Social Web Working Group Teleconference

27 Jun 2017

Attendees

Present: sandro, cwebber, eprodrom, tsyesika, ben_thatmustbeme, ajordan, rhiaro, tantek
Regrets Chair: eprodrom
Scribe: sandro, cwebber2

cwebber2: the test suite is moving along, but still not done
... the client-server: posting an activity, create, testing for non-activiries, media upload testing, following ... all works
... main things not there: add/remove/like/block
... needs more container support
... this week I'll be doing that
... I should be able to demo next week

<cwebber2> https://github.com/w3c/activitypub/issues/235

<Loqi> [cwebber] #235 Add a Tag type

cwebber2: a few issues
... (description of #235)
... Gargron and Puck expressed they'd like a type of Tags that are not Mentions
... AP has defined some new properties, but no new types yet

eprodrom: I'd like to check with James to confirm why it's not in there now

<eprodrom> #hashtag

eprodrom: I think we're talking *hashtags*, but I'd like to be clear

cwebber2: Yes, stringy-name-type-things

<Loqi> tantek: ajordan left you a message 1 day, 22 hours ago: just realized you would've been offline when I sent https://chat.indieweb.org/social/2017-06-24#t1498354046031000. nice to meet you :-)

eprodrom: I'm reluctant to have AP make this kind of change to AS. If it's necessary, then we can, but I'd like to be careful about this, and hear from James.
... like, What Is A Hashtag?
... is it anything more than a string? And if so, what?

cwebber2: The CG spent a lot of time talking about it, all we care about is name, and you localize it for your own instance
... for people who are using types, you might want to have a separate class for this kind of thing

eprodrom: Using the name 'tag' is a tricky, because you also tag a person in a photo, or tag a place, using a tag relationship, so calling this thing a tag might be "tricky"
... I'll ask James

<tantek> hence why we have indieweb.org/person-tag

ajordan: We're talking about adding something to AP, so I'm not sure why we wouldn't do this as an AS extension

cwebber2: Since AP kind of *is* an AS extension.... you're saying do it independently?

ajordan: Yes, tags are more broadly applicable than AP. Make it an AS extension that AP requires.

cwebber2: I'd be fine with that. No strong opinion

eprodrom: Yes, I'd second that, ajordan. This has nothing to do with Protocol per se, unlike inbox and outbox.
... You volunteering ajordan ? :-)

sandro: Let's wait to hear from James before writing it

<tantek> cwebber, did anyone who felt strongly volunteer to document the use-cases?

<tantek> there are no use-cases in the github issue

cwebber2: And let's see how needed it is

rhiaro: I use AS collections as tags on my site

<tantek> I am wary of plumbing-driven design

eprodrom: Checking, we're missing the websub folks today.

<cwebber2> https://github.com/w3c/activitypub/issues/231

<Loqi> [cwebber] #231 "Sensitive Media" tag

#231 Sensitive Media

cwebber2: CG didn't like WG's plan to do it on tags
... Mastodon is super happy putting it in Summary Field
... because they don't plan on changing their interface
... they want people to have a free field they just type into
... so I was going to withdraw that proposal
... on #231 Sensitive Media

<cwebber2> https://github.com/w3c/activitypub/issues/231

<Loqi> [cwebber] #231 "Sensitive Media" tag

scribe has been confused about sensitive media vs content warning

<tantek> same sandro

cwebber2: Diaspora has this too
... It's just a boolean on the whole post

<tantek> from the author? auto-marked by the server based on heuristics?

cwebber2: There's another bonus, in that this could solve Content Warning. If you mark a post as Sensitive, then the summary section becomes a Content Warning.
... Put this boolean on the tags, might solve both problems.

<ajordan> sandro: you're talking about YouTube?

<ben_thatmustbeme> is anyone scribing this?

<tantek> sandro noted a debacle about Youtube blocking LGBT content as "sensitive" in general

<tantek> I'm kinda shocked by the naivete in this conversation

<ben_thatmustbeme> scribenick: ben_thatmustbeme

sandro: i don't htink having one bit is socially feasable

<sandro> sandro: bottom line, I dn't think reducing things to one bit is socially feasible

<scribe> scribenick: sandro

tantek: I'll go farther than that -- it's horribly naive
... Like twitter marking any tweet with the word 'queer' was marked as Sensitive and then semi-blocked
... unless someone's got some research showing how this really works well for users, then ... we've no chance of getting this right

cwebber2: In mastodon, its the sender, not the service provider, who is doing the marking
... I see how a single bit is a problem

