I think theymos made a clarification for us all in other topic;
LoyceV's guide seems reasonable.
The system is for handling trade risk, not for flagging people for good/bad posts/personalities/ideas.
In part, the idea of the system is to organically build up & enforce a community consensus on appropriate trading behavior. However, those parts of the consensus which have less agreement should be more difficult to apply than those parts which have widespread agreement, and also subject to change. Everyone agrees that if Alice promises Bob 1 BTC for $8000 and doesn't pay it, that warrants flags & ratings, and it should be very easy to create these flags and ratings. If Alice promotes something without disclosing that she was paid to do so, and the thing later turns out to be a scam, then 65% of the community will call this highly shady behavior, and 35% will call it not a contractual violation and therefore more-or-less fine; it may be possible to make flags and/or ratings stick, but the people doing so should feel as though they are on less solid ground, and maybe the community consensus on this will shift against them (depending on the exact facts of the case, politicking by interested parties, etc.). I refuse to set down a single "correct" philosophy on ethical behavior, since this would permanently divide & diminish the community, and I am not such a wise philosopher that I feel the moral authority to do so.
For ratings and type-1 flags, proactive scam-hunting is good! But as explained above, if you're acting near the edge of community consensus, it should be more difficult. If the community is not overwhelmingly behind you on your scam hunting, then it's probably going to end up creating more drama, division, paranoia, and tribalism than the possible scam-avoidance benefit is worth.
Ratings
- Leave positive ratings if you actively think that trading with this person is safer than with a random person.
- Leave negative ratings if you actively think that trading with the person is less safe than with a random person.
- Unstable behavior could very occasionally be an acceptable reason for leaving negative trust, but if it looks like you're leaving negative trust due to personal disagreements, then that's inappropriate. Ratings are not for popularity contests, virtue signalling, punishing people for your idea of wrongthink, etc.
- Post-flags, ratings have less impact. It's only an orange number. Some amount of "leave ratings first, ask questions later" may be OK. For example, if you thought that YoBit was a serious ongoing scam, the promotion of which was extremely problematic, then it'd be a sane use of the system to immediately leave negative trust for everyone wearing a YoBit signature. (I don't necessarily endorse this viewpoint or this action: various parts of the issue are highly subjective. But while I wouldn't blame people for excluding someone who did this, I wouldn't call it an abuse of the system.)
- Exercise a lot of forgiveness. People shouldn't be "permanently branded" as a result of small mistakes from which we've all moved past. Oftentimes, people get a rating due to unknowingly acting a bit outside of the community's consensus on appropriate behavior, and such ratings may indeed be appropriate. But if they correct the problem and don't seem likely to do it again, remove the rating or replace it with a neutral. Even if someone refuses to agree with the community consensus (ie. they refuse to back down philosophically), if they're willing to refrain from the behavior, their philosophical difference should not be used to justify a rating. For example, in the YoBit mass-ratings example above, ratings should be immediately removed after the person removes the signature, even if they maintain and continue to argue that they didn't do anything wrong. If someone agrees to "follow 'the law' without agreeing to it", that should be enough.
Flags
- Use flags only for very serious and clear-cut things. They're an expression of ostracizing someone from the community due to serious, provable misconduct or really obvious red flags.
- Use type-1 flags when the message which will be shown to newbies/guests is appropriate: "the creator of this topic displays some red flags which make them high-risk. [...] you should proceed with extreme caution."
- Use type-2 and type-3 flags only if the person is absolutely guilty of contractual violations. Imagine a legal system in which there is no law but contract law, and consider if this person would owe damages.
Trust lists
- If you find someone who has sent accurate trust actions and has no inaccurate/inappropriate trust actions, add them to your trust list. Inclusion in trust lists is a more a mark of useful contributions than your trust in them, though at least a little trust is necessary.
- If you think that someone is not using the trust system appropriately, or if you disagree with some of their subjective determinations, exclude them from your trust list. If bad outcomes happen in DT, this is partly the fault/responsibility of: the bad actors themselves; DT1 who include the bad-actors; DT1 who don't exclude the bad-actors; DT1 who include or don't exclude failing DT1; anyone else who includes failing DT1. While it's best to spend some time trying to fix things at the lower levels before escalating it, it's reasonable to complain to any of those people, as I did regarding Lauda that one time, for example. (Of course, the system itself is probably also imperfect, and that's on me.)