XSF Discussion - 2021-01-18

Did anybody notice an increase of users on its server during the last weeks / related to the WhatsApp "privacy policy change thing"?

marc: Signal did...

but yeah, it looks like I have 3x the number of new users per week for the last two or three weeks

I know that signal did ;)

Ge0rG: cool!

Did we send that tweet yet?

I don't think so

Ge0rG: do you have public stats for yax.im?

marc: no

marc, here not much difference on active users, but on the other hand we never had that many MUC users

mathieui: I mean registered users on a server

Anyone looked at the public jabber.fr stats? Do they also increase as yax.ims?

mdosch: do they have public stats?

https://stats.jabberfr.org/

no effect in the past 90 days

there is a fake-effect in the past 30 days for active client connections, but if you go back further you notice that it’s just the recovery from the holiday drop off

Maybe that chapril is more popular now in France?

mdosch, I was talking about the jabber.fr stats

https://op-co.de/tmp/yax.im-registration-peak-whatsapp.png

marc: ^

It seems that I have a light bump on movim.eu

I'll try to do statistics

Ge0rG: wow, quite significant

yax.im is the new WhatsApp!

marc: I didn't filter for spam bots

I see

What's running on jabberfr?

prosody

I ~doubled the registration per day on my side

moving from 10 to 20

I have a captcha, and i've checked the JIDs, they seems legit

Would be nice to have more stats for all the public servers

that's what I was proposing

would be nice to have a XEP to allow a little IQ request

returning some general statistics

Yes

Like the module for the-federation.info?

kinda

But xmpp specific

There are already xmpp servers reporting.

39? Or what number the stats xep had

give me something to scrape, I’ll add it to s.j.n

Next project is better stats for ejabberd, then I will look into this topic more

jonas’: https://mdosch.de/.well-known/x-nodeinfo2

https://jabber.fr/.well-known/x-nodeinfo2

> Ge0rG escribió: > Did we send that tweet yet? Im working on this + a blog post

mdosch, can I get that over XMPP?

Afaik this module only serves the .well-known URL: https://modules.prosody.im/mod_nodeinfo2.html

But maybe another module does, dunno.

https://modules.prosody.im/mod_pubsub_stats.html Maybe

marc that would also greatly help with the "pick a XMPP server" thing

emus: I would have appreciated tweeting a first reaction when the situation was in the headlines, and to follow-up with a blog post later

half a week later it’s so old news, no point in tweeting anything

emus: as it looks now, WhatsApp's new ToS will come back in May, so we need to have something robust prepared for around mid-April

snikket with iOS? :)

jonas’: it's been ten days by now

Ge0rG, thanks

jonas’: but I fully agree with your point

(if that didn't become apparent from my previous statement, already)

How can we improve our social media "reaction time"?

have a paid person doing this a fixed amount of time per week

dwd: have a dedicated person for that

somebody who's familiar with the ecosystem or can consult the right people™ on short notice in a MUC

Ge0rG, I think we do have some people doing that. Are they sufficiently empowered and confident enoughto propose and post quickly?

dwd: apparently not

No, I agree. How can we change that? (And, FWIW, I don't know who these people are, which - as a Board member - is pretty rubbish of me).

Isn't Board the one with final responsibility?

dwd: I did a quick&dirty tweet suggestion on commteam@, on December 8th. I hoped there would be discussion/feedback on the comment, or somebody with twitter rights would pick it up.

dwd: probably what really is needed ia an authoritative person to drive the social media, so that there is no question on who should do the follow-up work

Zash, Yes. But it's sared across the Board. I think we probably need a single person to have the final say, with a policy driven by Board.

As it is now, there are multiple volunteers, and I can understand if they all don't have the resources / think that one of the other folks would do something™

Ge0rG, Shared responsibility doesn't multiply, it divides, indeed.

Tragedy of the commons 😕

dwd: and I suppose that having a 70% perfect tweet at the right time is much better than having a perfect tweet two weeks late

(no parallels to vaccination strategies intended)

100% agree that someone needs responsibility over social media. I'm starting to feel the same way about SCAM as well right now, to be honest. It's not clear what the status is, if anything of FOSDEM and summit.

part of the responsibility should be updating schedules on the wiki

Ge0rG, The enemy of the good tweet is several weeks bikeshedding about the perfect one, I agree. But Board does need to agree on what "Good" looks like.

dwd: it should suffice for Board to define the basic rules for what "Good" looks like, not to decide on each individual tweet

Otherwise we'll only ever tweet on friday evenings.

