Implementors Spec Meetup

[Monica Keller]

Time flies when you are busy… these are exciting times !

Who would be able to meet this coming Friday Aug 14th in downtown SF
around lunchtime to have a spec hack half day were we collectively
edit the implementers draft spec with the changes we agreed on at the
SJC meetup whilst reviewing of course.

http://groups.google.com/group/activity-streams/browse_thread/thread/483c8e9a2a566353

We can have this at the MySpace SF office on 2nd. Let me know. It is
just heads down work. If people want to come hack on validation tools
that would be great as well. Then the following week we can present to
the community.

Thanks for the help !

[Ryan Boyd (Google)]

Would love to attend.  Unfortunately, I'm going to be moving apartments at that time.

Cheers,

-Ryan

[Kevin Marks]

Sounds like a plan, we did say 3 weeks - Would be great to get
FriendFeed and FB people to this one.

2009/8/11 Monica Keller <monica...@gmail.com>:

[Monica Keller]

Hey Guys
Looks like we don't have enough people for tomorrow. What other dates
nearby work for you ?
Let me know this is an important effort thanks for your help

On Aug 11, 3:57 pm, Kevin Marks <kevinma...@gmail.com> wrote:
> Sounds like a plan, we did say 3 weeks - Would be great to get
> FriendFeed and FB people to this one.
>

> 2009/8/11 Monica Keller <monica.kel...@gmail.com>:

>
>
>
>
>
> > Time flies when you are busy… these are exciting times !
>
> > Who would be  able to meet this coming  Friday Aug 14th in downtown SF
> > around lunchtime to have a spec hack half day were we collectively
> > edit the implementers draft spec with the changes we agreed on at the
> > SJC meetup whilst reviewing of course.
>

> >http://groups.google.com/group/activity-streams/browse_thread/thread/...

>
> > We can have this at the MySpace SF office on 2nd. Let me know. It is
> > just heads down work. If people want to come hack on validation tools
> > that would be great as well. Then the following week we can present to
> > the community.
>

> > Thanks for the help !- Hide quoted text -
>
> - Show quoted text -

[David Recordon]

I could probably make it.

[Martin Atkins]

Monica Keller wrote:
> Hey Guys
> Looks like we don't have enough people for tomorrow. What other dates
> nearby work for you ?
> Let me know this is an important effort thanks for your help
>

Sorry. I had some mail filtering problems that caused me to miss this
discussion.

Tomorrow works for me.

[Kaliya]

I am interested in coming by.

On Aug 11, 8:57 am, Monica Keller <monica.kel...@gmail.com> wrote:
> Time flies when you are busy… these are exciting times !
>
> Who would be  able to meet this coming  Friday Aug 14th in downtown SF
> around lunchtime to have a spec hack half day were we collectively
> edit the implementers draft spec with the changes we agreed on at the
> SJC meetup whilst reviewing of course.
>

> http://groups.google.com/group/activity-streams/browse_thread/thread/...

[Monica Keller]

Guys sorry I didn't book my flight for tomorrow because I thought
people could not make it. I will be flying into SF saturday morning.
Martin says he can help hack over the weekend. Anyone else ?

How about Saturday lunch at 21st Ammendment ? or if not Monday I will
also be in SF and can meet.

And if you can't meet saturday or monday feel free to do some hacking
on friday too I just won't be able to join.

But let me know. Thanks !

[Kaliya]

Saturday works for me.

Kaliya
www.identitywoman.net

On Aug 13, 2009, at 6:46 PM, Monica Keller <monica...@gmail.com>
wrote:

[da...@sixapart.com]

I'd lean toward Monday, but Saturday could also work.

--
Sent from my iPhone

On Aug 13, 2009, at 6:48 PM, Monica Keller <monica...@gmail.com>
wrote:

>

[Chris Messina]

Would love to be there, but I've got a wedding to be at...! 

Friday would have worked for me too — bummer! — sorry I didn't respond sooner...!

Monday is a lost cause for me but I would like to attend a meeting sooner than later. :(

I think it might be worth looking at the discussions happening here in light of FriendFeed's acquisition:

http://friendfeed.com/clonefeed

Essentially a bunch of folks want to create an open source clone of FriendFeed... they've been looking at AS, so marshaling the developer potential from that group would be very excellent!

