Assuming that, as I have suggested with many other items, the various ‘x of this’ and ‘y of that’ are configurable on a per-instance & per-community basis, we can adjust these settings as needed.
2 Likes
Yes, they should be configurable at both levels. In addition, I recommend some pre-made “levels”, in much the same way that SE has privilege levels for private beta, public beta, established site, and SO. We won’t have those site types, but we’ll still have sites of different sizes, age, and activity level, and we should be able to change one setting rather than a couple dozen when tuning them. (I guess individual ones could still be tuned, but I hope we can define levels in a way that makes that not necessary most of the time.)
3 Likes
In general this sounds reasonable.
My only real objection is to trust level 3, “Constable”. You seem to view this as “moderator in training”, or “casual moderator”. However, it should also include “experienced user”. This level should also be earnable by someone who has shown commitment to the site and demonstrated they understand the norms by writing a lot of well-received posts or earned enough rep (or whatever we call it). These are the kind of people you want to have help policing the site. Right now you can only get there by doing a lot of janitorial work.
This level should be achievable with the equivalent of maybe 1000 - 3000 SE rep. Maybe 100 well-received posts, at least half answers, with less than 5% total posts negatively received. Basically, we’re looking for some metric that this is a user that knows what he’s doing, has provided significant good content, and hasn’t been causing trouble.
-
Rep doesn’t mean that someone is good at moderating. Plenty of people spend more time cleaning and helping others instead of writing answers to get rep and are arguably better at moderating.
-
Flagging and Voting to Close both create janitorial work for others, if someone doesn’t want to do janitorial work, then I don’t see why they should get to create it for others. I have seen plenty of people who never bothered to go through the Review queues complain about quality or questions not being closed and I eventually tuned them out because they weren’t going to vote to close or delete from the review queues which meant that they weren’t actually going to vote to close.
3 Likes
It is a rough guide to how familiar you are with the site and understand its norms. Doing janitorial work doesn’t mean you are good at it, or moderating in general, either.
Flaging and voting is janitorial work.
That’s a very unfair metric, at least for people like me. When I was active, I read most questions directly. I handled things right there as I thought they were required. In other words, I did review a lot of questions, just that this activity was not measured by the system.
All I’m saying is that being a “resident active user” should get you to being level 3 too. There are various ways someone can get familiar enough with the site and be trusted enough to be given those tools. I could even flip this argument around and say that people actively answering questions are in the best position to judge what a good question is.
3 Likes
We missed “create tags” in the list of privileges. I’m inlined to put it at trust level 3, which is where other curation starts happening. Is that too late? For MVP we don’t have tag-management tools (like merge and rename), so it’s probably better to be conservative now and we might drop it a level later. Thoughts?
1 Like
I just checked in DIY SE (the exact reputation numbers for each privilege sometimes vary, but that is good enough) and:
- Create Tags 300
- Create Tag Synonyms 2,500
- Approve Tag Wiki Edits 5,000
So almost anyone can create tags. I’d consider that ~ Trust Level 2. However,
- I think it is way too easy to create tags in SE, resulting in a lot of near-duplicates and other problems
- The initial communities will be modeled on SE sites and can therefore start off with a list of tags based on those sites
and so Trust Level 3 seems reasonable for MVP.
3 Likes
Currently, if people can only raise flags at Trust Level 1, that means that users at Level 0 have no way to get moderator attention if they need it on their post. IMO, Level 0 users should be able to raise flags on their own posts (and possibly on the answers / comments to their own post), up to, say, three flags in a two-day period, more if the flags are marked helpful.
The same with comments - they need to be able to comment on answers to their post and respond to comments on their own post.
8 Likes
“Registered” could also be called “Newcomer”, “Freshman” or the like…
“Registered” would create a misleading implication of a rather good reputation…
All these are just “names”. Which (if not already there) should be customizable strings in a database table - customizable per instance and per community. Inside the software they should be just numbers.
1 Like
I’ve actually been wondering if we even need names. I removed the first few in the functional spec (because there was disagreement and I didn’t see the need) and only kept the ones that involved moderation abilities.
2 Likes
I’m going to add “create tags” at trust level 3 for now. Users can always ask for help in creating tags when needed, as happens on SE.
2 Likes
Maybe this should be given to everyone during young community phase.
We do need names. But the names could default to “Trust Level 0”, “Trust Level 1”, etc.
4 Likes
I admit I can never remember the names for Discourse’s trust levels. “Trust Level 1” might be the best kind of name.
I like the proposed trust levels 0-2. I don’t like levels 3-4. Rather than gating a number of different abilities behind achieving milestones in some other different activities, I’d prefer each one be treated separately, with more use resulting in more ability, like leveling up in an RPG.