<eprodrom> https://gist.github.com/evanp/5968d874cee259a8c0e0059848b08ba3

cwebber2: but this is what's implemented currently, so if we leave it out, they'll have to use an extension
... their own extension

eprodrom: Two issues

<eprodrom> sensitive: true

eprodrom: for an activity for an object, there could be a Sensitive bit
... OR we could use it on tags, as in gist

"tag": [

{

"id": "http://www.cartoonnetwork.ca/tv/steven-universe/",

"name": "Steven Universe",

"sensitive": true

}

<tantek> doesn't matter if the intent is only for authors to specify it. if you put it in a format or protocol, services will use it, as they have been (ab)using sensitive/content warning in their existing UIs

]

<tantek> I still think it's a bad idea

cwebber2: That's my preferred direction to go

<saranix> "they'll have to use an extension" -- their usage is unique even amongst this working group, noone else does it the same as them, so I say it SHOULD be an extension for the way they do it

eprodrom: That says that Steven Universe is sensitive, rather than the post is sensitive

cwebber2: Good semantic point

eprodrom: So, let's be careful about that

<Zakim> ajordan, you wanted to respond to cwebber2

ajordan: cwebber2 you made a point earlier about this isn't a problem because in Mastodon users choose it themselves

<tantek> The larger problem is that "sensitive" is culture-specific, just as "NSFW" was

ajordan: but I would like to say, about that, we should not design with that assumption

<ben_thatmustbeme> ajordan++

<Loqi> ajordan has 11 karma

<tantek> so it's fundamentally flawed to assume a boolean communicates that

ajordan: because at some point someone will build an automated system using this bit

cwebber2: I agree, it's dangerous
... I was making the point that the people proposing this are using it themselves
... I agree, I'm not comfortable with sensitive flag as explained here
... Not sure what to do if there are existing instances using this property

sandro: I think it's a fine extension, but let's help them figure out what happens when it crosses boundaries to systems that don't implemment it. Does it drop the content, or drop the bit?

eprodrom: This is much more about AS than AP

<rhiaro> how do content warnings from mastodon translate to friendica?

<rhiaro> this sould be find-out-able..

<ajordan> rhiaro++ that's a good question

<Loqi> rhiaro has 149 karma in this channel (266 overall)

tantek: To focus on the positive use case cwebber2 expressed of Mastodon folks self-marking posts, as to how folks would expect that to work in a federated system, when the systems have a very different cultures and backgrounds.

sandro: But whoever is writing the spec for that extension can be the center for that discussion

cwebber2: I like Evan's suggestion, if there are vocab extensions that don't relate to protocol, we could make them be extensions to AS instead of changes to AP

<eprodrom> PROPOSED: vocabulary extensions unrelated to the protocol aspects of AP should be handled as AS2 extensions and not as part of AP

<cwebber2> +1

<ajordan> +1

<eprodrom> +1

+1 until/unless someone points out a problem we're not currently thinking about :-)

<rhiaro> +1

ajordan: The Likes and Shares collections should perhaps be in AS instead?

RESOLUTION: vocabulary extensions unrelated to the protocol aspects of AP should be handled as AS2 extensions and not as part of AP

cwebber2: I don't think so, because the protocol says when/how to put things into that collection, but we can figure it out later

JF2

<ben_thatmustbeme> changelog: http://dissolve.github.io/jf2/#changes-from-12-june-2017-wd-to-this-version

<scribe> scribe: cwebber2

<scribe> scribenick: cwebber2

<ben_thatmustbeme> can you not hear me?

<ben_thatmustbeme> ugh

<sandro> ben_thatmustbeme, we're not hearing you

<ben_thatmustbeme> hold on, reloggin on

<tantek> ok I'm not finding any journalists on this, just anecdotes

<tantek> https://twitter.com/Maltimoree/status/877976748947886081

<Loqi> [@Maltimoree] Again, just in case it wasn't clear. Twitter blocks #queer keyword. Talk to me about joining #mastodonsocial.

<eprodrom> PROPOSED: publish editor's draft 21 June 2017 of JF2 as new working draft

ben_thatmustbeme: when registering IANA, they had concerns about the security concerns section so this is an update to that section and removing one bit

<sandro> +1

<ajordan> +1 although I'm wondering why the At Risk note got removed?