Chris

--
Chris Messina
Open Web Advocate

Personal: http://factoryjoe.com
Follow me on Twitter: http://twitter.com/chrismessina

Citizen Agency: http://citizenagency.com
Diso Project: http://diso-project.org
OpenID Foundation: http://openid.net

This email is:   [ ] bloggable    [X] ask first   [ ] private

[Monica Keller]

Sorry about that Chris. We will catchup with you soon.

Ok for tomorrow 12:30 21st Ammendment
http://upcoming.yahoo.com/event/4230178/?ps=5

[Monica Keller]

Hi Guys we are done hacking on the first implementors draft spec.
Please check out the changes here

http://github.com/apparentlymart/activity-streams-specs/tree/master

And the 2 specs linked of here: http://www.activitystrea.ms/

Questions/TODOs that came during the session:
*Why did we have invite if no one has implemented it ?
*How do we represent friend relationships ?
*Need to finish modeling for a place includng location (free form)
*Finish modeling events including hosts and xcal ref
*Need to finish modeling for the review
*Finish modeling mood and annotation
*Discuss renaming note to status
*Should we model Message
*Can we rename file to document ?
*Add entity or user instead of person to make it more accurate
*Write security considerations
*Do we need to change song to audio ?

Let us know your thoughts before the next meetup used to accept this
first implementors draft

[Christian Crumlish]

sorry I couldn't make it today... home life beckoned:

On Sat, Aug 15, 2009 at 5:04 PM, Monica Keller <monica...@gmail.com> wrote:

*How do we represent friend relationships ?

this is a giant can of worms. given the various models (mainly reciprocal friendships vs. one-way following) it might be worth parsing out all the possibilities carefully.

 

*Can we rename file to document ?

vs. attachment (per email) or enclosure (per RSS)?

*Add entity or user instead of person to make it more accurate

I think I need to bone up on past discussions about entities vs. persons.

-x-

--
Christian Crumlish
I'm writing a book so please forgive any lag
http://designingsocialinterfaces.com

[Chris Messina]

On Sat, Aug 15, 2009 at 5:04 PM, Monica Keller <monica...@gmail.com> wrote:

Hi Guys we are done hacking on the first implementors draft spec.

Sweet!

Questions/TODOs that came during the session:
*Why did we have invite if no one has implemented it ?

Great question. When was it added? Was it intended for friend invites or event invites?

In either case, it doesn't seem to pertain to any public data — it seems more like a private activity type.

 

*How do we represent friend relationships ?

As Christian said, this is a huge can of worms that will spiral out of control unless we bound it to a particular context or purpose.

The basic purpose of revealing relationships was to mirror the [highly popular] activity type we saw on Facebook (and similarly on LinkedIn) when people confirmed each other as friends. Since Facebook only allows reciprocal relationships, it was useful to be able to state that "A and B are now friends", meaning that A had sent a request for friendship to B, who accepted.

In the non-reciprocal/follower model, it's a simpler statement to say that "A is following B" without regard for B's confirmation of a bi-lateral relationship. Even if B were following A and the relationship was being reciprocated, it wouldn't be necessary to point this out in the activity.

In the activity streams model, we may desire to model such mutually confirmed relationships, but I feel like it's beyond the scope of the first draft.

In the first draft, I think we should aim to stick to affirmative/active verbs and activities from the point of the individual. 

That is, if I subscribe the Monica's activities, I should be subscribing to things that she's doing or has done — rather than things that she is doing with others, or that require the involvement of others. Another way to evaluate prospective verbs and activities is to think, "If Monica said 'I just did X', would that make sense?"

Therefore, Monica could say "I just followed Martin", but she could not say "I just friended Martin and Martin friended me." (that is, Monica should not be able to make a statement where she becomes the direct object of someone's else activity) — this is simpler than I'm making it sound, but I'm struggling with the language.

Anyway, for now, let's err on the side of utter simplicity with relationships in this first draft and document where the holes appear. 

 

*Need to finish modeling for a place includng location (free form)

We need to document how to use both GeoRSS for location as well as our own free-form location tag. Presumably we'll want to support hcard where possible... or maybe just add an element called "adr"?

 

*Finish modeling events including hosts and xcal ref
*Need to finish modeling for the review
*Finish modeling mood and annotation
*Discuss renaming note to status
*Should we model Message

Can you describe how this would be different from a note/status? 

How would this differ? Was the idea that there would be a recipient/target?

 

*Can we rename file to document ?

What is motivating this? The difference is of course subtle, but it seems like a file pertains to a superset of types that include both documents (typically text-based) as well as binary files like MP3s.

What examples are there that suggest a need for this change?

