As I wrote at the phabricator task, I agree in principle. But the devil is in the details, of course, and as one of the couple of people who are
de facto running Incubator right now, I need to be involved in all of this.
One of the things that this discussion has me thinking about, though, is whether Incubator should actually be effectively closed and locked, or whether there should be three tiers: projects, incubating projects, and then
Incubator. Here's why I'm thinking along these lines (even if only as a transition step):
- As of the last major evaluation of Incubator (last winter), there were 1,020 tests on Incubator with at least one valid page of content. One was the most recently exported project, which we generally keep as a duplicate on Incubator for administrative reasons.
Of the other 1,019:
- 502 (49%) were either "active" (defined as one new page creation since the beginning of 2017) or "substantial" (defined as having at least 25 mainspace pages), or both. This included two that were approved but awaiting creation at the time.
- Of the remainder, only 15 had sufficient activity to meet the project approval activity requirement. Perhaps another 15 or so were pretty close.
- My estimate (purely an estimate) is that there are rarely more than 40–50 tests with substantial activity at any point in time.
- Incubator also provides a certain buffer zone around tests that are kind of borderline with respect to the current Language Proposal Policy. Many such projects are all the same very legitimate tests with communities working on them, and meet Incubator's
less restrictive rules for creating tests.
- Looking at the above, I'm pretty sure that at least at the beginning, we should only move out the most active projects, perhaps 20 to no more than about 50. This way, we can get the bugs out without having created 500 or so incubation subdomains. Certainly
during that period of time Incubator would stay open as usual for all other tests. After that, I think there are some serious things to think about:
- If a test is fairly substantial (25 pages? 100 pages?), do we create the incubation subdomain even if the test has been dormant for a while?
- Conversely, many tests open with a flurry of activity (over 1 day to 2 months), then go dormant. Going forward, do we really want to create incubation subdomains for these right away, and then have them go dark? Or do we want there to be some kind of threshold
for creating incubation subdomains? And if there's some kind of threshold, then Incubator needs to remain alive for projects not yet there.
What I think makes a lot of sense is for the most active, close-to-ready tests to move into incubation subdomains, where they can start having access to Wikidata, get rid of prefixes, and so forth. I'm not sure that means there isn't a place for Incubator
as a place for projects to get started in the extreme early stages.