Incident Reporting System/Updates/uk
August 2025: Conclusions from the conversations and next steps
[edit]As we previously posted, we planned conversations with a few groups (Stewards, Arbitration Committees, and U4C members) to hear their comments on the development of the Incident Reporting System. Based on what we heard, we are designing a flow for reporting incidents that are not emergencies.
The new flow needs to be legally and community-acceptable, and designed to work across all wikis. It will reflect and complement, rather than duplicate or disrupt, existing practices of different communities. In consequence, we have decided to make it configurable via Community Configuration.
We are also exploring an improved list of violation types, and customizable "Get Support" pages based on the type of incident. In parallel, we are researching different wiki communities to better understand existing practices. This is to make the default setting reflect the typical needs well, and to make it address compliance requirements.
Lastly, we wanted to note here that as the previous Trust and Safety Product team, we have been merged with the former Product Security and formed Product Safety and Integrity. People previously involved in the IRS project have not changed their roles, including the owner (Madalina). However, due to this merger, Madalina now has support from Eric (the owner of the broader Safety and Security area) and Szymon (the Movement Communications person for the Safety and Security area). We encourage you to contact any of us if you'd like to talk about IRS. 🖖
June 2025: Stakeholder conversations
[edit]To ensure that future development is aligned with community needs and what is feasible, we have paused deployments, and will be conducting conversations with key functionary groups (Stewards, ArbComs, and U4C members) to address concerns and gather input on the IRS's further development.
We will have structured calls with the groups in May-June 2025, with a focus on identifying major concerns and achieving a shared vision for the IRS.
December 2024: Incident Notification System is now available on Portuguese Wikipedia
[edit]
We implemented the Incident Reporting System on Portuguese Wikipedia on Wednesday, December 11, 2024. Users there can quickly report emergency incidents directly to the Wikimedia Foundation Trust and Safety team or be directed to submit community-managed reports to the appropriate community pages - all from a single, centralized platform.
To report an incident, users can access the reporting functionality in the Discussion Tools menu by clicking the ellipsis menu or the overflow menu (...) next to each comment, and then click Report.
If you have any comments, please share them on the project talk page.
December 2024: Upcoming pilot
[edit]We have released the first version of the Incident Reporting System which is now available on Beta. Visit a talk page, choose any comment and click the menu to access the reporting link.
In response to your comments, we have made some updates to the designs. You can check the latest designs here:
Emergency workflow
Non-emergency workflow
We will be enabling the Incident Reporting System on testwiki on December 9 for another round of testing and fixing bugs. If everything goes well we will be enabling the system on Portuguese Wikipedia likely in the week of 10 December.
We will share another update once we are live on testwiki and then on Portuguese Wikipedia.
November 2024: Improvements to the features
[edit]Our next step will be to develop a version that we can test on a few pilot wikis. We have made improvements to the design:
- Emergency incidents - Incidents relating to immediate threats of physical harm. These incidents need to be handled by the Wikimedia Foundation Trust and Safety team responding to emergency cases.
- Non-emergency incidents - Incidents that are not immediate threats of harm, for example bullying, sexual harassment and other unacceptable user behavior. These incidents are handled through local community processes.
The Emergency flow will direct users to file a report that will be sent to the Trust and Safety team.
The Non-emergency flow will direct users to reach out to their local community for support, as outlined in community policies. This will be done through a “Get support” page that will contain links and information specific to each community. The intention is to have configuration options on this page so that each local community can add the relevant links as necessary.
We have some screenshots to demonstrate how this might work. For the next deployment the main focus is to test the emergency flow in production.
Emergency flow:
-
IRS emergency report button
-
Design showing how users can describe an emergency incident when reporting
-
Design showing how users can add additional information when reporting immediate threat of harm
-
Design showing confirmation users see after reporting immediate threat of harm
Non-emergency user flow:
-
IRS non-emergency report button
-
Design showing how users can describe an incident when reporting
-
Design showing the types of unacceptable behaviour users will see when reporting
-
Design showing how users can get support for unacceptable behaviour on the wikis
-
Design showing how users can get support for unacceptable behaviour on the wikis
We’d love to hear your thoughts on the current designs! Please comment on the discussion page.
July 2024: User testing summary
[edit]In March 2024, the Trust & Safety Product team conducted user testing of the Minimum Viable Product (MVP) of the Incident Reporting System to learn if users know where to go to report an emergency incident, and if the flow makes sense and feels intuitive.
We learned the following:
- During user testing, all participants found the entry point to report an incident and the current flow is well understood.
- There was some confusion over two of the reporting options: “someone might cause self-harm” and “public harm threatening message”.