Ge0rG, Absolutely. Policy not control.

Ge0rG, s/December/January/?

jonas’: oh sorry, indeed.

I'm still caught in 2020

who isn’t

Wake me up when March ends

Zash: we are still in September. September 1993

Reminds me, I plotted the time between a message and a displayed marker on our network across 2020. (Actually average of the fastest 90%, since there's a really long tail). It's pretty muhc inversely proportional to COVID-19 caseloads.

dwd: can you publish a nice marketing write-up on that?

My server will only retain the last 14 days, and even then it's probably illegal for me to aggregate that data.

It's sad to see April 2020 as the first item under "Recent Events", and no mention of Board / membership elections

Upcoming: Google Summer of Code 2020

We really need to assign somebody to maintain the wiki frontpage

> Ge0rG escribió: > emus: as it looks now, WhatsApp's new ToS will come back in May, so we need to have something robust prepared for around mid-April I refuse to jump on each hype train on the opposite. Therefore I take sometime to think what I wanted to say to said. But also I dont have a lot time to do quick reactions always. Yes, if XSF wants quick reactions, we need to hire people for that and other things as well I guess. Anyway, I personally do not share those shorttime reactions/impressions as people usually do.

Ge0rG, how about you?

Ge0rG, I think we will be; however I did most of our analysis on a subsection of our data relating to Defence Medical Services, so I'll probably redo the data across all our users (and probably focus the Blog post on NHS data).

emus: I'm just comparing the public statements of a different federated IM network, and they are doing significantly better. This is not a criticism on anybody involved in our volunteer work, just a thought on what we could be able to improve

Ge0rG, "redo the data" is just re-migrate through the new server's archiving code which does most of this analysis on the fly.

I hate to ask this

emus: but I think that we need to be able to do such short-term hype reactions, even if we all dislike this kind of activity

but as it happens, I am around-ish most of the time anyway ✏

can I haz twitter access?

jonas': I can hand out Twitter access, but would rather be directed by someone on Board to do so.

dwd: that sounds really intriguing

I'd love to see jonas’ as our Social Media Representative

Ge0rG, haha, no

just as a twitter proofreader and proxy

I’m not going to write stuff

Zash: I absolutely lack the time to take any additional workload

Ge0rG, Fairly tightly coupled into our tech stack, though. But lots of its ideas could be reused in other servers.

Ge0rG: Sorry, till thursday last week I had no time for xmpp stuff 🤷‍♂️But as said Im on it. As there is no real holistic onboarding strategy within XMPP a quick thing wont bring us tot he front. Also each project advertises themselves for onboarding. So, if one wants to drop a short tweet is fine. I see my role in thinking about what we want actually say in general and give a proper response. I think that is one outcome in the blog post

emus: I agree on the holistic strategy, and it's a significant second part of marketing

emus: regarding project-oriented advertising, I think that there is a place for those, but there is also a place for advertising the federated ecosystem as a whole

Im sure this topic will be there on 8th of feb again or when ever they hit the ground with that. Im not afraid of "missing the right time" here. If we want to make a difference we shouldnt fear to just go crazy as all the others did. but give me another 24 hrs for the first review.

> Ge0rG escribió: > emus: regarding project-oriented advertising, I think that there is a place for those, but there is also a place for advertising the federated ecosystem as a whole Ok, but what exactly you mean

emus: the ToS enforcement has been delayed, at least for the EU, into May

emus: essentially what I proposed to tweet ten days ago. Advertising the federated XMPP network

I'm not against it, but it won't do much good

okay, ah thats what you meant

MattJ: what would you propose, that does more good?

Maybe we can start writing down an @xmpp Social Media Policy in the wiki?

Ge0rG: I know, but I prefered to write up something more thoroughly (no offence)

People have been pusing XMPP in various places the past few weeks, but pushing "XMPP" as an alternative to "WhatsApp" is a joke for all but 0.5% of WhatsApp users

They are looking for something they can search for on the Play/App store, and install it and move on

emus: and that's great as well. But we could pitch a short-term "hi we are here" first, and then follow-up with a blog-post and another round of twitter then

They're not interested in our little war against silos

MattJ: like... https://joinjabber.org/

Stuff like that is a step in the right direction, yes

> MattJ escribió: > They're not interested in our little war against silos And that is what I try to make a statement about

emus, What do you mean?

