W3C home > Mailing lists > Public > www-tag@w3.org > June 2021

Re: Principles of Identity in Web Architecture

From: Melvin Carvalho <melvincarvalho@gmail.com>
Date: Thu, 17 Jun 2021 12:09:35 +0200
Message-ID: <CAKaEYh+m6Ms-Gqh3WOqEQu3jiX_ccwqBg_FVAwEVMRTrYW1PdQ@mail.gmail.com>
To: Larry Masinter <LMM@acm.org>
Cc: "Patrick J. Hayes" <phayes@ihmc.org>, Philip Sheldrake <philip@eulerpartners.com>, TAG List <www-tag@w3.org>
On Wed, 16 Jun 2021 at 21:11, Larry Masinter <LMM@acm.org> wrote:

> The issue of identity metadata came up in the IETF’s RFC-interest. Should
> “Postal Address” be deprecated in the XML2RFC schema? Because it added
> complexity and potential instability to the software the RFC Editor would
> use to convert from (normative) XML to its rendering as HTML, PDF,
> (Unicode) Text.
>
>
>
> I called it Identity Metadata.
>
> “Identity metadata is data the author supplied to establish their identity.  Whether you could send postal mail to the person isn’t the point.”
>
>
>
> The rfc-interest June 2021 Archive by thread (rfc-editor.org) <https://www.rfc-editor.org/pipermail/rfc-interest/2021-June/thread.html#12105>
>
>
Yes, I think distinguishing between data and meta data is architecturally
important, though not always easy

The reason to do this is related to my point (5) "*Separate data and
protocol meta data from identity data"*

This probably could do with some rewording.  The ideas is that it is
undesirable to let leaky abstractions preventing different systems to grow
a network effect.  And that's an easy trap to fall into.

For example, in an HTML page, you have a HEAD and a BODY which generally
separates data from meta data, tho there is some overlap.  You also have
headers which are protocol level meta data, again with some overlap

If you wanted to swap out HTML or HTTP in future and have it interoperate
with another protocol or serialization, then treating the data, and meta
data as separate concerns makes the coding orders of magnitude easier

The example outlined of a postal address, highlights how tricky a mental
task this can be.  To what degree are street, region, code meta data about
a postal address, and to what degree are they 'actual' attributes of a Thing

This is done quite inconsistently on the web today, even with proposed
newer protocols.  For example, with the decentralized identifier (DID)
system, there was at one point an attempt to insert a predicate "created",
and it was unclear whether that meant that the document was created at that
time, or the identifier was created at that time.  The other famous example
of this is HR14

Rather than a philosophical debate, we could reframe it in a more practical
and positive way.  ie that if create tools to identify and fix leaky
abstractions, more easily, we can increase interoperability on the web

In this case the clean separation of "identity" data (agents), data
(things) and meta data (protocol specific items), together with naming


