[WG-UMA] Definition of "Trust" (borrowing from TCG)

Mark Lizar mark at smartspecies.com
Sun Mar 6 19:09:32 EST 2011

On 6 Mar 2011, at 14:44, Eve Maler wrote:

> Hi Mark-- Your aspirational high-level sound bite is very helpful.   
> If I understand correctly, you're saying that *all* layers  
> (technical, social, and business) involve "reliance" and, done  
> right, produce "confidence", and the latter can rise to the level of  
> "trust" if there are enough mechanisms (technical, operational, and  
> legal?) in place to make "trustworthiness" transparent (and  
> enforceable in some fashion?).

on the button, thanks for translating. :-)

> Given this, we could maybe tweak our wording choices around the  
> edges; I defer to the group on that but suggest that we not spin our  
> wheels on it. Substantively, it seems like we still have two  
> important top-level tasks:
> - Cleanly separate "reliance on parties" and "reliance on tools"
>  . Which involves close review of the protocol messaging boundaries  
> for all these
> - Review and clean up the verbs describing the nature of the reliance
> Once this is done, I think it will be possible for us to push  
> forward on analyzing and adding to the constellations as necessary,  
> and to map notions of LOA, LOP, and LOC to various ones.

Finally the fun stuff.. !!

> 	Eve
> On 4 Mar 2011, at 10:51 AM, Mark Lizar wrote:
>> My Thoughts.
>> I do think it is important to get distinct verbs that are salient  
>> across the three distinctions and I understand the language of  
>> trust as I have worked previously on a community interest company  
>> called Identity Trust for 6 years.  As you can imagine the  
>> experience of trying to define what this organisation was about  
>> took 5 years of effort.
>> From this experience I understand that If we want the language to  
>> coincide with other efforts in trust terms, we should use the  
>> technical vernacular of confidence, control and reliance to replace  
>> the term of technical trust.    But, rather than get into that now  
>> at this point, I very much like Tom's approach.
>> Lets first concentrate on the relationship and needed messages and  
>> not get hung up on language at this point. Even though I recommend  
>> not using the word trust at the micro technical level - eg. TR1  
>> etc. and to use trust in reference to concepts that are more  
>> macro.  (e.g. a trusted system of technology)  I dont  think its  
>> important to define these terms at this stage.
>> For now why not continue to examine the relationships (as Rainer  
>> and Jeff have been doing) and messages then decide on what terms  
>> actually reflect the messaging and relationships UMA is talking  
>> about?  e.g. Reliance relationship 1, (RR1)  I think we all  
>> understand that we are talking about a trust framework so that we  
>> don't necessarily need to make a lot of terminology decision at  
>> this point or over use the word trust when other options are  
>> available.
>> If we had to address the terms now I would hope that when the  
>> effort builds it would end up sounding like this from a high level:
>> "To address the needs of a trust framework UMA defines levels of  
>> reliance in the  relationships of the AM,Host and Authorising User  
>> to produce messaging that developes confidence that can be trusted  
>> and used in transparently  trustworthy manner.
>> Trusted in that it address liability by indemnifying all parties  
>> and trustworthy through appropriate use of messaging that people  
>> can use to build trust.  In this way UMA work towards a solution  
>> that evolves appropriate control and administration of consent  
>> (over time) for the appropriate parties when the technology is  
>> properly deployed.
>> The meta-model effort (in my opinion) can then concentrate on  
>> addressing global trustiness that is usable to be trustworhty.
>> Does this help align all of our thinking?
>> - Mark
>> On 4 Mar 2011, at 16:34, Eve Maler wrote:
>>> This exchange has been chock-full of wisdom so far.  Obviously we  
>>> can't solve all of the philosophical questions about trust that  
>>> remain (and it seems like many questions have been answered in  
>>> decades of literature). Also, different communities have been  
>>> reusing the same word for different purposes. But I wonder if we  
>>> can agree on a limited set of terms and concepts that allow us to  
>>> push our agenda forward.
>>> We've already divided the UMA players into tools vs. parties,  
>>> which suggests a technical/legal split, something that's come up  
>>> in the broader trust framework efforts as well. The technical/ 
>>> social/legal trust distinction we discussed yesterday seems  
>>> promising. Here are those notes again:
>>> 1. Technical trust (APIs/"contracts" between software or hardware  