 

*Add entity or user instead of person to make it more accurate

Examples in the wild? Can you also provide a example of what this kind of feed object would look like?

 

*Write security considerations
*Do we need to change song to audio ?

That would be fine with me if we model song-titles in the AtomMedia format (I presume we do).

Chris

 

Let us know your thoughts before the next meetup used to accept this
first implementors draft

[Christian Crumlish]

On Sun, Aug 16, 2009 at 5:09 PM, Chris Messina <chris....@gmail.com> wrote:

*Add entity or user instead of person to make it more accurate

Examples in the wild? Can you also provide a example of what this kind of feed object would look like?

 

I wasn't privy to this conversation, so I'm not sure of the connotation of "entity or user' (the user part suggests it is a search for a generic subscriber type with actual people being a specific type, but I'm not sure). I don't quite picture how an entity "follows" unless it simply means subscribing to a feed, which of course any machine or process can do.

I can imagine use cases for wanting to be able to follow entities and not just people. A common analogy might be the difference between becoming a fan of a Comcast page on Facebook and following an individual posting on Twitter, say, as a Comcast employee.

[Chris Messina]

On Sun, Aug 16, 2009 at 5:15 PM, Christian Crumlish <xi...@pobox.com> wrote:

On Sun, Aug 16, 2009 at 5:09 PM, Chris Messina <chris....@gmail.com> wrote:

*Add entity or user instead of person to make it more accurate

I can imagine use cases for wanting to be able to follow entities and not just people. A common analogy might be the difference between becoming a fan of a Comcast page on Facebook and following an individual posting on Twitter, say, as a Comcast employee.

Right. I can see this making sense for bands, companies or other "aggregate entities", where there the language would need to be changed to "it" instead of "he" or "she".

Indeed, the "Diso Project" or "OpenID Foundation" would be entities — so we could start there as the canonical examples.

Chris

[Kevin Marks]

On Sun, Aug 16, 2009 at 5:09 PM, Chris Messina<chris....@gmail.com> wrote:
> On Sat, Aug 15, 2009 at 5:04 PM, Monica Keller <monica...@gmail.com>
> wrote:
>>
>> Hi Guys we are done hacking on the first implementors draft spec.
>
> Sweet!
>>
>> Questions/TODOs that came during the session:
>> *Why did we have invite if no one has implemented it ?
>
> Great question. When was it added? Was it intended for friend invites or
> event invites?
> In either case, it doesn't seem to pertain to any public data — it seems
> more like a private activity type.
>

Invitations and responses are a key kind of activity, well represented
in streams already— for example on upcoming.yahoo.com and Facebook. I
think public vs private is orthogonal to the purpose of the activities
spec - we should be able to keep the expression of activities in
streams separate from the 'who gets to see what' aspect (which we can
presumably say 'use OAuth' to).

The Opensocial location type is a union of human-readable address and
geopoint, which is handy.

[Monica Keller]

I feel like we should remove verb "invite" from the implementors spec
because no one has implemented it as far as I know.

Totally agree that events are very important. But usually in
activities which are descritions of relevant actions we see Monica is
hosting event X (verb=post) and Monica is attending event
Y(verb=rsvp-yes)

I have never seen Monica invited Chris to event A and we have not even
discussed how we would model "Chris" (target?)

This is a bit stalkerish and so I would like to not include for phase 1

Aso talking about events for implementors draft 1 I feel we need to
fully document xCal/hCalendar support
which means we can add dtStart, dtEnd, location and adr from hCard.

MySpace has implemented this and I think it helps our v1 to be usable.
<vevent xmlns="urn:ietf:params:xml:ns:xcal">
<uid>tag:myspace.com,2009:/Bandshow/38538527/38538527</uid>
<summary>beverly hills</summary>
<dtStart>2009-07-01T03:00:00Z</dtStart>
<location>407 N Maple Dr, beverly hills, 91607, US</location>
</vevent>


Please see
http://microformats.org/wiki/hcalendar

And
http://developerwiki.myspace.com/index.php?title=ActivityStream_Events#Band_Shows

[Monica Keller]

Oh and hCalendar also has geo as you know.

Overall I have found its easier to map activitystrea.ms objects to
microformats than opensocial objects....they are a bit further along

[Kaliya *]

On Sun, Aug 16, 2009 at 5:18 PM, Chris Messina <chris....@gmail.com> wrote:

On Sun, Aug 16, 2009 at 5:15 PM, Christian Crumlish <xi...@pobox.com> wrote:

On Sun, Aug 16, 2009 at 5:09 PM, Chris Messina <chris....@gmail.com> wrote:

*Add entity or user instead of person to make it more accurate

I can imagine use cases for wanting to be able to follow entities and not just people. A common analogy might be the difference between becoming a fan of a Comcast page on Facebook and following an individual posting on Twitter, say, as a Comcast employee.

