Problem:
We are trying to figure out the ideal position of mul in the termbox. The layout might change substantially depending on how many Babel boxes users use.

Question:

• What is the distribution of the Babel box count?
• What are examples for use cases that require a lot of Babel boxes?

How the data will be used:

• If most users use up to 6 Babel boxes then putting mul last in the list of "In other languages" would work. Otherwise we would need an alterative approach.
• If there are use cases that require a lot of Babel boxes, then we need to allow power users to modify the position of mul in the termbox (e.g. by specifying the desired position via Babel box).

Data:

Sample A

• 500 edits from recent changes by registered non-bot editing via the Wikidata UI (URL)
• This resulted in a sample of 102 users.
• Sample A is biased towards power users.

Sample B

• 10000 random Wikidata users from the database (FROM wmf.mediawiki_user_history WHERE wiki_db = 'wikidatawiki' AND snapshot = '2023-01' AND caused_by_event_type = 'create').
• Sample B is biased towards inactive accounts.

Results:
All users in sample A:

• 56% did not use Babel boxes.
• 44 % had one or more Babel boxes.

Only users with Babel boxes in sample A:

• 13% of sampled users with Babel boxes used more than 6 Babel boxes.
• 11% of sampled users with Babel boxes used more than 9 Babel boxes (range 10 to 49).

(Histogram)

Only users with > 9 Babel boxes in sample A:

• They usually had a lot of "-1" and "-0" Babel boxes.
• They appeared very active.
• Example users: A, B, C, D, E

All users in sample B:

• 100% did not use Babel boxes
• 0.3% had one or more Babel boxes (33 users)

Only users with Babel boxes in sample B:

• 9% of sampled users with Babel boxes used more than 6 Babel boxes (3 users).
• 6% of sampled users with Babel boxes used more than 9 Babel boxes (2 users had 11 Babel boxes).
• Example users: A, B

Wikidata Analytics, 15.02.2023