This has been a while coming; thank you for your patience. I’m very happy to be able to share the final four candidates for Mozilla’s new community-facing synchronous messaging system.
These candidates were assessed on a variety of axes, most importantly Community Participation Guideline enforcement and accessibility, but also including team requirements from engineering, organizational-values alignment, usability, utility and cost. To close out, I’ll talk about the options we haven’t chosen and why, but for the moment let’s lead with the punchline.
Our candidates are:
- Mattermost,
- Matrix/Riot.IM,
- Rocket.Chat, and
- Slack
We’ve been spoiled for choice here – there were a bunch of good-looking options that didn’t make it to the final four – but these are the choices that generally seem to meet our current institutional needs and organizational goals.
We haven’t stood up a test instance for Slack, on the theory that Mozilla already has a surprising number of volunteer-focused Slack instances running already – Common Voice, Devtools and A-Frame, for example, among many others – but we’re standing up official test instances of each of the other candidates shortly, and they’ll be available for open testing soon.
The trial period for these will last about a month. Once they’re spun up, we’ll be taking feedback in dedicated channels on each of those servers, as well as in #synchronicity on IRC.mozilla.org, and we’ll be creating a forum on Mozilla’s community Discourse instance as well. We’ll have the specifics for you at the same time as those servers will be opened up and, of course you can always email me.
I hope that if you’re interested in this stuff you can find some time to try out each of these options and see how well they fit your needs. Our timeline for this transition is:
- From September 12th through October 9th, we’ll be running the proof of concept trials and taking feedback.
- From October 9th through the 30th, we’re going discuss that feedback, draft a proposed post-IRC plan and muster stakeholder approval.
- On December 1st, assuming we can gather that support, we will stand up the new service.
- And finally – allowing transition time for support tooling and developers – no later than March 1st 2020, IRC.m.o will be shut down.
In implementation terms, there are a few practical things I’d like to mention:
- At the end of the trial period, all of these instances will be turned off and all the information in them will be deleted. The only way to win the temporary-permanent game is not to play; they’re all getting decommed and our eventual selection will get stood up properly afterwards.
- The first-touch experiences here can be a bit rough; we’re learning how these things work at the same time as you’re trying to use them, so the experience might not be seamless. We definitely want to hear about it when setup or connection problems happen to you, but don’t be too surprised if they do.
- Some of these instances have EULAs you’ll need to click through to get started. Those are there for the test instances, and you shouldn’t expect that in the final products.
- We’ll be testing out administration and moderation tools during this process, so you can expect to see the occasional bot, or somebody getting bounced arbitrarily. The CPG will be in effect on these test instances, and as always if you see something, say something.
- You’re welcome to connect with mobile or alternative clients where those are available; we expect results there to be uneven, and we’d be glad for feedback there as well. There will be links in the feedback document we’ll be sending out when the servers are opened up to collections of those clients.
- Regardless of our choice of public-facing synchronous communications platform, our internal Slack instance will continue to be the “you are inside a Mozilla office” confidential forum. Internal Slack is not going away; that has never been on the table. Whatever the outcome of this process, if you work at Mozilla your manager will still need to be able to find you on Slack, and that is where internal discussions and critical incident management will take place.
… and a few words on some options we didn’t pick and why:
- Zulip, Gitter.IM and Spectrum.Chat all look like strong candidates, but getting them working behind IAM turned out to be either very difficult or impossible given our resources.
- Discord’s terms of service, particularly with respect to the rights they assert over participants’ data, are expansive and very grabby, effectively giving them unlimited rights to do anything they want with anything we put into their service. Coupling that with their active hostility towards interoperability and alternative clients has disqualified them as a community platform.
- Telegram (and a few other mobile-first / chat-first products in that space) looked great for conversations, but not great for work.
- IRCv3 is just not there yet as a protocol, much less in terms of standardization or having extensive, mature client support.
So here we are. It’s such a relief to be able to finally click send on this post. I’d like to thank everyone on Mozilla’s IT and Open Innovation teams for all the work they’ve done to get us this far, and everyone who’s expressed their support (and sympathy, we got lots of that too) for this process. We’re getting closer.