> Ge0rG escribió: > emus: and that's great as well. But we could pitch a short-term "hi we are here" first, and then follow-up with a blog-post and another round of twitter then I know, but I had absolutely no time. Sorry for that.

emus: not criticising you, just saying that it would be great to have somebody who has a time budget allocated to this

MattJ, FWIW, I think people broadly do care about decentralization. Just not as much as they care about talking to their friends.

> dwd escribió: > emus, What do you mean? That it is a question of technology, not of the messenger. but as said, gonna sent my suggestion soon

dwd, I think that among the general population (let's say WhatsApp users), you'll find less than 20% who care or understand decentralization

and a far smaller fraction of them are willing/able to actually do something about it when presented with the xmpp.org website

MattJ: but it makes sense to target the nerds with content regarding "easy setup of an encrypted-by-default family chat", and those will convince the larger populace

yep

I agree that the XMPP marketing target audience is not regular people but nerds

for example, a Snikket

MattJ, I think that more people care in general, broad terms. But yes, I agree few will do anything about it - in part because fundamentally, people just want top talk to their friends and family.

> Ge0rG escribió: > MattJ: but it makes sense to target the nerds with content regarding "easy setup of an encrypted-by-default family chat", and those will convince the larger populace good point

How's that working out so far with everybodies friends/family/etc.? (I'm being facetious of course, but I don't think that's true at all: the nerds are almost never the ones to be the early adopters of a technology that gets it off the ground I suspect)

I've not succeeded in introducing xmpp as a family chat because of the lack of featureful clients (especially for iOS)

SamWhited, I think you're wrong, but as someone who has spent over a year trying to make that happen...

My family are all using XMPP, except for the few iPhone users

So 11 family members, and hopefully 3 more to be added once iOS is ready

MattJ: and are they spreading it outside of your family to their friends and your extended family?

Haven't heard much complaints, tho there's that one iOS user who unfortunately is still left out 😕

SamWhited, no, but I'm hoping to fix that very soon

Sponsored the new feature in Conversations that allows you to invite new contacts

SamWhited: I've introduced XMPP as the sole mean to do IM with me, and it looks like five family members use it to communicate with me, except when their videos exceed the http file upload size

and my wife is looking forward to that

I suspect it's exactly the opposite. The nerds all join new shiny things quickly, then something else picks up steam, the nerds resist it for as long as possible, but then everyone else has gotten on Facebook Messenger or whatever and eventually the nerds capitulate and use it. Advertising to a tiny fraction of people is never going to cause something to spread widely.

There's a lot of small groups and "except's" in these examples. Facebook (or whomever) wins because they have large group and very few "excepts". The "excepts" that they do have are likely the nerds like us who don't want to use it for <nerdy reasons>.

Although I also don't understand the goal. It sounds like we're saying "the goal is a general audience, but advertise to the nerd crowd". Whereas if the goal is just to expand XMPP use among the nerd crowd it might make more sense to advertise to them. Maybe we need a clear goal first.

SamWhited: yes, that'd be good

From my perspective: I have my own clear goals, XMPP is coming along for the ride. I don't think you can successfully advertise XMPP to non-nerds (and goodness knows I spent a *long* time doing that, with minimal success)

So I'd rather advertise something more concrete, and use that as the gateway to federation, decentralization, software choice and all the good stuff

Advertising the latter features, which are just concepts, doesn't really help

I don't disagree. I've wanted to do Snikket(sp?) (equivalent) for years, but haven't been able to resource it.

People will see you pushing decentralization, and agree. And they'll stop there. Maybe they'll make an account on a public server. But then they'll stop there. Or maybe they manage to convince a friend. But it 99.9% of the time stops there.

The friction to onboard friends and family (let alone strangers) is too high, there are too many factors and choices involved

Decentralization sounds good, but the problems it solves are kinda abstract

But as I said, I'm not against us (as the XSF) pushing decentralization, it's good to make people aware of it

But as a reaction to the WhatsApp thing, if we're expecting users to flock to "XMPP" like they flocked to Signal, that's just not even remotely possible

I think the XSF pushes (or should) XMPP as a solution to people building things. And then those things are pushed to users.

I agree

And we surely have a lot of work to do in that area alone

Building things might be chat for an MMO, or it might be a private federated chat network, or it might be an XMPP client, but ... yeah.

Make XMPP good so people want to use it to build things, the XSF TL;DR mission statement?

