Goodo Moderation Policy

Draft for legal review — not yet legal advice

This policy explains how Goodo staff review content and accounts, how decisions are made, how people can challenge them, and how we hold ourselves to account. It is the working procedure behind the Community Guidelines. The most serious safety cases follow the Safeguarding and Incident Response Policy.

1. Principles

1. Safety first, fairness always. We act quickly to stop harm, and we give people a way to challenge our decisions.
2. Humans decide. Automated tools find and sort. A trained person makes the decision to remove a badge, restrict, suspend or ban.
3. Be consistent. Similar cases get similar outcomes. Reasons are written down.
4. Be proportionate. Use the smallest step that keeps people safe, and escalate when needed.
5. Be transparent. People are told what happened and why, in plain language. We publish figures about our work.
6. Protect privacy. Staff see only what they need. Private messages are read only when reported or flagged.
7. Look after the people doing the work (section 10).

2. Who moderates

• Goodo staff (moderators and admins) review the queues. Only staff can remove badges, restrict, suspend or ban accounts, or read a reported or flagged conversation.
• Group moderators (people who created a group, or who were given the role by it) can approve or remove members, ban members from their group, and remove content in their group. They must follow the Community Guidelines. They cannot override staff. Staff can remove a group moderator, and can act on any group, event or content. Group moderators are volunteers: we give them guidance, a clear way to escalate to staff, and support (section 10).
• Workers lounge moderators are verified workers. Their rules are the same, plus: never identify a participant, even indirectly.

3. The queues

Staff work from four queues. Every queue item is assigned, timestamped and logged.

An appeals list sits alongside (section 7).

Priority. Reviewers set priority from what they read, not only the reason picked by the reporter. Raise priority when in doubt. P1 and P2 work is done at all hours (see the Safeguarding Policy for cover arrangements).

Triage steps:

1. Open the item. Read the content in context.
2. Confirm or change the priority.
3. Check for danger. If someone may be in danger, follow the Safeguarding Policy.
4. Decide using the matrix (section 4).
5. Record the action and a clear written reason. Link it to the report.
6. Notify people (section 6).

4. Decision matrix by report reason

"Strike", "restrict", "suspend" and so on are the enforcement steps defined in section 5. Where a row says "per ladder", use the ladder: warn for first minor breach, strike for a clear breach, and escalate for repeats. A row's "at once" actions are for severe cases and can be used without earlier steps.

What can be acted on:

5. Enforcement ladder and strikes

Other actions: remove_content, restore_content, reinstate (end a restriction or suspension), revoke_badge, and shadow_limit (reducing how widely a piece of content is shown, used only for spam and borderline content, and the person is told).

Default thresholds (starting points that reviewers can adjust with a written reason) [numbers to be confirmed]:

• 2 active strikes: restrict for 7 days.
• 3 active strikes: restrict for 30 days or suspend for 14 days.
• A further breach while restricted or after a suspension: suspend 30 days; ban review.
• A fourth active strike: ban review.

Skip the ladder (suspend or ban immediately) for: sexual exploitation or grooming; anything sexualising a minor; credible threats; exploitation of a person with disability; fraud; false identity or false credentials; evading a ban.

Badge revocation (revoke_badge) is separate from the ladder. It is used when someone no longer meets a badge requirement because of conduct, or they gave false information, or they seriously or repeatedly breached the guidelines in their role as a provider. Badges that lapse because a credential expired are removed automatically and this is not a penalty. See the Verification Policy.

Approvals. A ban or a conduct-based badge revocation must be approved by a second staff member or the safeguarding lead before it takes effect, except in an emergency when it can be applied first and approved afterwards. [To be confirmed.]

Evading a ban. We use identity-check references and account signals to stop banned people returning, and do not store more data than that needs.

Escalation to the NDIS Commission, police and other bodies is decided as in the Safeguarding Policy.

6. Telling people what happened

The person affected by a decision gets a notice in the app (and email if they have one) that says, in plain language:

• what we did (removed, warned, restricted, etc.)
• what content or behaviour it was about
• which guideline it breaks
• how long it lasts, if it is for a set time
• how to appeal, and by when (7 days)
• where to get help, if relevant

We do not tell someone if doing so would put a person at risk or interfere with a police or regulator investigation.

The reporter is told when we have looked at their report and whether we acted, without sharing private details about the other person.

The target of abuse (if different from the reporter) is told what we did where it is safe to do so.