Right. I can see this making sense for bands, companies or other "aggregate entities", where there the language would need to be changed to "it" instead of "he" or "she".

Indeed, the "Diso Project" or "OpenID Foundation" would be entities — so we could start there as the canonical examples.

I asked this question - noticing there would be activity stream output from a twitter handle /userID from an organization (company or nonprofit) that was not a "person" - or a band for example.


The identity gang lexicon has two terms that can be helpful in thinking about this. (http://wiki.idcommons.net/Lexicon) 

--------
Entity (http://wiki.idcommons.net/Entity)

Definition: A person, physical object, animal, or juridical entity

Comment1: In an identity system implementation an Entity is abstract, conceptual, non-modelled.

Comment2: An Entity can be an application or service.

----------
Party is the term for legal persons -  http://wiki.idcommons.net/Party

Definition: A natural person or a juridical entity.

The thing that the word party does not cover is an "informal group" (without any legal standing) or a sub group - like a department of comcast a company like comcast - I guess entity does.

-Kaliya

 

[Martin Atkins]

Chris Messina wrote:
>
> *Add entity or user instead of person to make it more accurate
>
>
> Examples in the wild? Can you also provide a example of what this kind
> of feed object would look like?
>

My thought here was that we would simply replace "person" with "user"
and keep all of its properties the same.

Rationale: Not all user accounts represent people. Some user accounts
represent companies or other entities. To keep things simple, we should
only have one type for v1 -- since that's all that most apps can support
today anyway -- and that type should be called "user". Later, if it
becomes useful to do so, we may add "person", "entity", etc as sub-types
of "user".

There are of course some counter-examples. Facebook already has the
user/page distinction, and MySpace has various non-person account types
such as "band", etc. But on the whole most sites just have user
accounts, and non-Person entities are represented by a normal user
account where multiple users happen to have the password.

Right now there seems to be no use-case for distinguishing these things
in an activity stream, since both are generally just referred to by
their name anyway. Later it may become useful to distinguish them so
that, for example, people can be referred to by gendered personal
pronouns, though in many cases sites treat gender as private information
or don't have gender at all, and some individuals don't have genders
that fit into the range of gendered pronouns supported by the English
language.

[Dan Brickley]

On 18/08/2009 00:35, Kaliya * wrote:
>
>
> On Sun, Aug 16, 2009 at 5:18 PM, Chris Messina <chris....@gmail.com
> <mailto:chris....@gmail.com>> wrote:
>
> On Sun, Aug 16, 2009 at 5:15 PM, Christian Crumlish <xi...@pobox.com
> <mailto:xi...@pobox.com>> wrote:
>
> On Sun, Aug 16, 2009 at 5:09 PM, Chris Messina
> <chris....@gmail.com <mailto:chris....@gmail.com>> wrote:
>
>
>
> *Add entity or user instead of person to make it more
> accurate
>
>
>
> I can imagine use cases for wanting to be able to follow
> entities and not just people. A common analogy might be the
> difference between becoming a fan of a Comcast page on Facebook
> and following an individual posting on Twitter, say, as a
> Comcast employee.
>
>
> Right. I can see this making sense for bands, companies or other
> "aggregate entities", where there the language would need to be
> changed to "it" instead of "he" or "she".
>
> Indeed, the "Diso Project" or "OpenID Foundation" would be entities
> — so we could start there as the canonical examples.
>
>
> I asked this question - noticing there would be activity stream output
> from a twitter handle /userID from an organization (company or
> nonprofit) that was not a "person" - or a band for example.

Such distinctions are well worth making.

> The identity gang lexicon has two terms that can be helpful in thinking
> about this. (http://wiki.idcommons.net/Lexicon)
>
> --------
> Entity (http://wiki.idcommons.net/Entity)
>
> Definition: A person, physical object, animal, or juridical entity
>
> Comment1: In an identity system implementation an Entity is abstract,
> conceptual, non-modelled.
>
> Comment2: An Entity can be an application or service.

Can you give any example of something that is *not* an entity (in this
sense)?

