[Social] Problem related to namespaces... ID proliferation.

[Joe Cascio, Jr.]

I'd like to hear some discussion on a related topic, which is ID
proliferation. I would think Chris Messina might have something to say on
the topic, being involved in the DISO project. In addition to the problem of
having more than one person having "@susan" there is a growing problem of a
single Susan being Susan (AT) twitter (DOT) com, Susie (AT) identi (DOT) ca, SueBlue (AT) whatever (DOT) com,
etc.

I'm bothered by the fact that most discussion seems to assume that identity
is merely a projection of XMPP's ID mechanism. Wouldn't it be better to have
someone's OpenID meta-data provide a discovery mechanism for any of several
servers that they might be contacted on? Then the XMPP address/identity
would be a lower level routing, perhaps invisible to the end user?

This surely doesn't solve the SMS problem, which is due to the fact that
simply more characters are needed to create globally unique addresses, but
I'd like to make one observation about short (domain-specific) vs. long
(domain-independent) names.

In the set of all people that I talk to, on Twitter, email, or IM, I have a
hard time coming up with any two that have the same local short name. Yes,
there are multiple JoeCascios out there, but I don't think any of my online
contacts know them. Ok, so my name is fairly uncommon. What about a
JohnSmith? How many JohnSmiths do you know? Couldn't potential conflicts be
handled by a personal nickname? The globally unique low level id could be
mapped to my own local short alias for that person. The default short alias
is their own defined short name, but I could override that to let me
distinguish messages arriving at my device.

JoeC

[Henry H]

You just summarized the last post (as of now - sans the summary post) of the
thread you're referring to - or you can just look at the last line of the
summary post :-)

On Thu, Jul 31, 2008 at 8:11 AM, Joe Cascio, Jr. <joec0914 (AT) gmail (DOT) com> wrote:

> I'd like to hear some discussion on a related topic, which is ID
> proliferation. I would think Chris Messina might have something to say on
> the topic, being involved in the DISO project. In addition to the problem of
> having more than one person having "@susan" there is a growing problem of a
> single Susan being Susan (AT) twitter (DOT) com, Susie (AT) identi (DOT) ca,
> SueBlue (AT) whatever (DOT) com, etc.
>
> I'm bothered by the fact that most discussion seems to assume that identity
> is merely a projection of XMPP's ID mechanism. Wouldn't it be better to have
> someone's OpenID meta-data provide a discovery mechanism for any of several
> servers that they might be contacted on? Then the XMPP address/identity
> would be a lower level routing, perhaps invisible to the end user?
>
> This surely doesn't solve the SMS problem, which is due to the fact that
> simply more characters are needed to create globally unique addresses, but
> I'd like to make one observation about short (domain-specific) vs. long
> (domain-independent) names.
>
> In the set of all people that I talk to, on Twitter, email, or IM, I have a
> hard time coming up with any two that have the same local short name. Yes,
> there are multiple JoeCascios out there, but I don't think any of my online
> contacts know them. Ok, so my name is fairly uncommon. What about a
> JohnSmith? How many JohnSmiths do you know? Couldn't potential conflicts be
> handled by a personal nickname? The globally unique low level id could be
> mapped to my own local short alias for that person. The default short alias
> is their own defined short name, but I could override that to let me
> distinguish messages arriving at my device.
>
> JoeC
>

[Mark Morgan]

On Thu, Jul 31, 2008 at 2:11 PM, Joe Cascio, Jr. <joec0914 (AT) gmail (DOT) com> wrote:
> I'm bothered by the fact that most discussion seems to assume that identity
> is merely a projection of XMPP's ID mechanism. Wouldn't it be better to have
> someone's OpenID meta-data provide a discovery mechanism for any of several
> servers that they might be contacted on? Then the XMPP address/identity
> would be a lower level routing, perhaps invisible to the end user?

There are a few things which are being assumed by this thread and the
one it forked off from:

a/ there would be a method of discovering their presence on each of
the different services at any given time, so you know what one to
route to. Possible with some that are stateful, but not such things
as email, sms, twitter (I don't think anyway, I'm not too sure how
twitter operates).

b/ if there is a way of discovering the above, the user you're trying
to contact may have a preference as to which one things should be
routed to, which would ideally need to be taken into account.

c/ conversation context would also need to be taken into account. If
a conversation started on one medium, it normally would be best to
continue it on that same medium, even if a higher-priority 'route'
becomes available. Not necessarily as important for private
conversations, but moreso for public ones, else a chat may suddenly
appear on the given serivce in the middle of a conversation.

d/ invisible routing would always be preferred. This often won't be
the case, for instance, if the highest priority route (if that could
be determined from the above) is one that's public (twitter, blog,
whatever else), you don't want to be sending them very personal
messages. Even having an option for specifying the routing, with a
default if none given is not ideal, as it would then be up to the
sender to always take the routine into account anyway, in case it was
a 'private' message, with a potentially very embarrassing result if
that is forgotten.

