TL;DR version:

For the sake of argument below, a user showing negative behaviour on this site is called a "Troll" and a user exhibiting positive behaviour a "Unicorn".

I've been pondering about creating a niceness system that would allow the trolls on this site to handle the trolls, the unicorns handle the unicorns and the neutrals handle everyone if they choose so! :-)


The long version:

Today, we only have a single type of reputation on the system that is based on knowledge only!

I've been thinking of a system where we add an additional kind of rep that is based on a user's "niceness" where everyone starts at 0 where:

  • Past behaviour does not count
  • Any behaviour on the site will have an influence on a user's "niceness" score completely separate from a user's "knowledge" score (today a.k.a "Reputation") and would allow all of the existing site features to have an influence on a user's niceness score as well:

    • spam flags
    • rude/offensive flags
    • Chat bans
    • User bans
    • ...
  • We use a scoring system that is similar to downvoting a question or an answer:

    • downvoting a user's niceness score will give a penalty of -2 to the downvoted user
    • will affect the downvoter with -1 (as the fact of downvoting isn't "nice" neither).
    • Upvoting a niceness score will remain similar to upvoting a question: no score change for the upvoting user, only for the upvoted one.
  • the niceness interactivity is based on 2 signed INTs denominating:

    • User's niceness score
    • User's niceness threshold
    • Allowing an adaptive system to naturally evolve over time depending on MIN(niceness) and MAX(niceness) on a particular site.

Once we have that in place, users can start up/downvoting both according to the question / answer separate from the user's behaviour

Calling a spade, a spade has some additional fringe benefits:

  • Users can choose which kind of users they want to interact with on the site according to the niceness they have (and that is what the niceness threshold is used for):
    ☑ I want to interact with users who are unicorns  :-D
    ☑ I want to interact with users who are nice  :-)
    ☑ I want to interact with users who are neutral  :-|
    ☑ I don't mind interacting with users who are a bit naughty  :-(
    ☑ I don't mind interacting with trolls  ;-(
  • Users that do not completely overlap with a user's individual sense of niceness will show up in light grey if there is only a little difference and be completely invisible to a user including their/your questions/answers/comments!

  • This is not a complete overhaul of the way the SE network works today:

    • Very negative answers are shown in grey already
    • Deleted answers are invisible as well. (except to 10K+ users)
    • Additional storage capacity in the SE network's DB is negligible (~60 MB at the time of this writing, excluding indexes)
    • We have to do something as the sites grow and we're seen as not very "welcoming" to new users
    • the above compartementalises the users according to their behaviour.
    • We could create an additional privilege that allows very trusted users to see the niceness score for all users (50K? halfway to 100K and the t-shirt, hat and stickers?) and to the other users (except mods) these scores would be invisible.

Of course, a troll user limiting itself to interacting only with nice people will put itself in the unfortunate position of seeing very little on this site but then that's their own personal and subjective freedom of choice!

Thoughts? Remarks? Improvements?

P.S. Please don't just downvote: :-) leave a comment at a minimum. A well-funded answer is preferred... 0:-)

  • 2
    Two problems: Ease of abuse of the proposed system (evidenced by the number of times people have abused the existing voting system), and how/where would this be shown? If anything, this would be a hidden aspect, in my opinion, and would require its own privilege levels in order to be less-absued. (20k+ anyone?) – Thomas Ward Jul 15 at 0:24
  • @ThomasWard Thanks for the constructive feed-back. adding that to the Q. – Fabby Jul 15 at 0:56
  • 2
    "once a year we can vote for that year's Troll King and Unicorn Queen" - yeah this is a "no no" because it sounds like targeted discrimination of some sort and that is technically already against policy. – Thomas Ward Jul 15 at 1:02
  • SOHF... Removed... :P ;-) @ThomasWard – Fabby Jul 15 at 1:08
  • 4
    as you are suggesting a new feature, it's usually "not the best idea" to include things that could be misinterpreted - not a 'sense of humor' problem but a 'this could be interpreted literally since it's part of the proposal' problem. 'Tis all. :) goes back to reading more – Thomas Ward Jul 15 at 1:09
  • 1
    thresholds seem to be another issue here - how would we quantitatively determine what fits into each level. Personally I don't see that you can quantify the niceness of a user, as what some users find to be nice others will interpret it to be the opposite. While I understand where the idea for this originates, I don't see it being a useful datapoint because niceness of behavior is a qualitative datapoint that varies based on point of vies, not a quantitative one. – Thomas Ward Jul 15 at 1:31
  • 1
    Further, mods would have no way to really handle targeted voting problems on this either, nor would we have a method to handle abuse of this system, so from the moderation control viewpoint of this feature I foresee it being impossible to moderate or protect from abuse effectively. Unless we require that only extremely well established members of the community can vote on this aspect of a user (then goes back to the 'how do you quantify this' question though) – Thomas Ward Jul 15 at 1:32
  • @ThomasWard I can see you've got a security mindset: you bang on things till they break. :-) I love it! So we'll not only need a niceness score but niceness threshold as well. 2 signed Integers. Let me ponder a bit on how to make this workable for a mod. – Fabby Jul 15 at 13:28
  • @Fabby about the "Thanks", we can just add a little button (i don't know where it must be placed), which works like the upvoted of a comment, and which refered to a "Thanks, you help me a lot" or something like that; and i think this option can be used by all users, with default reputation too because a little thanks is a very good reward for the answerer, or others members which helps a lot too (in some case, it's not only 1 member which solve a problem, but only one is reward with upvote + accepted answer – damadam Jul 16 at 8:28
  • @damadam: We have that already and it's called "upvoting" ;-) – Fabby Jul 16 at 18:20
up vote 7 down vote accepted

This is indeed an interesting concept and I did smile while I was reading it. As tempting as it is to have this system, I'm not sure if it would work, as we're rating behaviour and strikes that are kept private for moderator and community team view only. That is, records of abusive comments, suspensions and other forms of unwanted behaviour.

The site strives to work on improvement, as opposed to punishment and for committed users it works well. Suspensions and bans are designed to give users time to reflect and rate limit the problem behaviour, ultimately weeding out disruptive members who cannot or will not abide by the site's policy. As the site encourages improvement, it's important we keep each user's moderated activity private.

From my moderation activity on other sites, I've noticed (anecdotal evidence only) that there's been an improvement in the behaviour of our long term users since the network has pushed to make changes to the code of conduct. The propensity for your proposed system to shame people is not in the spirit of what the site is trying to achieve. However, I wouldn't let this stop the brainstorming, as it's from posts like these, we grow to understand what the community prefers, helps us discover underlying issues and, ultimately (hopefully), create solutions.

  • Thanks for the funded answer. I agree that the "niceness score" should be kept private, but nowadays mods only see the negative side: you know when you have a troll in your basket, but you cannot see whether the troll is harassing another troll or a unicorn... (pondering your answer and probably editing question with this feed-back in the near future) – Fabby Jul 15 at 17:20
  • Meh... Not worth pursuing. +6 -4 for the Q +4 for the A... – Fabby Jul 18 at 20:47

You must log in to answer this question.

Not the answer you're looking for? Browse other questions tagged .