- Two participants also made assumptions about the system being automated. One participant was concerned about automation and wanted a human response, whereas the other participant felt assured by the idea it would check if the abuser had any past history of threats and offences, and delete the offensive comment accordingly.
- All participants expected a timely response (an average of 2-3 days) after submitting a report.
Test the Incident Reporting System in Beta – November 2023
[edit]We invite editors to test the initial version for the Incident Reporting System. It makes it possible file a report from the talk page where an incident occurs. This version is for learning about filing reports to a private email address (e.g., emergency
wikimedia.org or an admin group). It doesn't cover all scenarios, like reporting to a public noticeboard. We need your opinions to see if this approach is effective.
To test:
1. Visit any talk namespace page on Wikipedia in Beta that contains discussions. We have sample talk pages available at User talk:Testing and Talk:African Wild Dog you can use and log in.
2. Next, click on the overflow button (vertical ellipsis) near the Reply link of any comment to open the overflow menu and click Report (see slide 1). You can also use the Report link in the Tools menu (see slide 2).
-
Slide 1 -
Slide 2
3. Proceed to file a report, fill the form and submit. An email will be sent to the Trust and Safety Product team, who will be the only ones to see your report. Please note this is a test and so do not use it to report real incidents.
4. As you test, ponder these questions:
- What do you think about this reporting process? Especially what you like/don’t like about it?
- If you are familiar with extensions, how would you feel about having this on your wiki as an extension?
- Which issues have we missed at this initial reporting stage?
5. Following your test, please leave your feedback on the talk page.
If you can't find the overflow menu or Report links, or the form fails to submit, please ensure that:
- You have logged in
- Your Beta account email address is confirmed
- Your account has been created for over 3 hours and you have at least 1 edit.
- You have enabled DiscussionTools because the MTP is integrated with DiscussionTools
If DiscussionTools doesn’t load, a report can be filed from the Tools menu.
If you can't file a second report, please note that there is a limit of 1 report per day for non-confirmed users and 5 reports per day for autoconfirmed users. These requirements before testing help to reduce the possibility of malicious users abusing the system.
Sharing incident reporting research findings – September 2023
[edit]
We have completed research about harassment on selected pilot wikis. The research, which started in early 2023, studied the Indonesian and Korean Wikipedias to understand harassment, how harassment is reported and how responders to reports go about their work. In summary, we received valuable insights on the improvements needed for both onwiki and offwiki incident reporting. We also learned more about the communities' needs, which can be used as valuable input for the Incident Reporting tool.
Чотири оновлення проєкту «Повідомлення про інциденти» – 27 липня 2023
[edit]Hello everyone! For the past couple of months the Trust and Safety Product team has been working on finalising Phase 1 of the Incident Reporting System project.
The purpose of this phase was to define possible product direction and scope of the project with your feedback. We now have a better understanding of what to do next.
1. We are renaming the project as Incident Reporting System
[edit]The project is now known as the Incident Reporting System, with the word "Private" removed.
In the context of harassment and the UCoC the word “Private” refers to respecting community members’ privacy and ensuring their safety. It does not mean that all phases of reporting will be confidential.
We have received feedback that this term is confusing and can be difficult to translate in other languages. Therefore we are removing it.
2. We have some feedback from researching some pilot communities
[edit]We are conducting research on harassment in the Indonesian and Korean Wikipedia communities. With their feedback, we have been able to document how users in these communities report harassment and created maps out of the information. These maps represent, to the best of our knowledge, how community members on both wikis currently report incidents of harassment and abuse.
-
How Korean Wikipedia reports harassment
-
How Indonesian Wikipedia reports harassment
If you have any feedback on these maps, you can give it on the talkpage.
3. We have updated the project’s overview
[edit]What we want to build moving forward
- The Trust & Safety Tools team will be developing an extension for reporting incidents/UCoC violations.
- The extension is intended to be configurable, communities should be able to adapt it to their local processes
- The extension name is ReportIncident
- The purpose of the extension is to:
- Facilitate the filing of reports about various types of UCoC violations by Wikimedians
- Route those reports to the appropriate entities that will need to process them
- Facilitate the filing of reliable reports and filter out/redirect the unactionable ones.
- Facilitate the filing of both private (e.g. to an email address) as well as public (e.g. on-wiki to an Admin noticeboard) reports according to local processes.
- Extension is intended to be incident agnostic (ability to support the reporting of different types of incidents)
What we won’t be doing
- The system is intended for reporting and routing only, we will not be dealing with processing reports
- The system is intended for incidents with regards to UCoC violations. We will not use this for other type of requests (such as technical support requests, account access etc)
- The system is NOT meant to replace existing processes on wikis. Our purpose is to make it easier to follow existing processes.
