Users trying to translate the same topic at the same brings interesting challenges that can be addressed in different ways. Before we define any collaboration model, some information on how and how much those conflicts are happening can be really useful.
This task proposes to define and implement metrics that inform on how often and in which circumstances the mentioned overlap happens. Some possible signals to base metrics on:
- Repeated drafts. Drafts that are started on a topic for which there is already another draft for the same language pair. This allows to understand how many users are working on the same time on the same topic.
- Publishing conflict. An article exists at the time the user tries to publish a translation. It may be interesting to be able to check how many of the "repeated drafts" lead to publishing conflicts, and whether the conflicting article already existed when the translation was started or not.
- Abandoned drafts. Drafts not edited in a while which have not been published. A reverse "active drafts" can be used instead.
The above signals need to be made specific to become metrics. I expect these metrics to provide us information such as "Few users happen to choose the same topic to translate (5% in general but up to 10% in smaller Wikipedias). From those most (70%) end up in a publishing conflict that we may want to prevent doing X". Note that this is just an example to illustrate how the signals are related, we don't know how big/small the overlap will be.