Chapter 1 of 5 · Welcome Messages and Social Cues0:00 / 16:56
Chapters
Meeting purpose
The purpose of the meeting was to discuss various Talk page workflows, user onboarding experiences, and editor interaction tools before breaking into small working groups.
Key takeaways
Welcome messages are most effective when kept simple and signed by an active user who can promptly answer questions, rather than using complex boilerplate templates.
Editors handling vandal warnings and reversions need more targeted notification options so they only receive alerts when a user provides a substantive reply.
Workflows often stem from passively monitoring other user Talk pages or using the New Pages feed to discover articles needing collaborative improvements or translations.
There is a desire to expand acknowledgment features like Thanks, WikiLove, and Barnstars, as well as replace Edward's Bot for project newsletters and invitations.
Topics discussed
The group discussed how users perceive welcome messages, noting that tests show editors often cannot distinguish between automated bots and real humans unless messages are kept simple and conversational.
Speaker A shared personal experiences on the value of early human connections and mobilizing a global network of editors to collaborate on complex article topics.
Speaker D detailed the workflow of reverting vandalism, noting that automated watchlists create noise when editors only want notifications if a user provides a thoughtful response.
Participants analyzed passive Talk page monitoring, where editors discover and improve articles simply by observing conversations on other editors' pages.
Jake proposed enhancements to recognition tools such as the Thanks feature, Barn-O-Meters, and Barnstars, and called for an improved alternative to Edward's Bot for bulk notifications.
Speaker F described utilizing the New Pages feed to find newly created articles, locate relevant sources, and coordinate with original creators on translations.
Next steps
The facilitator will read through the summarized list of discussed workflows shown on the board.
Attendees will divide into small breakout groups to select and examine specific workflows.
Speakers
A
28% · 5m
D
15% · 2m
G
12% · 2m
J
AI guess11% · 2m
E
10% · 2m
C
10% · 2m
B
8% · 1m
F
7% · 1m
Fabrice Florin and Erik Bernhardson for the Wikimedia Foundation, via Wikimedia Commons (CC BY-SA 3.0). Transcoded to H.264. Source