What wiki gets the features first (assuming they don't go out in one wave)

I'll ask these communities if they want to volunteer. That's already tracked by T146972: Announce and follow up with communities about the New Filters for Recent changes Beta deployment.

How many communities do we want to have for that first test?

Do we want to contact English Wikipedia for that first test, that big community that will give us a lot of work and feedback? I would say no, just to work with a short community to skim UX and technical feedback.

What would be the order after that?

I would suggest to follow the TCG's recommandations.

If in the meantime we have some communities that volunteer and do their best to have ORES tests ready, we will include them to the list of communities that will have that beta feature.

In any way, we need to document first. Please see T146669: Create dedicated pages for ERI Recent Changes Beta project to coordinate about that and consider to draft a glossary (T141449: Define Edit Review Improvements glossary) ;)

For a beta feature (i.e. an opt-in feature for communities to try) that does not generate new kinds of content, I'd consider making it available to as many places as possible, announce it, and ask for feedback. Limiting the availability of the beta feature only prevents communities to make an informed decision based on trying the feature in practice (why are we preventing some communities to even try it?). An initial test deployment of the beta feature to one/few wikis in order to check that there are no sever technical issues, seems ok to me, but I'd consider providing the feature to as many wikis as possible as soon as possible, even considering those where ORES services are not available yet.

Releasing as a default experience (i.e., moving it out of beta) is a very different story and we may want to have a more elaborate plan of gradual deployments basing each step on the feedback we receive from the previous one. I consider this to be the point where "when features are released to targeted users" that the TCG guide mentions.

In T150715#2794898, @Pginer-WMF wrote:

An initial test deployment of the beta feature to one/few wikis in order to check that there are no sever technical issues, seems ok to me, but I'd consider providing the feature to as many wikis as possible as soon as possible, even considering those where ORES services are not available yet.

You mean release the filters, no matter if ORES is available? I remind (I may be wrong) we agreed not to do that. But I was supporting the idea of releasing filters with or without ORES.

No matter what, I would recommend to have a two week long test period with a few volunteering wikis first, and them open the Beta to more wikis.

In T150715#2795650, @Trizek-WMF wrote:

In T150715#2794898, @Pginer-WMF wrote:

An initial test deployment of the beta feature to one/few wikis in order to check that there are no sever technical issues, seems ok to me, but I'd consider providing the feature to as many wikis as possible as soon as possible, even considering those where ORES services are not available yet.

You mean release the filters, no matter if ORES is available? I remind (I may be wrong) we agreed not to do that. But I was supporting the idea of releasing filters with or without ORES.

No matter what, I would recommend to have a two week long test period with a few volunteering wikis first, and them open the Beta to more wikis.

That sounds good. All I am asking for is not to delay this process for too long. Given a certain amount of time to work with the communities, it seems much effective to spend it in the process of getting feedback about the available beta feature and how/when to make it default, rather than on discussing whether a community wants to opt-in into an opt-in feature to try and discuss about it.

I think the process of a beta feature could perfectly start with a communication along the lines of "This new tool is intended to X, you can try it now in the beta section. It is already available for you to try and provide feedback about. It is in an early development stage, because we want to incorporate your early feedback. Let us know if it works well, and help us undercover cases we may have not considered.".

We discussed this in the Collab Team meeting today. The team made the following determinations, with the proviso that we would like the CLs to weigh in with their opinions. The thinking was:

• Go with one wiki first, just to make sure nothing is really broken. Then release to the rest of the ORES wikis above pretty quickly after that.
• All things considered, the best candidate for going first seemed to be Portuguese, which is not too large and generally accepting of new technology.

@Quiddity and @Trizek-WMF, what do you think? Anything to add or ammend?

Works for me. We have to ask them asap (saying what?) and have a possible backup. Polish is also a good candidate.

As discussed with @jmatazzoni in our 1:1, the release would follow this order:

1. pt.wp T156157: Contact Portuguese Wikipedia about testing ERI Recent Changes Beta project
2. wikis with ORES enabled, excluding en.wp
3. en.wp
4. other wikis without ORES

I've discussed with @Johan about the question of providing support to communities who wish to integrate ORES to their wiki.
We may have an additional step with Wikidata, I'll see that with @Lea_Lacroix_WMDE.

Not that we're just about to finish T151611: Enable ORES Review Tool on Czech Wikipedia so that'll add a wiki to the list.