Implementing Community Moderation

Question

i am building a social network platform, as a platform to learn web development. i have some questions on how should community moderation be implemented properly.

flagging

should it be totally automated? eg. if a post is flagged xx times, it will be closed/deleted?

some problems i see is that, if somehow a grp of users decide to wreck havoc, they can just go arnd flagging everything? of course hopefully the community dont have such many users?

a more non-destructive method is hide the post by default, and have users click to see the flagged post. this will happen until trusted moderators delete the post?

closing/deleting posts

i think this should be reserved for trusted moderators/admins? or will it be too much work?

how is it implemented commonly? like in Forrst or stack overflow?

Answer 1

The short answer is: It depends. For example on hostility and spirit of your community. Your authentication and reputation mechanisms. On incentives to behave or misbehave. On the damage done, and if it is permanent or reversible. The amount of moderators or engaged community members willing to help with such tasks.

It is generally a good thing to assume trust and cooperation in your community, but to keep every action recorded and reversible, and thus every person accountable for his actions while minimizing the damage that can be done.

Answer 2

It's a tricky balance, but some form of mechanism based on the reputation of the person flagging the post might be a reasonable "one size fits all" solution. (e.g.: A "standard" user can affect a post's score by X, whereas a moderator can affect by Y, an admin Z, etc.)

That said, I'd be tempted to at least add some form of IP address logging, etc. (if this isn't handled during the flagging stage) if you decide to go for any form of automatic closure. (This should at least prevent one form of "group attack".)

In essence, whatever administration tools and utilities you can build in at the beginning will most likely pay dividends later on. Whilst it might not be that relevant in this instance (as it's for the purposes of learning, I'm guessing it won't see significant deployment for a while, if ever) it's a good heuristic to follow.