4. We have the first iteration of the reporting extension ReportIncident
[edit]This is just an initial version with very minimal basic features, to get us started. This is not a finalized product. |
In November last year we talked about how we should start small with a very limited scope, so for our first iteration we thought about creating a very basic experience.
What’s included in this initial iteration?
- Ability to report from User Talk page
- Report a topic header
- Report a comment
- Ability to complete a basic form and submit
- The report will be sent to an email address (a dummy email for testing purposes).
Designs
The first version of the MTP (minimum testable product) will let a Wikimedian report an abusive topic header or comment on a talk page. Here are the designs.
- On Mobile
- On Desktop
Implementing Designs – What’s next
[edit]The Trust and Safety Product team is now working on developing these initial designs as an MTP, a proof of concept that will be deployed to Beta-cluster and tested internally. The purpose of this is to assess technical viability. If everything goes well the next step is to deploy to test.wikimedia.org for usability testing and feedback.
Looking forward to your feedback about this first iteration on the talk page!
November 2022
[edit]Our main goal for the past couple of months was to understand the problem space and understand what people are struggling with, what they need, and their expectations around this project. We did this by:
- Reviewing and synthesizing harassment research, surveys and other relevant documentation (going back to 2013)
- Having user interviews with volunteers who have experienced or witnessed harassment on Wikipedia
- Having discussions with Staff members, UCoC drafting committee and wiki functionaries.
Our purpose was to identify priorities, scope and a possible product direction.
Findings and next steps
[edit]Focus on Safety
[edit]The recommendation from the Movement Strategy discussions is to provide for safety and inclusion within the communities. As our ultimate goal is for people to feel safe when participating in Wikimedia projects, we will use this as the guiding principle for what to focus on in the minimum viable product (MVP).
Project Approach: start small
[edit]There are a lot of things to take in consideration when thinking about this project.
- Many types of Users: reporter, responder, observer, accused, monitor
- Many Use cases: doxing, abuse of power, content violations, security breaches, legal issues etc.
- A lot of Complexities: admins as harassers, off wiki harassment, government interference etc.
This project will grow and become more complex over time. So we need to start really small, with a very limited scope before we dive into anything more complex.
Focus on two types of users
[edit]We have identified a few different types of users:
- Reporters: Users who have experienced harassment, and are filing a report.
- Responders: Users who receive the report, and want to help.
- Accused: The users who are named in the report.
- Monitor: People who are interested in tracking the progress of reports, to understand the problem better or to ensure that people are treated properly.
Since we want to start small, we will focus on reporters and responders first.
MVP Approach (Short-term)
[edit]The way we would like to approach this is to build something small that will help us figure out whether the basic experience actually works.
Principles of the MVP:
- We will design for and test and release on a few pilot wikis
- Since our goal is to address safety we are going to focus only on 3.1 (Harassment) in the UCoC.
- We will explore a basic experience for two user groups only:
- Reporters will understand how to file a report, and feel comfortable enough to complete the report process.
- Responders will receive clear reports, giving them the information that they need in order to understand the problem.
- MVP will connect to current systems as they are (we are not changing any existing processes)
This experiment should also help us explore and answer some important questions and learn things as we go:
- Entry points (where reporting starts) – what are they, should we have one or more?
- Users – do people easily discover the entry point? What do they think will happen when they engage it?
- Scale – can we do this at scale? Will we overwhelm the responders? etc.
- Data – can we build something that will help us collect the data we need in order to make decisions? What can we measure to know we’re moving in the right direction?
What we are not doing (yet)
[edit]The idea is to start with a really small scope, try a few things and learn as we go. Therefore we need to be very clear about what we are not going to do yet:
- We are not solving for bad admins and/or other complex use cases
- We are not fixing existing flawed processes
- Not everything in the UCoC is about safety but we are focusing only on safety
- Agnostic reporting – we cannot do this without validating a basic reporting experience works with a specific type of incident
What happens after the MVP (long-term)
[edit]We have some ideas about v2 and v3 but we want to experiment with an MVP first and see how people feel about it. What we learn now will be useful to make decisions about future versions.
Some v2 and v3 ideas include:
- Private reporting (creating a private space for reporters and responders to interact)
- Escalation (having the ability to route cases to a different entity for further support)
In order to explore these two ideas we need to ensure the basic/core experience actually works. If it does we will build on top of it.
Discussions points
[edit]- What do you think about this approach?
- What scares/concerns you about this project?
Looking forward to your feedback on the talk page!
September 2022
[edit]We have been collecting feedback, reading through existing documentation and conducting interviews in order to better understand the problem space and identify critical questions we need to answer. We are currently synthesising the information we have collected in an effort to start defining a more clear scope for the project. It is a lot of information to go through so this might take a while, there's so many things we need to learn!