(Or it might be a Snikket)

Zash: +1

Targeting WhatsApp users is far less productive for XMPP than targeting developers with the potential to build the next successful decentralized messaging app

Right now most of those people are being attracted to Matrix (and I don't blame them)

I note that the XSF doesn't have as goal to maximize users, while the Matrix foundation do.

I don't think that's unfixable, and I think there's plenty of room for both protocols to exist. It would be good to provide some resources comparing the two, for example

Zash: I think the XSF should resonably have a goal to attract users, but users of XMPP are people building things.

and an intro to the XMPP ecosystem, which is something I also would like to see produced

I did try to do that some years ago, when I pushed for a certain book. I'm not sure how successful it was though.

Sure, the book helped

But it was one thing, we need to keep up that kind of thing... times have changed, a new book is needed :)

Or a new revision ;)

Or both!

I tend to agree with all of that and that we *should* advertise to the nerd crowd FWIW, but if so we need to be clear about what our goals are. It sounded before like we were saying the goal was the general population (in which case we should not advertise to the nerd crowd)

I think we are able to attract both sides. We just lack the people. Everyone I told who got the idea was convinced its good. Techie or not. And I think we should not refuse to have a valid statement with what we do but also be proud.

The XSF is notoriously bad at providing content for end-users.

It's slightly less bad at providing content for developers. No idea if we should focus our marketing on either as long as we don't have the content resources

But OTOH it's better to have marketing to attract people who might help improve the situation, than to get lost

If we lack people, we should start by attracting people who can attract more people... I have no idea how to do that.

Ge0rG: I see that. I just wanted to keep the horizon open for thinking

PSA: Going to upgrade the wiki, with me luck

Zash: please go ahead!

MediaWiki 1.35.1 LocalSettings.php not found. Please set up the wiki first.

Related to the discussion above: isn't advocacy better managed by a separate, product organisation? That is, one that has a primary remit that isn't producing and promoting technical solutions.

Like, an XMPP Software Foundation?

Zash: a Jabber Software Foundation

PSA: Wiki upgrade should be done now.

Perhaps, yeah (not sure if that already exists and I'm betraying my ignorance here), but perhaps something closer to what Snikket is.

Ge0rG, try it on mobile? 😉

+1 to deuill's comment about advocacy. I've looked into that once or twice in a semi-serious way, but always decided there wouldn't be enough interest from people with appropriate skill sets to start it, unfortunately. It was all very unscientific though, so who knows.

Zash: thanks, much better now!

the main page with two columns is still not perfect, but all the others are much nicer

deuill: I think we should not split up again ressources

Having a separate organization can help focus, but there is of coures some overhead

Yes, but I specifically disagree to decentralise anything else in organisational questions within XMPP/XSF

Actually, I just saw Mongoose wrote something, so even they took a while, which is fine in my view. ✏

I realize this discussion has probably been had to death before, but I don't necessarily think it's about splitting resources. Whether the XSF umbrella should grow or whether there should be a separate organization, there is nothing on the XSF roadmap that is anywhere close to advocacy of things built on top of XMPP (rather than the specifications themselves).

And there's no reason for it to be there -- the XSF is (AFAICT) aimed at developing standards, not necessarily the promotion of products built upon these standards.

Myeah, compare with the W3C or IETF.

(the issue being, people in the XSF are already in other orgs, such as software projects, or server hosting, or companies, and having yet another hat that has to find funding for its activities has to be a pain in the neck)

(I’m not involving myself in promotional stuff because I am terribad at it, and also have not time)

Oh yeah absolutely. I'd probably also say that promoting a specific piece of software as "blessed" might be contentious in and of itself.

let’s not go too far into the topic of the XSF’s "neutral stance", too much time has already been spent on this

The reason I say this, again, as an outsider, is: even if your audience is engineering types, the most common entry-point into the XMPP ecosystem is very likely the products built on top of it.

Exactly. We've already decided to have a neutral stance, for better or for worse, so if we want something to promote anything with a non-neutral stance we have to create it. Unless we want to rehash the XSF's position on that yet again.

SamWhited: is that equivalent to "we should not do any promotion that fits our neutral stance"?

