Triple Phinix
Moderation

Reputation and Trust Levels: How Communities Decide Who Can Do What

New accounts can usually read everything and do very little. The abilities are unlocked gradually by a score or a level, and the order in which they unlock is deliberate.

Why permissions are not all granted at once

On a well-run community, a brand-new account can read almost everything and do almost nothing. Posting a link, attaching a file, sending a private message, starting a new thread, editing a shared page: any of these may be locked until the account has done something else first. This is not suspicion of the individual. It is a response to a simple fact about open platforms, which is that accounts are cheap to create and most of the harm one account can do happens in its first hour.

What a reputation system measures

A reputation system turns an account's history into a number, a badge or a level. The inputs differ between platforms, but they are drawn from a short list.

Those inputs are usually collapsed into a small number of tiers rather than exposed as a raw figure, because a tier is easier to explain and easier to attach rules to.

What the tiers typically unlock

Why the gates sit where they do

Each gate corresponds to an abuse that costs somebody something. Link posting is restricted because the cheapest way to exploit an open community is to arrive, drop an address and leave, which is the pattern described in why open sign-ups attract profile spam. Private messaging is restricted because it reaches a person directly and privately, where other members cannot see it and report it. Bulk actions are restricted because an attacker who can register many accounts at once can otherwise outvote or outpost everyone else, a problem known in the literature as a Sybil attack.

Read in that light, the order of unlocking tells you what the operator has been hurt by before.

Where the measure stops being useful

Any number that members can see is a number some of them will play to. Points, streaks and badges are a form of gamification, and they reliably produce more activity — including activity that exists only to raise the score. Short agreeable replies accumulate approval faster than careful ones, so a score tends to measure how much the majority of voters agreed with an account, not how accurate or useful it was.

Scores can also be manufactured. Several accounts controlled by one person, each voting for the others, are sock puppets, and detecting them is a routine part of the work that lands in moderation queues. Operators respond by weighting old votes less, ignoring votes between accounts that always appear together, and capping how much one account can add to another in a day.

Reading a community's levels as a new member

If your level stalls or drops

A level can fall when posts are removed or a report against the account is upheld, and the drop is often the only notice given. Ask which rule was applied rather than asking why the number changed; the answer identifies the behaviour, and the number follows from it. Where restrictions are applied automatically, there is usually a route to ask a person to look again, and it is worth using once rather than repeatedly.

Last reviewed 2026-11-20