-
Notifications
You must be signed in to change notification settings - Fork 5
Moderation Suggestions
These are a few of our goals:
- Allow Twitarr to continue to exist as an enjoyable means of shipboard communication
- Prevent whack-a-mole situations, where banning a user allows them to create a new account in a way that we can't track
- Provide a way for THO to identify the person behind an account for Code of Conduct violations
- Ensure account creation and system use remains easy enough for novice users
- Minimize the risk of "identity theft" - for example, people voluntarily giving an activation code to another person, or a nefarious person using the names/credentials of other people when creating accounts
- Allow parody accounts
There are three main issues:
- How to moderate posts
- How to validate accounts/users
- How to present it to users/educate users on acceptable behavior
- Volunteers (maybe a subset of Helper Monkeys who are interested in moderating/have moderation experience)
- Volunteer moderators may require multiple "votes" to unhide a post or thread, we can make the numbers configurable
- THO
- Anybody specified by The Home Office
- Admins
- People who built the system and work closely with The Home Office in its operation
The overall goal is to be able to deescalate/hide reasonable conversations that get out of hand (political discussions?), allow people to hide others who they do not want to interact with, and give moderators the ability to remove problem content and users from the platform altogether.
- Develop a set of moderator guidelines/policies. Get THO sign-off on the policies, distribute to all volunteer moderators (and THO/Admin moderators), ensure all moderators are trained on the moderation system and able to follow the guidelines.
- Allow tweets and forum posts to be reported. Reported messages are added to a moderation queue. From there, moderators can take appropriate action.
- Potentially, after some defined number of reports, the reported tweet/forum post is automatically hidden from the timeline/forum.
- Additionally, if multiple tweets or forum posts in the same thread are reported, hide the entire thread.
- For example: 3 reports hides the tweet, 3 reports on 2 separate tweets hides the thread, 3 reports on 2 tweets by the same person hides the person. Numbers will be configurable in an admin console in case tweaks are needed throughout the cruise.
- Then, from the moderation queue, moderators can decide if the post(s) in question are actually a violation (permanently removing the tweet or post, or potentially the entire thread), or if they are fine and can be reinstated. (QUESTION: Once an individual tweet/forum post is reinstated, should future reports on that individual tweet/forum post be ignored?)
- If necessary, the users who made reported post(s) can be warned by the moderator, or banned.
- QUESTION: If a post is reported (once? multiple times?) should we send some sort of an automatic message to the user who made the post, to let them know that someone thinks they are out of line, and that their post will be reviewed by a moderator?
- Grant users the ability to block/ignore other users. The blocked user's posts in the tweet stream, forums, and seamail will no longer be visible to the blocking user.
- Madagascar Protocol (SHUT DOWN EVERYTHING): If for some reason there is a sudden flood of reports on multiple posts in the tweet stream from many users, automatically disable and hide the tweet stream until THO/Admin cleans it up and re-enables it. Similarly, if there is a sudden flood of reports on multiple forum threads from many users, automatically disable and hide the forums until a moderator cleans it up and re-enables it.
- Report/time threshold will be configurable in the admin screens.
- When content has been flagged for review/is in the moderation queue, clicking into it will present a view of the content's context, along with a moderator discussion area. Volunteer moderators can discuss what actions (if any) should be taken regarding the content and the users.
- The screen will indicate if the content has been reported enough to trigger an auto-hide threshold.
- There could be vote buttons, to either permanently remove or reinstate the content. Volunteer moderators could require some configurable number to agree before either action is taken.
- THO/Admin moderators will be able to act unilaterally.
- If the consensus is that a user should be permanently banned, moderators can escalate it. An Admin/THO-level moderator will make the final decision.
- Ideally, banning a user is a last resort. It's for egregious violations, or for users who do not improve their behavior after being warned.
There are a number of ideas for how we could accomplish this. With THO's guidance, we will decide on one (or a combination of a few) to implement - or discard entirely if THO does not want these in place. Our goals here are to allow THO to track serious Code of Conduct violations back to real people in case further action is needed, and to prevent banned users from creating new accounts for further trolling.
- Simple collection of user information during account registration
- Full name, stateroom number. Required, stored on the account, only visible if the user explicitly makes it visible.
- The least secure option - users could enter whatever they want, no real way to validate
- Print and distribute twitarr registration codes
- These could be handed out at the same time as ship cards, or distributed in swag bags
- Doesn't provide a way to positively identify a user, but could help prevent banned users from re-creating accounts
- Scan the barcode on the user's ship card
- Parse and store the number encoded in the barcode (phone camera, or uploaded photo)
- If necessary in the case of a problem user, THO could work with Holland America to identify the person who corresponds to the barcode
- Based on the 2017 barcodes, there is a pattern to the numbers. We would validate that the scanned barcode matches the pattern. May require updating ship-board if the pattern or numbers change from year-to-year
- Could be easy for sophisticated users (such as everyone who goes on the JoCo cruise) to fake out the system.
- Some sort of pre-generated user list
- Would require THO to create/upload a listing of users, or some sort of identifying info that can be validated against
- Probably not going to happen, given THO's reluctance to share anything that can be used to determine cruiser personal info and/or booking counts
- Potentially shut off new user registration after some number of days
- Potentially create an account/profile system - One account per user, but can have multiple identities on the account.
- Each identity would be able to make its own posts and seamails
- There would be an easy mechanism for switching between identities on an account without having to log out/log in
- Useful for people who enjoy parody accounts (examples: Taco Bar, The Oosterdamned, Steve the Devourer, Cruise Director Erin)
- If one identity is banned, the entire account is banned
- On first launch of the CruiseMonkey app once on board, display a "welcome" pop-up that talks about the culture, links to the Code of Conduct text
- During account registration, have a series of statements that users will agree to (and potentially checkboxes) - Keep them short, a high-level overview, so that users actually read them instead of skipping over
- Allow admins/THO/moderators(?) to send warnings/guidance to users who make borderline or problematic posts - attempt to educate them/bring them back in line
THO provides Twitarr with a trained bloom filter that allows name+date-of-birth information to be validated.
Twitarr allows anyone to create accounts but every account, to be created, must have a name and date-of-birth given that the bloom filter says is valid. The name and date of birth are hashed and stored with the account, but otherwise not stored, so Twitarr cannot map accounts to names or dates of birth.
If an account needs to be banned due to code-of-conduct violations, it, and any other accounts associated with that hash, are switched to read-only mode. Users cannot create accounts with a name+date-of-birth that matches a banned hash. The hash will be provided to THO on request; it can be compared by THO to all the known names+date-of-birth pairs that went into creating the bloom filter to determine the account holder's identity.
Admins have the ability to delete posts on the main feed and on the forums.
There is no moderation beyond the ability to delete posts and permanently disable accounts that violate the code of conduct. Behavior that violates the code of conduct is not tolerated and results in bans without additional discussion.