In FOAF we have long had a class "Agent", but (just as in the Dublin
Core project, who have a similar definition) it turned out to be hard to
say much about what counts, beyond "things that do stuff". Pretty much
anything can be usefully described as an agent in some context. Despite
that, the term is pretty useful, eg. for describing common properties of
people, pets, software bots, organizations etc. We got by without
anything approximating identity gang's "Party", although it is
appealing. And we also have Group (broadly defined such that it includes
self-aware groups, as well as groups defined by queries) and
Organization. In FOAF, we've managed to avoid saying much about where
Group and Organization overlap and which to use when.

So fwiw, in FOAF, Agent is a superclass of Person, Organization, Group.

Back to the question of trying to find an example of something that is
*not* an Agent (or "Entity"), for a while I thought we could at least
say that agents and documents were disjoint. But then if you think about
pages with embedded software (or people with tattoos), even that
distinction seems tough to sustain.

If you do end up using a class - eg Entity - that covers all things that
exist, I'd suggest mentioning that explicitly in the spec up front;
it'll save people wondering whether their object fits the definition...

cheers,

Dan

[Elias Bizannes]

+1 on entity.

It's something ingrained in common law which exists in North American,
UK, and Commonwealth countries like Australia.

An entity is something that has it's own legal identity, like a human
being or a corporation. An agent, is an entity that acts on behalf of
another entity.

It may be worth while comparing legal definitions from the various
company law acts in countries to give a common definition, rather than
reinventing the wheel here.

Sent from my iPhone

[Chris Messina]

I'm not sure that bringing legal definitions and terminology into this work would be useful.

Instead, we want to model what is already being done — and make this format easy to implement without having to learn too much new jargon.

I think "entity" is okay from a general perspective — that is, it generically seems to cover all the bases — but doesn't appear to be human-friendly... additionally "user" implies a hierarchical relationship, as opposed to an original position of creation or authorship.

Martin, can you be specific about where "person" is used that needs a substitution? It seems that using "actor" generally is sufficient — as it covers all kinds of "entities" — and "nouns that do things".

Chris

[Martin Atkins]

Chris Messina wrote:
>
> Martin, can you be specific about where "person" is used that needs a
> substitution? It seems that using "actor" generally is sufficient — as
> it covers all kinds of "entities" — and "nouns that do things".
>

Imagine an activitystrea.ms version of the feed for the sixapart Twitter
account.

What would the type of activity:subject and/or activity:actor be?

Right now we have no type suitable for describing Six Apart.
Furthermore, Twitter doesn't have any way to determine whether a
particular user account is a person or not.

While "actor" could be used, it seems confusing to have both an element
and an object type both called "actor", and "user" matches how this kind
of object is generally referred to in applications and APIs that are
already deployed, including Twitter.

Therefore assuming that we want to generalize person -- which it seems
like we do -- I think "user" is the best term for it based on existing
practice. We would still retain the term "actor" for the property of an
activity whose value is the object that did the activity, which will in
most cases be an object of type "user".

[Chris Messina]

Yeah, I see.

In the case of Twitter, of course you're right — they don't distinguish between people, companies, bots or other types of "entities". 

Can you give me a "before and after" example of how this discussion of "users" vs "entities" would currently effect an implementation? I'd like to see what an entry would look like dealing with a "company entity" — perhaps the case of @sixapart on Twitter?

Chris

[Kaliya *]

On Tue, Aug 18, 2009 at 8:55 AM, Elias Bizannes <elias.b...@gmail.com> wrote:

+1 on entity.

It's something ingrained in common law which exists in North American,
UK, and Commonwealth countries like Australia.

An entity is something that has it's own legal identity, like a human
being or a corporation.  An agent, is an entity that acts on behalf of
another entity.

Nope - an entity is a "thing" you can point to - person, rock, animal application.

A Party is a something with a legal identity - human or corporation.
 

[Kaliya *]

On Tue, Aug 18, 2009 at 10:57 AM, Chris Messina <chris....@gmail.com> wrote:

I'm not sure that bringing legal definitions and terminology into this work would be useful.

Instead, we want to model what is already being done — and make this format easy to implement without having to learn too much new jargon.

I think "entity" is okay from a general perspective — that is, it generically seems to cover all the bases — but doesn't appear to be human-friendly... additionally "user" implies a hierarchical relationship, as opposed to an original position of creation or authorship.

I think changing to "user" makes sense and then if next rev the subtypes of "people" and "comapnies,orgs,groups" happens - great.   Just didn't want to have it be "person" when this was not accurate of who actual users were.
 