- Have 10 new tags be accepted to be allowed to create new tags directly
- Flag-to-close 100 questions which ended up being closed to be allowed to directly vote to close
- Suggest 100 accepted edits to be allowed to edit directly
- After 1000 edits be allowed to mark as a “minor” edit to not bump
- Suggest 10 posts be protected to be allowed to protect directly
- Flag 100 answers as NAA which ended up deleted to be allowed to vote to delete answers
When each ability is tied to demonstrated proficiency in a weaker version of that ability we’ll avoid giving people abilities which they haven’t shown any interest in before.
Of course if something like this was used, then there’d need to be some tricky maths for success ratios etc. If someone gets up to 100 accepted edits after 2000 rejected ones they shouldn’t get the ability. Maybe there’d be a minimum number of edits as well as a minimum approval rate of 80%. And there should be a time limit on these too - only counting those from the last year maybe?
7 Likes
I know this is a popular idea, and it seems reasonable at first glance. However, think about what qualifications you really want in someone you give a privilege to.
We give people privileges so that they can help maintain the site. One thing SE got right was trying to have the community largely moderate the site, with actual moderators stepping in only as exception handlers. From this point of view, you want as many people as possible being able to step in and help with site operation.
However, you also need to have reasonable confidence these people will use the power properly. Someone has to:
- Understand the site norms.
- Know what they're doing.
- Not be malicious.
The question then becomes how to judge someone against these criteria.
#1 comes from experience. More site activity means greater likelihood of understanding the site norms.
#2 also comes from experience, and can be measured by the number of negative actions (as judged by the rest of the community) relative to their whole. If someone wrote 100 answers and 98% had net positive votes, they seem to know what they are doing. Conversely, someone with 30% negative answer or questions doesn’t seem to understand how the site work, and can’t be trusted (yet).
#3 can be measured by number of upheld flags for malicious content relative to their total number of posts. We have to be careful to distinguish truly malicious content here from ordinary bad posts, disagreements as to how PC one should have been, etc. Truly malicious content is rare, and is indicated by spam flags, hate-speech flags, and probably little else.
Note that none of this is specific to particular site activities. Let’s say your total site activity is writing 100 answers, most well received. Does that mean you don’t know how to tell a good question from a bad one? Are you unqualified to edit someone else’s question? Don’t know how to tell whether it’s off topic, spam, or hate-speech? Of course not. In fact, having answered a lot of questions well probably puts you in a better position to judge what you need a good question to be in order to answer it well.
The same logic applies to other actions. If someone has only asked 50 well receive questions, they are actually in a better position than most to judge what a good answer looks like. Anyone that has in some way or another been involved with 100 questions would certainly know whether a comment is appropriate, trying to answer the question, etc.
After you’ve been around the site a bit, you know what a good tag is, when a question should be closed, when to edit and what’s appropriate to edit, when a post should be protected, when a question is off topic, when an answer is good, etc. You don’t need to have actually done all these specific things to be able to judge them.
Let’s keep it simple. Users move up the trust levels with site activity, and that activity is well received. That “well received” needs to be a ratio of total activity, and should probably be weighted towards recent activity. After all, we assume people change their outlook as they get more familiar with the site, else we’d be giving everyone full trust from day one.
2 Likes
Except this is demonstrably false for many people on many SE sites. Gating abilities behind something that isn’t just rep points will help, but we can do more. I also don’t think it will be hard to convey a user’s progress, a simple graph for each privilege could be shown in the profile like the progress graphs for badges are.
3 Likes
Reflecting on this more, its seems that there may be a high level philosophical disagreement about trust levels. It seems some view high trust levels as something you earn, as a reward for performing specific site actions. I believe our view should be:
Trust levels are invitations to help run the site.
2 Likes
I very strongly agree with the last couple of notes from @Olin .
I feel like the “reputation points” model of SE was in some way an effort to basically say “if you do a lot of good things, you gain trust and therefore gain privileges”. The catch is that it is not perfect. We are trying to come up with a perfect alternative - and maybe there isn’t one - short of having a moderator actually individually review every user for every privilege, which of course is not practical.
1 Like
Counterpoint,
There is a user on Outdoors.SE with enough rep to edit,close and reopen. However their contributions were so low quality that I have helped deleted 20-30 of their answers and the rest of their contributions needed substantial editing by others.
Giving them privileges by rep instead of activity means that said user got privileges they wouldn’t have had if they had to earn them individually.
Letting someone with poor grammar edit without oversite seems a recipe for disaster.