<eprodrom> +1

<tantek> +1

<ben_thatmustbeme> i can update the changelog to explain that

<sandro> ben_thatmustbeme: with At Risk removed, link headers, head, and body can all be used

RESOLUTION: publish editor's draft 21 June 2017 of JF2 as new working draft

+1

<sandro> or link to the previous draft's at-risk section

eprodrom: what happens next with jf2?

ben_thatmustbeme: well I've been experimenting with it, and once it's registered I'd like to bring it to more feed readers as an alternative, and see if I can get the jsonfeed crowd to look at it and experiment with it

<tantek> +1

ben_thatmustbeme: I think they're discovering issues with not having a mimetype, etc. hopefully it'll also add more credence being under w3.org rather than random website, even if it's a note rather than a rec-track spec
... after that? I could see options for adding an actual profile for individual items, but I'm not sure about that yet

<tantek> what's the implementation status?

ben_thatmustbeme: I know aaron uses it that way

<tantek> any?

<tantek> (of current draft)

I don't see any major changes coming now

eprodrom: will there be a test suite?

ben_thatmustbeme: this isn't on rec-track, so it's only going to a note... i could make a test suite or validator, I think kevinmarks already started one

sandro: just to clarify, putting it on rec-track would violate the spirit of the extension, but going through the same motions of validator / test suite are good things

tantek: I was going to echo what sandro said, the only way we could reasonably handle it beyond an extension is.. the way we saw it around an extension is that clearly we saw implementation around mastodon etc... so I'd want something like that be to be true of JF2 or anything else before we considered an fpwd towards a rec
... we should meet the bar of what we generally set for an extension, which is a bit higher bar for this WG but I think that's the bar we've set up for
... I think we'd want a good number of implementations, short of that we'd want to keep it as a note

ben_thatmustbeme: I did want to ask... since we're on what's allowed under the extension is one thing that came up in the CG is how to do or modernize webfinger into actual urls and etc to see if everyone could get on the same page...

eprodrom: we're at 3 minutes before the hour and we have to talk about scheduling, I'm reluctant to go into webfinger

<tantek> can we talk summer time?

ben_thatmustbeme: I was just asking about adding notes

sandro: short thing is I think it's best for the CG to tackle it

eprodrom: yes

summer schedule

eprodrom: we're in the heat of summer, this is often when people go on vacation and it's tough to get everyone on the same irc schedule. that said we're running pretty lean these days, usually between 5-10 people on the telecon, so again I'd love to get proposals out by september or october time frame rather than slipping into the fall and be under a lot of pressure then
... that's my main feeling is I'd like to keep us going on an aggressive schedule during the summer, but open to other ideas

<sandro> +1 sticking to aggresive schedule

<ben_thatmustbeme> i would be +1 to going every other week

tantek: next week tuesday is july 4th in the us

<sandro> +1 skip next week

eprodrom: let's skip next week as an edge case

ben_thatmustbeme: I would be in favor of going to every other week because I'm in the CG now, which is taking up a lot more time since those go long

tantek: based on # of attendees recently it's not a bad suggestion

sandro: but we need the websub people in here

eprodrom: for me my big concern is that these are working months for us, and if we get to end of october / beginning of november and we're like oh gosh we need to get this out, we'll be under time pressure... that said a lot of what we need is implementations. I would be ok with every other week as long as we feel like we're making progress

tantek: at least july and august might be good

sandro: maybe we could just look at particular weeks?

tantek: starting july 11th

sandro: yup.....
... any regrets for july 11th?

<eprodrom> meetings on 7//11 and 7/18

<ben_thatmustbeme> works for me

<ben_thatmustbeme> +1

<eprodrom> 7/11 and 7/25

eprodrom: let's say meetings on 7/11 and 7/25

<rhiaro> july all 80% likely okay for me

<ben_thatmustbeme> +1

<eprodrom> August schedule TBD

<tantek> regrets 2017-08-01 if that's a possibility

sandro: question is when we can get aaronpk and julien on the call

<tantek> regrets 2017-08-08 also

eprodrom: in that case let's wrap up

<eprodrom> \o/

<rhiaro> o/

ajordan: I think we found out a good way to bridge indieweb sites and activitypub

trackbot, end meeting

<rhiaro> better late than never RRSAgent

<ben_thatmustbeme> wat?

<rhiaro> by the way ben_thatmustbeme, if for some reason the html minutes aren't generated the perl script that generates them is knocking around somewhere (scribe.perl) and can be run locally

<rhiaro> https://dev.w3.org/2002/scribe/scribedoc.htm

<sandro> Link to resolution: https://www.w3.org/2017/06/27-social-irc.html#T17-52-22

<sandro> not the final one, but from IRC which is good enough