Proposal: Add a more explicit incident response lifecycle to the Moderation Policy
还没有人认领这个 Issue。
评估
- 难度
- 5/5
- 预计耗时
- 一周以上
- 新手友好度
- 30/100
- Issue 类型
- 文档
- 描述清晰度
- 基本清楚
- 活跃度
- 活跃
调研方向
从 Moderation-Policy.md 开始,将其中现有的升级、申诉和意图考量部分与链接的 CNCF Incident Resolution Procedures 进行比较。审阅提案中新增的生命周期、利益冲突和禁止报复内容,以及 moderation 圆桌讨论的反馈。达成共识并由一个 PR 相应更新 policy,即视为完成。
由索引模型根据 Issue 内容生成。
描述
Part of the moderation policy review framed by openjs-foundation/summit#511, for discussion at the Moderation Roundtable on October 1 (if not resolves to consensus prior).
[!NOTE]
This proposal has a bit more ambiguity than the others, and should be considered later in the session.
The policy covers content moderation but not larger CoC concerns
The Moderation Policy handles the common cases well: inappropriate posts in a public repo, and spam. Consideration of intent, self-correction opportunities, documentation requirements, and worked scenarios are all defined. Escalation and appeal paths exist.
What's missing is a default or suggested lifecycle for reports that aren't resolved by moderating a single post. Larger-scoped CoC violations have less written guidance for us all to align against. This routinely leaves moderation team members to continually do their best to evaluate many factors at once, and the ambiguity can impact SLAs for moderation and repair.
Some gaps against the policy as written, in my opinion:
- Someone emails
report@nodejs.orgabout an incident, pattern of behavior, a private conversation, or a serious CoC violation. We don't mention acknowledgement to the reporter or the accused. - A report involves conflicting accounts. There is no process for resolving factual disagreements.
- The team needs to check jurisdiction: is this ours, the TSC's (PR #990), or the OpenJS CoC Team's?
- A moderator has a relationship with the accused but is not "directly involved." The recusal requirement only covers direct involvement.
I've done a bit of research and noted that other projects, in particular the CNCF, have better outlined procedures. We can mature here.
Gaps relative to CNCF
The CNCF Incident Resolution Procedures cover these cases. Not all of it applies to a volunteer team, but the structure is useful reference.
| Step | CNCF | Node.js |
|---|---|---|
| Acknowledge receipt | Yes | Reporting channels defined, no acknowledgment step |
| Confirm jurisdiction | Explicit check; transfer or escalate | Not addressed (escalation/appeal paths exist but no intake check) |
| Investigation | Review evidence, interview witnesses | Consideration of intent only |
| Notify the accused | Yes, balances retaliation risk | Collaborators get a chance to self-correct; no general notification step |
| Interim protective measures | Explicit authorization during investigation | Temp blocks exist but not framed as interim measures |
| Conflict of interest | Hard/soft definitions with procedures | Recuse if "directly involved" |
| Communicate results | Defines what to tell reporter and accused | Not addressed |
| No retaliation | Explicit | Not addressed |
The below here would turn into a PR against the Moderation Policy. I include it here for early feedback, and then expect to draft a PR.
Add to the Moderation Policy:
Proposed lifecycle
1. Receive and acknowledge. A moderator acknowledges receipt to the reporter.
2. Confirm jurisdiction. Is this governed by the Node.js CoC? Does the accused's role change who handles it (TSC members per PR #990)? Should this be escalated to the OpenJS CoC Team? If not ours, transfer and notify the reporter.
3. Assess interim protective measures. If there is active harm, the team may hide content, set interaction limits, or temporarily block before the investigation is complete.
[!NOTE]
This is a substantial addition, for us to consider carefully.
4. Investigate. Review available evidence. Contact the reporter for clarification. Interview witnesses where appropriate. Consider intent per existing policy. Write down our evidentiary threshold, per comments in https://github.com/openjs-foundation/cross-project-council/issues/1701#issuecomment-4063822741
5. Notify the accused. Inform them a report was received. Give them an opportunity to respond. Do not disclose the reporter's identity without consent.
6. Decide. Determine action, referencing the enforcement ladder in the Code of Conduct. Document the decision and rationale in nodejs/moderation.
7. Communicate results. Inform the reporter of the outcome. Inform the accused of any action taken and the appeals process.
Conflict of interest
A moderator has a conflict of interest if they are personally involved (reporter, accused, or witness), have a close personal or professional relationship with someone involved, or have another interest that would affect their impartiality. Conflicted moderators disclose and recuse.
No retaliation
Retaliation against someone who reports a CoC concern in good faith or provides information during an investigation is itself a CoC violation.
Not included
This proposal does not cover specific timelines (see SLA proposal), but the two proposals would intertwine and be stewarded by a chair per that proposal.
Prior art
- CNCF Incident Resolution Procedures
- Python PSF Enforcement Procedures
- Contributor Covenant 3.0 "Reporting an Issue" section
Action items
- Feedback period before and during Collab Summit
- PR to add the lifecycle to the Moderation Policy, where more consensus explored
Assisted by: Claude Opus
- 主要语言
- JavaScript
- 星标
- 202
- 派生
- 183
- 平均合并
- 13 天 12 小时
- 30 天内合并 PR
- 2
贡献指南
从这里开始
- 先读完整个 Issue,再读项目的贡献指南。
- 在 Issue 下留言说明你要接手 —— 这能避免两个人做同样的事。
- Fork 仓库,在一个分支上完成修改。
- 提交 Pull Request,并在描述里引用这个 Issue 编号。
nodejs/admin 的其他 Issue
-
难度 1/5 1 小时以内 新手友好度 90/100
-
难度 5/5 一周以上 新手友好度 35/100
-
难度 5/5 一周以上 新手友好度 35/100
-
难度 5/5 一周以上 新手友好度 30/100
-
难度 1/5 1 小时以内 新手友好度 35/100
相似的 Issue
-
难度 2/5 1-3 小时 新手友好度 65/100
antfu-collective/icones#398 ·
-
ECmail.com 未关闭
难度 1/5 1 小时以内 新手友好度 90/100
wesbos/burner-email-providers#554 ·
-
难度 2/5 1-3 小时 新手友好度 65/100
radiantearth/stac-browser#1023 ·
-
难度 2/5 1-3 小时 新手友好度 75/100
-
难度 2/5 1-3 小时 新手友好度 75/100
components-web-app/docs#92 ·