[go: up one dir, main page]

Page MenuHomePhabricator

[SPIKE] Identify the "inputs" that can be used/adjusted to determine when the reference check will become activated
Closed, ResolvedPublicSpike

Description

This task involves the research of identifying the inputs that we can consider using/adjusting/tuning to determine when the reference check will become activated.

Where "inputs" in this context could mean things like: bytes added, time since last keystroke, paragraphs added, sections added, etc.

And where "we" in this context could mean both the Editing Team, and in the future, volunteers, perhaps by way of T323811.

Related Objects

Event Timeline

ppelberg moved this task from Untriaged to Upcoming on the Editing-team board.
ppelberg moved this task from To Triage to Triaged on the VisualEditor board.
VPuffetMichel renamed this task from [SPIKE] Identify the "inputs" that can be used/adjusted to determine when the reference check will become activated to [edit check] [SPIKE] Identify the "inputs" that can be used/adjusted to determine when the reference check will become activated.Dec 8 2022, 1:36 PM
ppelberg renamed this task from [edit check] [SPIKE] Identify the "inputs" that can be used/adjusted to determine when the reference check will become activated to [SPIKE] Identify the "inputs" that can be used/adjusted to determine when the reference check will become activated.Dec 12 2022, 5:55 PM
Restricted Application changed the subtype of this task from "Task" to "Spike". · View Herald TranscriptJan 3 2023, 4:38 PM

DECIDED
We're going to prioritize work on this task once we complete work on T323145 and T326856.

Reason being: through T323145 and T326856, we'll learn what "inputs" are available to us to tune/configure.

Meeting notes from today:

  • This happens later once we have an initial heuristic (T324730)

Alright, a couple of updates...

Updates

  1. This week, the Editing Team will start collaborating with the Growth Team to figure out the extent to which we might use Community Configuration to provide the kind of configuration experience @Xaosflux described on-wiki [i] and @Mathglot described in T327330#8607428 and T327959#8954388.
  2. I've added "⭐️" to the configuration options in the task description I'm proposing that we initially expose on-wiki:
    • A) Account attributes for whom Edit Check will be active
    • B) Where the citations Edit Check automatically inserts will be placed: before/after periods
    • C) What categories Edit Check is not activated on pages within
    • D) Edits in what sections Edit Check is not activated within. Note: this proposal will not be final until the Editing and Growth engineering teams reviews it. See "Thinking" below for rationale.

Thinking

The four configuration options proposed above are grounded in the initial principles that surfaced in the Editing Team meeting on 12 July 2023:

  • Impact: configurations ought to contribute to volunteers, on a per-project basis, being able to adjust Edit Check in ways that meaningfully impact things like the rate of false positives and false negatives and the automated actions Edit Check makes without depending on the Editing Team.
  • Awareness: volunteers ought to be equipped with the information they need to independently evaluate the impact of the configuration adjustments they make so that they can iterate towards a state where the feature is causing behavior that aligns with projects's stated goals, policies, and conventions.
  • Distinctness [ii]: configuration options should not cause volunteers to become confused about the distinction between the various places on-wiki where volunteers could potentially configure how Edit Check works.

i. "Having rules live on-wiki (like at MediaWiki:EditCheckRules.json or the like) would help projects adjust to their needs in a lighter weight format than the abusefilter (inspired by the newcomer tasks setup)."
ii. There is likely a more accurate word than "Distinctness." Tho, it's not currently coming to mind.

ppelberg claimed this task.

Per T324731#9024902, this work ended up happening in T327959 (see the task description's Use cases section).