[Nate Benes]

You folks in and around SF are lucky, I feel so left out over here in Nebraska.

Nate

nate...@gmail.com

[Monica Keller]

Martin why did we change Person to User already ?
http://martin.atkins.me.uk/specs/activitystreams/activityschema#anchor13

This breaks all the current implementations which were using
http://activitystrea.ms/schema/1.0/person

I think we need to add person back in and have be a specialization of
user

Then later on we can add other users like "Band"

On Aug 18, 10:57 am, Chris Messina <chris.mess...@gmail.com> wrote:
> I'm not sure that bringing legal definitions and terminology into this work
> would be useful.
> Instead, we want to model what is already being done — and make this format
> easy to implement without having to learn too much new jargon.
>
> I think "entity" is okay from a general perspective — that is, it
> generically seems to cover all the bases — but doesn't appear to be
> human-friendly... additionally "user" implies a hierarchical relationship,
> as opposed to an original position of creation or authorship.
>
> Martin, can you be specific about where "person" is used that needs a
> substitution? It seems that using "actor" generally is sufficient — as it
> covers all kinds of "entities" — and "nouns that do things".
>
> Chris
>

> On Tue, Aug 18, 2009 at 8:55 AM, Elias Bizannes <elias.bizan...@gmail.com>wrote:
>
>
>
>
>
>
>
> > +1 on entity.
>
> > It's something ingrained in common law which exists in North American,
> > UK, and Commonwealth countries like Australia.
>
> > An entity is something that has it's own legal identity, like a human
> > being or a corporation.  An agent, is an entity that acts on behalf of
> > another entity.
>
> > It may be worth while comparing legal definitions from the various
> > company law acts in countries to give a common definition, rather than
> > reinventing the wheel here.
>
> > Sent from my iPhone
>
> > On 18/08/2009, at 2:17 AM, Dan Brickley <dan...@danbri.org> wrote:
>
> > > On 18/08/2009 00:35, Kaliya * wrote:
>

> > >> On Sun, Aug 16, 2009 at 5:18 PM, Chris Messina <chris.mess...@gmail.com
> > >> <mailto:chris.mess...@gmail.com>> wrote:
>
> > >>    On Sun, Aug 16, 2009 at 5:15 PM, Christian Crumlish

> > >> <x...@pobox.com

> > >>    <mailto:x...@pobox.com>> wrote:
>
> > >>        On Sun, Aug 16, 2009 at 5:09 PM, Chris Messina

> > >>        <chris.mess...@gmail.com <mailto:chris.mess...@gmail.com>>

> This email is:   [ ] bloggable    [X] ask first   [ ] private- Hide quoted text -
>
> - Show quoted text -

[Monica Keller]

I am trying to have MySpace stay as compliant as possible with the
spec but we keep making non backward compatible changes. This person
to user thing should be on the next rev or if we think its very
critical can we please make it backward compatible. Please dont pull
the rug from under our feet :(

On Aug 21, 6:44 pm, Monica Keller <monica.kel...@gmail.com> wrote:
> Martin why did we change Person to User already ?http://martin.atkins.me.uk/specs/activitystreams/activityschema#anchor13
>
> This breaks all the current implementations which were usinghttp://activitystrea.ms/schema/1.0/person

> > - Show quoted text -- Hide quoted text -

[Chris Messina]

While the spec can change from time to time (especially before we get to a "blessed" 1.0), I think it is worth considering the ramifications of a change like this — even if it's meant to simplify or clarify the usage of a particular element of the spec.

Before Monica's complaint, no one really spoke up about the consequences of the change — but perhaps that's because few people have implemented the spec at scale like Monica has.

Given that we need to work with existing implementors while improving the spec, I'd suggest reverting the change back to "person" for now — and through the the first release of the spec. As Monica said, if we want to change it — or perhaps add more "actor types" — later, we should consider that when we have actual use cases rather than preparing for ones that are merely speculative so far.

Chris

[Phil Wolff]

re: person / user / entity

We have real use cases in that facebook groups and pages and linkedin groups and the upcoming "twitter corporate accounts" are all non-person entities generating feeds today. if this is a big change, best to make it early.

Phil Wolff
managing editor, Skype Journal
http://SkypeJournal.com
pwo...@skypejournal.com
skype:evanwolf
+1-510-444-8234 San Francisco
+1-510-316-9773 mobile
http://www.linkedin.com/in/philwolff
http://www.facebook.com/profile.php?id=724232370
http://twitter.com/evanwolf
http://dataportability.org