>>> components – this approaches "compliance" to specs or profiles)
>>> 2. Social trust (expectations about responsibilities)
>>> 3. Business trust (a model of liability/indemnity that applies to  
>>> parties whose responsibility rises to the level of "duty")
>>> Would it be outrageous to consider distinct verbs for each, like  
>>> this?
>>> 1. "Technical trust" = Tool A "relies on" tool B for X
>>> 2. "Social trust" = Party A "expects" party B to do X (maybe  
>>> useful for certain TRs we've listed already)
>>> 3. "Business trust" = Party A "trusts" party B to do X (presumes  
>>> testable compliance to assure "technical trust" underneath --  
>>> obtains only within hard protocol boundaries)
>>> If we think this has the potential to work, we could have one  
>>> diagram the tools and what they rely on each other for, and  
>>> another diagram for the parties and what they expect/trust each  
>>> other to do.  We could review the specific action verbs in the  
>>> second column of the trust model document in this light.
>>> Thoughts?
>>> 	Eve
>>> On 4 Mar 2011, at 7:47 AM, Mark Lizar wrote:
>>>> ++1
>>>> I very much agree.  I support the relies upon approach for message
>>>> flows as this reliance dictates the needed notice for each  
>>>> relationship.
>>>> As Jeff and Rainer have done well to point out the flow and  
>>>> direction
>>>> of the Host and AM relationship needs to be clear for that to be
>>>> effective to progress these forward.
>>>> On 4 Mar 2011, at 14:11, Thomas Hardjono wrote:
>>>>> Thanks Mark and Rainer,
>>>>> Yes the definition of "trust" can be many and broad :-)   For  
>>>>> UMA, I
>>>>> believe we need to focus on building-blocks or modules that can be
>>>>> identified and which has a clear API purpose (eg. expected,  
>>>>> inputs,
>>>>> expected outputs/behaviors). I think Domenico's diagram is on  
>>>>> track to
>>>>> doing this.
>>>>> So, for example, if a UMA spec states that a User "introduces a  
>>>>> Host
>>>>> to a AM", we need to be very exact as to the message flows,  
>>>>> parameters
>>>>> exchanged in the message, and the status of information stored  
>>>>> at the
>>>>> AM/Host (ie. expected privacy-behavior of entities (AM/Host) when
>>>>> holding User-related data), etc.
>>>>> By doing so I believe:
>>>>> (a) UMA will go beyond OAUTH2.0 as it stands today, and
>>>>> (b) UMA will facilitate (make life easier for) the folks  
>>>>> developing
>>>>> Trust Frameworks and writing contracts based on these  
>>>>> frameworks, and
>>>>> therefore get faster adoption.
>>>>> /thomas/
>>>>> _______________________
>>>>>> -----Original Message-----
>>>>>> From: Mark Lizar [mailto:mark at smartspecies.com]
>>>>>> Sent: Thursday, March 03, 2011 2:57 PM
>>>>>> To: Rainer Hörbe
>>>>>> Cc: Thomas Hardjono; 'UMA WG WG'; Thomas Hardjono
>>>>>> Subject: Re: [WG-UMA] Definition of "Trust" (borrowing from TCG)
>>>>>> Rainer,
>>>>>> I just skimmed this doc and found it faintly useful for this
>>>>> discussion
>>>>>> of Trust.  I highly recommend getting Piotr Cota's book as it is
>>>>> very
>>>>>> technically detailed and is useful for developing a trust  
>>>>>> framework
>>>>>> meta-model as it technically discusses all of the elements and  
>>>>>> their
>>>>>> relationships for what we are proposing with a trust framework.
>>>>>> Trust has 17 different basic definitions.  I think for the most  
>>>>>> part
>>>>> we
>>>>>> use trust in relation to the definition where expectations exist
>>>>>> between the AU and the AM.  In this context I would suggest  
>>>>>> that the
>>>>>> term confidence, or reliance is used instead of trust.
>>>>>> as for Trust Relationships referred to here in UMA   is this
>>>>> referring
>>>>>> to a technical trust between technical actors?
>>>>>> With my sociologist hat on, I am of the opinion that the intent  
>>>>>> of a
>>>>>> trust framework for identity is so that it can be trustworthy for
>>>>> the
>>>>>> management of identity in society.   IF this is true than we  
>>>>>> should
>>>>>> avoid mixing the technical trust, with the social trust as this  
>>>>>> may
>>>>>> lead to creating a significant challenge to any effort to  
>>>>>> create a
>>>>> meta
>>>>>> model of trust frameworks.
>>>>>> When possible I suggest using the term control or confidence  
>>>>>> instead
>>>>>> of trust  in the technical context we are applying it to here.
>>>>> E.g.
>>>>>> a confidence framework can be developed so that it is  
>>>>>> trustworthy..
>>>>>> I found this table of contents and foreward summary of Piotr's  
>>>>>> book.
>>>>>> (attached)
>>>>>> Best Regards,
>>>>>> Mark
> Eve Maler                                  http://www.xmlgrrl.com/blog
> +1 425 345 6756                         http://www.twitter.com/xmlgrrl

More information about the WG-UMA mailing list