Take care,
Mark.

[Seth Fitzsimmons]

Long, long ago I ran into a similar (but much on a much smaller scale)
problem. Here's how I solved it:
If I'm using identi.ca, @susan refers to susan (AT) identi (DOT) ca; for twitter,
susan (AT) twitter (DOT) com. If I'm using identi.ca and want to refer to
susan (AT) twitter (DOT) com, I would use @susan (AT) twitter (DOT) com (ugly with the double @'s;
in my old system, we used !'s; e.g. !susan (AT) twitter (DOT) com! or
!susan (AT) twitter (DOT) com:susan on twitter!).

When my note is syndicated to another system, the reference is rewritten
from the perspective of the system it's being read on, so a viewer on
twitter would see a message from "me (AT) identi (DOT) ca" to "@susan", but a viewer on
identi.ca would see a message from "me" to "susan (AT) twitter (DOT) com".

Using the !user[@node][:description]! variant (or something similar, perhaps
with []'s), you can define nicknames (or just infer from someone's contact
list) and rewrite them into that form automatically.

seth

On Thu, Jul 31, 2008 at 9:11 AM, Joe Cascio, Jr. <joec0914 (AT) gmail (DOT) com> wrote:

> I'd like to hear some discussion on a related topic, which is ID
> proliferation. I would think Chris Messina might have something to say on
> the topic, being involved in the DISO project. In addition to the problem of
> having more than one person having "@susan" there is a growing problem of a
> single Susan being Susan (AT) twitter (DOT) com, Susie (AT) identi (DOT) ca,
> SueBlue (AT) whatever (DOT) com, etc.
>
> I'm bothered by the fact that most discussion seems to assume that identity
> is merely a projection of XMPP's ID mechanism. Wouldn't it be better to have
> someone's OpenID meta-data provide a discovery mechanism for any of several
> servers that they might be contacted on? Then the XMPP address/identity
> would be a lower level routing, perhaps invisible to the end user?
>
> This surely doesn't solve the SMS problem, which is due to the fact that
> simply more characters are needed to create globally unique addresses, but
> I'd like to make one observation about short (domain-specific) vs. long
> (domain-independent) names.
>
> In the set of all people that I talk to, on Twitter, email, or IM, I have a
> hard time coming up with any two that have the same local short name. Yes,
> there are multiple JoeCascios out there, but I don't think any of my online
> contacts know them. Ok, so my name is fairly uncommon. What about a
> JohnSmith? How many JohnSmiths do you know? Couldn't potential conflicts be
> handled by a personal nickname? The globally unique low level id could be
> mapped to my own local short alias for that person. The default short alias
> is their own defined short name, but I could override that to let me
> distinguish messages arriving at my device.
>
> JoeC
>

[Bob Wyman]

On Thu, Jul 31, 2008 at 12:00 PM, Seth Fitzsimmons <seth (AT) mojodna (DOT) net> wrote:
> a viewer on twitter would see a message
> from "me (AT) identi (DOT) ca" to "@susan", but a
> viewer on identi.ca would see a message
> from "me" to "susan (AT) twitter (DOT) com".

If your Tweets/Dents were being logged to an RSS file at Identi.ca, how
should things look in that file? The file is "dumb" and can't tell whether a
reader is an Identi.ca user or a Twitter user... Am I correct in assuming
that one should consider readers of the file to be "outsiders" and thus both
names would be fully expanded with no nicknames? (e.g. a message from
me (AT) identica (DOT) ca to susan (AT) twitter (DOT) com)

bob wyman

[Ralph Meijer]

Bob Wyman wrote:
> On Thu, Jul 31, 2008 at 12:00 PM, Seth Fitzsimmons <seth (AT) mojodna (DOT) net
> <mailto:seth (AT) mojodna (DOT) net>> wrote:
> > a viewer on twitter would see a message
> > from "me (AT) identi (DOT) ca <mailto:me (AT) identi (DOT) ca>" to "@susan", but a
> > viewer on identi.ca <http://identi.ca/> would see a message
> > from "me" to "susan (AT) twitter (DOT) com <mailto:susan (AT) twitter (DOT) com>".
>
> If your Tweets/Dents were being logged to an RSS file at Identi.ca, how
> should things look in that file? The file is "dumb" and can't tell
> whether a reader is an Identi.ca user or a Twitter user... Am I correct
> in assuming that one should consider readers of the file to be
> "outsiders" and thus both names would be fully expanded with no
> nicknames? (e.g. a message from me (AT) identica (DOT) ca <mailto:me (AT) identica (DOT) ca>
> to susan (AT) twitter (DOT) com <mailto:susan (AT) twitter (DOT) com>)

That seems good to me. Just like it is usually discouraged to use
relative links in feeds. Also, as somebody else suggested, those nicks
could instead be marked up, to point to a JID or web page. I wouldn't
use e-mail addresses as in your example.

