552DE73
x-forefront-antispam-report: SFV:NSPM; SFS:(10009020)(69224002)(24454002)(377454003)(13464003)(53754006)(19609705001)(81166005)(19580405001)(5002640100001)(19580395003)(92566002)(3280700002)(76576001)(5630700001)(3660700001)(75432002)(106116001)(5640700001)(76176999)(90282001)(99286002)(54356999)(2501003)(50986999)(19617315012)(16236675004)(74316001)(19300405004)(88552002)(450100001)(586003)(6116002)(102836003)(3846002)(2900100001)(790700001)(16601075003)(87936001)(5004730100002)(5003600100002)(1220700001)(107886002)(5008740100001)(122556002)(110136002)(10400500002)(1730700002)(11100500001)(1096002)(189998001)(93886004)(66066001)(2906002)(2351001)(33656002)(86362001)(106356001)(15975445007)(19625215002)(89122001)(77096005)(2950100001); DIR:OUT; SFP:1101; SCL:1; SRVR:BN3PR0601MB1954;; FPR:; SPF:None; MLV:ovrnspm; PTR:InfoNoRecords; LANG:en;
spamdiagnosticoutput: 1:23
spamdiagnosticmetadata: NSPM
Content-Type: multipart/alternative; boundary="_000_BN3PR0601MB1956E6B81FD35211464C2032FB8C0BN3PR0601MB1956_"
MIME-Version: 1.0
X-MS-Exchange-CrossTenant-originalarrivaltime: 18 Mar 2016 18:32:25.3153 (UTC)
X-MS-Exchange-CrossTenant-fromentityheader: Hosted
X-MS-Exchange-CrossTenant-id: 6737ae8b-7f79-4cda-90a2-1861f16fa830
X-MS-Exchange-Transport-CrossTenantHeadersStamped: BN3PR0601MB1954
Archived-At: <>
Subject: Re: [VoT] How to express duplicate checks with VoT?
X-Mailman-Version: 2.1.17
Precedence: list
List-Id: Vectors of Trust discussion list <>
List-Unsubscribe: <>, <>
List-Archive: <>
List-Post: <>
List-Help: <>
List-Subscribe: <>, <>
X-List-Received-Date: Fri, 18 Mar 2016 18:32:50 -0000

I would expect (though could be wrong) that the issue isn’t just imposters, but as much about enforcement of separation of duties for the same entity. If I give one of your accounts review rights and another approval rights, then there’s an SoD issue.

So the only thing I was thinking would be worth measuring within the IdP is whether the LoA registration process includes duplicate resolution. I don’t know that any assurance profiles/VoT’s I’ve seen include that as an explicit requirement. I.e., what’s to stop me from having two accounts both vetted to a strong degree and the IdP doesn’t recognize them as the same person.

But your point is compelling: Is identity matching internal to an IdP terribly irrelevant in a federated world? What value does an RP get by knowing that this identity was checked against duplication within an IdP if I can separately create an account (with an equally high LoA) through a completely different IdP? I don’t see what IdP-managed trust measure/indicator is going to help unless you limit your use cases to those involving a single IdP, in which case the need for a profile and standardization is less pressing.

--- Eric

From: vot [] On Behalf Of Julian White
Sent: Friday, March 18, 2016 6:19 AM
To: Joanne Knight
Cc: Rolf Brugger;
Subject: Re: [VoT] How to express duplicate checks with VoT?

I would say, as I have in the ISO 29003 discussions, that uniqueness is in fact irrelevant in terms of the level of proofing. I see no need to know whether that set of attributes is proven to be unique by the proofing provider. Uniqueness is highly dependent on the context, which raises all sorts of problems if you have an identity federation (how do I know its unique across the federation? If, as an RP, I can can cope with non-uniqueness across the federation then non-uniqueness within the proofing provider is also not a problem).

In my view it is for the RP, or identity federation scheme, to decide which attributes are required in their context. They select sufficient attributes that give them uniqueness (in their context) if that is required, e.g. SSN for some services (though I would advise against it!). The purpose of the proofing process is to express to what level of confidence that those attributes are true and belong to the person presenting them, be they name, address, dob, email address, SSN etc.

In my dealings with RPs they are confused about this and ask for "uniqueness", this is usually followed up with "because we can't let people register multiple identities". There is a lack of understanding that, assuming the proofing process is working correctly, it doesn't matter how many times you enrol you are proven to be the same person each time; i.e. multiple enrolments lead to the identification of the same person being bound to the same attributes. Therefore it isn't actually a problem in terms of uniqueness.

More often than not the ask for uniqueness is really an expression of how well the system defends itself against impostors (different people registering for the same attributes) which is actually the point of the different levels of proofing (the higher the levels of proofing the harder this should become). Demanding uniqueness doesn't actually fix the impostor problem, you just introduce a time trial to the process; the first person to get bound to those attributes locks out other people; if the first person to claim them was the impostor you've now locked out the genuine person - they now have a hell of time unpicking a mess not of their making. If the system defends against impostors properly at the higher levels then the uniqueness adds little value since I can only bind those attributes to the correct person anyway.

The thing that may need to be unique is the account/record at the RP, and its the RP's responsibility to manage that. This is a problem they have regardless of whether the proofing provider tells them its unique or not because they often have multiple on-line and offline channels that deal with their customers, this forces them to understand how to match different representations of the identity to the same account which tends to work across channel and within the same channel.

So in summary it doesn't drive any real value in the proofing provider to do this, its just a cost of effort for little gain that the RP is going to repeat regardless if they care about uniqueness.



On 17 March 2016 at 21:49, Joanne Knight <<>> wrote:
The short answer to your question - No.