>
>
> --
>
> https://LarryMasinter.net https://interlisp.org
>
>
>
> *From:* Melvin Carvalho <melvincarvalho@gmail.com>
> *Sent:* Wednesday, June 16, 2021 11:45 AM
> *To:* Patrick J. Hayes <phayes@ihmc.org>
> *Cc:* Philip Sheldrake <philip@eulerpartners.com>; TAG List <
> www-tag@w3.org>
> *Subject:* Re: Principles of Identity in Web Architecture
>
>
>
>
>
>
>
> On Mon, 14 Jun 2021 at 03:41, Patrick J. Hayes <phayes@ihmc.org> wrote:
>
>
>
>
>
> On Jun 6, 2021, at 8:50 AM, Melvin Carvalho <melvincarvalho@gmail.com>
> wrote:
>
>
>
>
>
> On Sun, 6 Jun 2021 at 14:02, Philip Sheldrake <philip@eulerpartners.com>
> wrote:
>
> ….
>
>
> It is reassuring to see you distinguish “identity” and identifiers in the
> context of humans, but your email here indicates to me that you still
> consider Alice to have just the one identity. This aligns of course with
> the comparatively recent (centuries) bureaucratisation of identity, aka
> legal identity, and the imperative for Sybil resistance in democratic,
> taxation, and wealth distribution contexts, but I have yet to find another
> discipline beyond law and information technology conceiving identity as
> either singular or enduring. Quite the opposite.
>
>
>
> You raise a good point.  Alice can have many identities.  …
>
>
>
> Ahem. Allow me to call BS at this point. Alice does not have many
> identities. LIke everyone else who has ever drawn breath, and indeed like
> every endurant object, Alice has one identity. She has it, uniquely and
> irrevocably, from the moment she was born, to the end of her life. She is
> one person.
>
>
>
> I understand what Philip is saying here, but he wildly overstates his
> point. In his minority opinion chapter for the SSI book he starts by
> talking about players in Monopoly being represented by the board pieces,
> then in a few sentences segued all the way to:  " Intuitively following
> some self-reflection, we all know that we have different identities, call
> them personae if you like, or avatars, that we adopt in different
> contexts.", then after reminding us that "All the worlds a stage." he goes
> on to "I referred to this earlier in the context of ‘being’ different
> across different social media in different contexts, and constantly
> revising those identities based on the corresponding contextual
> relationships and interactions. Consider your professional avatar, your
> parental avatar, your spousal avatar, your student avatar, etc. Overlapping
> and interacting in some respects no doubt, but always evolving and always
> contextual." And he sums up: "Your ‘you’ in the performance review meeting
> at work differs to your ‘you’ on your wedding day. Your avatars at age 35
> will differ from those age 25, or indeed from those age 34. Or 34½. Or
> yesterday. Psychologists and sociologists understand all this well."
>
>
>
> Psychologists and sociologists here, of course, as contrasted with
> unimaginative computer nerds. But also as opposed to clear-thinking
> philosophers.
>
>
>
> Philip in his email response is careful to use scare quotes, speaking of
> "identities" rather than identities. He also refers to personae and
> avatars. Fine, no doubt we all can have many of these things that
> sociologists study, and they are of course contextual and perhaps (though I
> would dispute with him on this) highly transient and flexible. But these
> things, whatever they are, are not identities. They are not what is talked
> about when people use owl:sameAs, or what we mean when we say that a
> supreme court judge, a cancer patient, a tennis player and a worshipper in
> the Episcopal Church are all the SAME person. These "identities", pieces on
> some super-Monopoly game, avatars of ourselves in social games, are not
> distinct human entities. If the cancer patient dies, so do the SCOTUS
> justice and the tennis player. If the tennis player is tried for fraud or
> theft, so is the SCOTUS judge. Identity means /being the same thing/, or in
> this more limited sense /being the same person/, and sameness of personhood
> is something far more fundamental, and in the end far simpler, than these
> social avatars/projections/roles that Philip is talking about. It means
> simply being the same person. In a legal sense to be sure but also in a
> biological, personal-identity, continuity of memory, continuity of physical
> identity sense. And, I claim, in an ordinary everyday common sense. Alice
> is Alice, all one of her. She is herself, and nobody else.
>
>
>
> So perhaps it could be beneficial to separate the digital world and the
> human world, here
>
>
>
> Alice could have many digital identities (one for social nets, one for
> microblogging etc.) and one physical or legal identity
>
>
>
> The way that the web has evolved, is that these digital identities largely
> operate in silos, even when the stated goal is interoperability
>
>
>
> The aim of raising this architectural point is to try and make it easier
> for different systems to interoperate, and benefit from unexpected reuse.
> Many systems want to do this, but seemingly dont know how to
>
>
>
> I know with identity, you can always say, "its complicated", but that
> doesnt get you too far
>
>
>
> It would be nice to create a document and say to digital platforms, 'If
> you follow these points, you're going to be easier to interop with'
>
>
>
> That could apply to web based systems, P2P, block chain etc. all of which
> have a notion of identity, albeit, slightly different
>
>
>
>
>
> Pat Hayes
>
>
Received on Thursday, 17 June 2021 10:10:25 UTC

This archive was generated by hypermail 2.4.0 : Thursday, 17 June 2021 10:10:36 UTC