Note that I think that (Atom) feeds should use the Atom threading
extensions to point to the post responded on, and that one might also
use <link rel='related'/> elements in the entry to point to related
entities.

ralphm

[Henry H]

At this point, we are into visual semantics.

One could argue susan (AT) twitter (DOT) com is better because it's a well-known
convention for email. Others could argue that http://identi.ca/susan is
more friendly for use in RSS feeds. Someone else could argue using a
different convention like susan/twitter.com without an @ sign is better.

Personally I don't care either way, but I'd avoid creating something
entirely new because it's an age-old problem with existing and proven
mechanisms for addressing. Why re-create the wheel with some "new"
identification system? The problem is simple so keep the solution simple.
At its roots this is a messaging platform just like every other messaging
platform out there (real-time, near real-time, etc.) which needs to address
messaging routing, tracking, security, privacy, etc.

The Internet is based on network layers (packets, mac addresses, IP
addresses, domain names, users within domains) - the only reason to add
another layer is for usability (because you are not providing more
functionality). The functionality is already there and being used. If you
want better searching, less chatty protocols, better tracking then you can
add additional layers on top of the network stack but it always comes at a
cost. Unless you of course plan to build this messaging platform on top of
something besides the Internet (not likely for too many reasons to name).

Hey, I'd love to put an @mom on an envelope and drop it in any U.S. post
office mailbox and have that routed to my mother, but I would hate to be the
U.S. post office if that was ever implemented. And if my mother lived
outside the U.S., I would hate the job of the person who is required to
translate mail addresses between other countries and the U.S.

The biggest challenge here is everyone has to agree on it a "standard", but
standards usually occur AFTER the fact, not before. Adoption first,
standards second as the space matures. Naturally something that works well
and is adopted quickly usually becomes the basis for a standard - people are
lazy. Therefore the solution needs to be seamless or invisible and usually
based on something familar. Make something complex and it won't be adopted
by anyone but the most tech savvy (i.e. the people on this list). Standards
driven by simplicity naturally become the most widely adopted - it's a
self-fulling circle.

In the battle between natural standards (e.g. think evolution) versus
artificial standards (e.g. you will now use RFC 31234115 to define message
length, blah blah blah because it's needed to ensure flexibility for the one
day in the future where everyone will be hooked directly up into the Matrix)

:-)





On Thu, Jul 31, 2008 at 11:38 AM, Ralph Meijer <jabber.org (AT) ralphm (DOT) ik.nu>wrote:

> Bob Wyman wrote:
>
>> On Thu, Jul 31, 2008 at 12:00 PM, Seth Fitzsimmons <seth (AT) mojodna (DOT) net<mailto:
>> seth (AT) mojodna (DOT) net>> wrote:
>> > a viewer on twitter would see a message
>> > from "me (AT) identi (DOT) ca <mailto:me (AT) identi (DOT) ca>" to "@susan", but a
>> > viewer on identi.ca <http://identi.ca/> would see a message
>> > from "me" to "susan (AT) twitter (DOT) com <mailto:susan (AT) twitter (DOT) com>".
>>
>> If your Tweets/Dents were being logged to an RSS file at Identi.ca, how
>> should things look in that file? The file is "dumb" and can't tell whether a
>> reader is an Identi.ca user or a Twitter user... Am I correct in assuming
>> that one should consider readers of the file to be "outsiders" and thus both
>> names would be fully expanded with no nicknames? (e.g. a message from
>> me (AT) identica (DOT) ca <mailto:me (AT) identica (DOT) ca> to susan (AT) twitter (DOT) com <mailto:
>> susan (AT) twitter (DOT) com>)
>>
>
> That seems good to me. Just like it is usually discouraged to use relative
> links in feeds. Also, as somebody else suggested, those nicks could instead
> be marked up, to point to a JID or web page. I wouldn't use e-mail addresses
> as in your example.
>
> Note that I think that (Atom) feeds should use the Atom threading
> extensions to point to the post responded on, and that one might also use
> <link rel='related'/> elements in the entry to point to related entities.
>
> ralphm
>