I think being neutral does not mean we should act as a mute person. Just imagine how not clear the xmpp network is for externals (and also internals)

Ge0rG: no, I don't understand how that would follow

I don't know. Maybe I'm still looking for ideas on what the XSF *should* do in terms of marketing

I think we already talked about that. Decide what audience we want to reach first

It is possible to have a Websocket BOSH hybrid for XMPP? Websockets is modern and for HTML5, while it lacks the speed of BOSH for intermittent networks.

sidereal: what do you mean by speed of BOSH?

reliability

sidereal, where does BOSH introduce reliablity that WebSockets do not have?

What's reliable about it?

probably only if you would use a new TCP connection for every BOSH request

but I always assumed that modern HTTP libraries try to reuse the TCP connection

latency, I suppose

I suppose the way out is using XEP-0198 over WebSocket

Well BOSH can survive network changes better than WS or TCP

Zash, how's that?

sidereal: latency, reliability and speed are all different things

Since it goes over HTTP requests

Zash, right, but aren't those HTTP requests performed over the same TCP connection in some (many?) cases? ✏

How do browsers realize that an existing connection is dead?

thanks Zash and Ge0rG.

Ge0rG, not sure if they do it on "desktop" platforms like on linux

yes, Ge0rG, they're different, on an intermittent network, they go hand in hand

but Android tells Chrome when there is a network change, and I think Chrome reacts to that

flow: my chromium is hanging when I try to open github in a new tab, because an old tab is hanging somewhere

sidereal: but they have a significant overhead cost

Both on the server and on the mobile data counters

Also I'm not sure how much better BOSH is behaving when a given request hangs

conversations.im uses both stream management (XEP-0198) and Websockets. So does stream management make up for difference between BOSH and Websockets?

Ge0rG, in theory you can re-try a failed request.

sidereal: partially

Zash, I guess you could not only re-try failed requests, but also re-issue outstanding requests ✏

yep, because sequence numbers

For what would be best for a mobile phone, or an intermittent network. XMPP holds a stream alive to have a conversation. Text messages on the other hand, send a message then don't hold a connection stream open, which uses up bandwidth. When there are local emergencies for instance or during rush hour, texts allow people to communicate without clogging the network system

sidereal: an open connection requires very little traffic as opposed to periodic polling with BOSH

Bosh and maybe Websockets make sense for that, to not require a stream on the mobile end.

it does? with XMPP, the connection drops easily on its own, by the nature of the network, then XMPP reestablishes the link and resends the XML data. It's not only the message, it's the message plus the handshake and resending headers and XML tags

So Websockets and Stream Management together are good enough, or better than BOSH?

and with BOSH it's resending the TLS handshake and HTTP headers

Seems like what you're asking for is push notifications, which are already supported in all servers and mobile clients

It's the same thing essentially, but it's the mobile OS that maintains the connection

Both Android and iOS do this already

it would have to, to remain secure? I would think that would be less traffic, as opposed to resending all of the data with XML and handshakes.

Apple's was (is??) based on XMPP

Wasn't Googles too? Firebase?

Google's was never XMPP to the device

But you could push to it via XMPP

many used XMPP

i.e. to Google

XEP-0357 Push Notifications: informs users of new messages, when their clients aren't online. These indications are delivered through a mobile network

MattJ: Android around 1.0 exposed an XMPP API, so I'm pretty sure it was internally using that for push

Fun

Maybe not for push after all https://android-developers.googleblog.com/2008/08/some-information-on-apis-removed-in.html

the idea of Danger Inc. was to have inter-device intents that are transported via XMPP

But it became a danger?

> But you could push to it via XMPP You can still I think?

But nobody is using that? Instead, people run for the http api

Holger, I vaguely recall some deprecation notice, but I may be wrong

https://firebase.google.com/docs/cloud-messaging/xmpp-server-ref

> Holger, I vaguely recall some deprecation notice, but I may be wrong Same here though 🙂

MattJ, Google was delivering push notifications through Google Talk for a few years. After they deprecated Google Talk, at some point they migrated to a protobuf based protocol for c2s. This protocol as of today still has iq and message stanzas and after login you're assigned a "JID" which has the form user@domain/resource. The hostname of the server endpoint is still mtalk.google.com and I'd guess that internally their infrastructure is still based on whatever their XMPP server was. ✏

Probably a ton of cool code archeology in that code base by now 🙂

Isn't the WebRTC stack related to Jingle too?

yes

webrtc is based on porting google talk browser extension natively into the browser

IIRC there are still jingle related code fragments in libwebrtc in chromium

https://source.chromium.org/chromium/chromium/src/+/master:third_party/libjingle_xmpp/xmpp/xmppclient.cc 😉