Your use case is silent on the level of vetting required (how real does the identity have to be). You can have any level of identity and still collect a biometric for the purposes of one and only one uniqueness. I would pose however that it is very unusual for one and only one assurance to be required without some other driver requiring an elevated level of realness.

In your use case I would suggest that there is also a requirement that a qualification is issued to
a) the person carrying out the study and
b) an identity of such rigour that their (or the university's) eligibility for public funding can be established and
c) an identity that is going to be tested by future employers.

Therefore you are probably looking at P2 or P3 depending on the level/type of qualification. (I might want P3 if the qualification is for brain surgeon or commercial pilot).

As one and only one is really only able to be achieved through the use of a biometric, we now have to consider 'acceptability'.
Does the collection of a biometric (and all that it entails - cost, privacy, security, user acceptance etc.) outweigh the impact caused by a duplicate entity in the system?
In other words, do you really need one and only one (the 3rd level of uniqueness) or could sufficient singularity be gained from leveraging the social security number's uniqueness and an equivalent level of validation and ownership test (the 2nd level of uniqueness) i.e. no Pu?

We must remember the mutual exclusivity between attributes collected, vetting under taken and level of uniqueness.

Happy to discuss further if this makes no sense.


-----Original Message-----
From: Rolf Brugger [<>]
Sent: Thursday, 17 March 2016 11:35 p.m.
Subject: Re: [VoT] How to express duplicate checks with VoT?

Hi all,

Thanks for all your responses and ideas. We have discussed your inputs in our project team (which is why it took me some time to respond)

I was wondering what actually the differences are between the three levels of uniqueness described by Joanne. In each of the three levels a set of attributes is used to determine if an identity is unique within a population of identities. The difference is 1) what exact set of attributes is used in each level and 2) the vetting level of the attributes. Right?

A concrete example might be

First level:
- attributes: Name, email, phone number
- vetting level: P1

Second level:
- attributes: Name, email, phone number, social security number
- vetting level: P2

Third level:
- attributes: finger print or retina scan
- vetting level: P3

The appropriate set of attributes to be used in each level is certainly context-specific. As Eric points out it will be difficult to define and agree on a generic definition of the scope.

Maybe it's time to describe our use cases here. In the IdM service we are planning individuals self-register their identities. With such an identity, individuals can register for studies at a university. But one individual may only register once at the same university. Another case is a user's library account that was blocked because of violations of the library's usage terms. Such a user should not be able to just create a new account to use the library services again.

With Justin's suggestion of a Pu category, the requirements in our use cases could be expressed as P1.Pu or P2.Pu. The semantics of "P2.Pu"
would then be: Proofed unique based on an attribute set with the vetting level P2.



On 11.03.16 02:56, Joanne Knight wrote:
> Hi All
> There are three aspect or levels when looking at uniqueness to be
> considered.
> First level is that the identity is unique - that is, within a context
> there is a set of attributes that are unique for each identity
> registered.
> Second level is sole claimant - this is a check that only one entity
> has claimed a particular set of attributes. At the higher levels of
> identity proofing where authoritative sources are used it may be
> possible to achieve this. This is sufficient in most RPs cases. At
> this level while it is possible for a single entity to claim more than
> one identity, they do so at the risk of causing a counter-fraud flag
> should the real owner (or any other party) also attempt to claim the
> identity.
> The final level is one and only one - This is a check - usually
> biometric - that an entity has only one claim in the context. This is
> usually only reserved for the highest level of identity and would also
> require equally high levels of credential and credential issuance
> processes.
> As to how this relates to VoT - The first should be innately built
> into all levels of P - it is the sole requirement of all levels
> The second could be built into P3 if the wording was amended slightly.
> The last item only is substantively missing and to date (in the
> conversations I have been having elsewhere) there has been
> insufficient appetite to add it as an explicit requirement.
> Should we have a PU? Maybe, but steer clear of the term 'unique' If we
> do, in my mind it would only have two values - P?0 - Claims per entity
> not checked, P?1 - Claims per entity restricted to one.
> Joanne
> -----Original Message----- From: Rolf Brugger
> [<>] Sent: Friday, 11 March 2016 5:51 a.m.
> To:<> Subject: [VoT] How to express duplicate checks with
> VoT?
> Hi all,
> I'm new to this list and I hope my question is not totally irrelevant
> here.
> We have plenty of use cases where RPs need to have confidence, that a
> person does not have multiple identities in one IdP. I don't see how
> this aspect of identity quality can be expressed, and I believe it is
> pretty orthogonal to the P, C, M and A dimensions that are currently
> specified in the VoT draft.
> We could imagine multiple ways to gradually prove that an identity has
> been checked against duplicates. The most straightforward approach
> would be to make sure that unique personal attributes are used only
> once within one IdP or an IdP federation, like - email
> address(es) - mobile phone number - home postal address - social
> security number - ID / passport number - the combination of name and
> birth date - etc.
> Would it make sense to express this in VoT?
> best regards
> Rolf
> -- SWITCH -------------------------- Rolf Brugger, project Swiss
> edu-ID Werdstrasse 2, P.O. Box, 8021 Zurich, Switzerland phone +41 44
> 268 15 15, direct +41 44 268 15 89<tel:%2B41%2044%20268%2015%2089><>,
> _______________________________________________ vot mailing list

Rolf Brugger, project Swiss edu-ID
Werdstrasse 2, P.O. Box, 8021 Zurich, Switzerland phone +41 44 268 15 15<tel:%2B41%2044%20268%2015%2015>, direct +41 44 268 15 89<tel:%2B41%2044%20268%2015%2089><>,

vot mailing list<>