False and abusive reports. Reports made in good faith are never penalised, even if we decide there was no breach. Repeated false reports made to harass someone are treated as harassment.

Accessible notices. Notices use plain language, work with screen readers, and are available in Easy Read.

7. Appeals

• Who can appeal: anyone who receives a warning, strike, content removal, restriction, suspension, ban, badge revocation or credential rejection. A reporter can also ask us to look again if we dismissed their report.
• How: in the app, by tapping Appeal in the notice, or by email to [appeals email TBC]. You can ask someone to help you, such as an advocate, nominee or support person, and we will deal with them if you tell us to.
• Time limit: within 7 days of the notice. We will consider late appeals where there is a good reason, such as hospitalisation or lack of access.
• Who reviews: a different staff member from the one who made the original decision. For bans and conduct-based badge revocations, the reviewer is a senior member of the team.
• How long: we aim to give an answer within 7 days of receiving the appeal. [To be confirmed.]
• While the appeal is pending: content removed for safety stays removed; other actions may be paused where it is safe to do so. For credential rejections we keep the evidence file until the appeal is decided (see the Verification Policy).
• Outcomes: upheld, changed (for example a lesser step), or overturned (content restored with restore_content, account reinstated with reinstate, strike removed, badge restored). The decision and reasons are given in writing.
• Overturned decisions are used to improve guidance and training.
• After an appeal: if you are still not happy, you can complain to the director at [complaints email TBC]. You may also contact relevant external bodies, such as the eSafety Commissioner or the OAIC, depending on the subject.

8. Records and accountability

• Every action is recorded with who did it, what, when, to what, why and which report it responds to. Actions on credentials and badges are also recorded in the audit log.
• Strikes expire after 12 months but the record is kept for the period in the Data Retention Schedule.
• Staff cannot change or delete the record of their own actions.
• A sample of decisions is reviewed each month by a second person for accuracy and consistency, and for bias.
• Staff do not moderate content from people they know, or in which they have a personal or financial interest.

9. Transparency reporting

We commit to publishing a transparency report every six months from public launch, in plain language and an accessible format. During the invite-only phase we produce the same figures internally each quarter and share a summary with our co-design group. Each report will include:

• reports received, by reason and by what they targeted
• median and 90th percentile time to first review, by priority, and the share that met the targets in this policy
• content removed, and accounts warned, struck, restricted, suspended and banned
• badges revoked for conduct, and credentials rejected or expired
• the share of actions started by automated tools compared with users, and how often a human reversed an automated decision
• appeals received, decided, and the share overturned
• referrals to the NDIS Quality and Safeguards Commission, police, eSafety and other bodies (counts only)
• legal requests and notices from authorities
• changes to this policy and the Community Guidelines
• moderation team size and cover hours

Counts below five are shown as "fewer than 5" so no one can be identified. We will also provide any reports required of us under the Online Safety Act.

10. Moderator wellbeing

Moderators see distressing content, and many may have lived experience of disability or caring. We are responsible for their safety.

• Exposure limits. No one reviews harmful content for more than [4] hours in any day, with breaks. Work is rotated between queues.
• Protective defaults. Images and video thumbnails in review are blurred, audio is muted by default and message text is collapsed, until the reviewer chooses to open them.
• Severe material. Child sexual abuse material is not routinely viewed by staff. Automated matching and the escalation path to the AFP and eSafety are used, and no one is required to view it.
• Opt out. Moderators can say they cannot review a category at that time, without penalty.
• Never alone. P1 and P2 cases always have a second person to call. No one carries the on-call phone without backup.
• Support. Access to confidential counselling, with a provider trained in trauma and vicarious trauma, paid for by us. Debrief after serious cases. Regular check-ins. Paid time to recover.
• No unhealthy targets. We do not set speed quotas that push people to rush or skip breaks. Speed targets in this policy are service levels for the whole team.
• Training. Disability awareness and rights, trauma-informed practice, privacy, bias, cultural safety for Aboriginal and Torres Strait Islander people, and the NDIS Code of Conduct. Refreshed every year.
• Group moderators get a short guide, a "how to handle hard moments" sheet, and a direct route to staff for anything serious. They are never expected to deal with self-harm, abuse or exploitation alone.
• Review. We ask our moderators about their wellbeing at least every quarter and act on what they say.

11. Review of this policy

This policy is reviewed at least once a year, after every serious incident, and when features change. Changes to the matrix and thresholds are approved by the director and noted in the next transparency report.