My thoughts about ‘reputation’ or ‘karma’


(Jeff Atwood) #27

If you don’t read, you aren’t listening, and if you aren’t listening, that’s not much of a conversation… is it?


#28

I think there’s a misunderstanding there. I do read, but I don’t also wonder about the various user stats while reading.

I’m listening plenty, but I’m not hearing as much meaning as is possible.


(Jeff Atwood) #29

I suggest cough reading

https://blog.codinghorror.com/because-reading-is-fundamental-2/


#30

You cough assume that I haven’t already. :slight_smile:

My previous posts here were by no means advocating post-count as a metric. And while I do appreciate the value of reading-time, it alone isn’t sufficient to represent a member’s overall value to the community. In my humble opinion anyway.


(Mittineague) #31

So what is being suggested is a popularity metric?


(Evgeny) #32

I myself will not use karma. The Q&A system has both pluses and minuses. I like another function, for example, as in Quora - follow the participants. But this is only my personal opinion. And about karma: you can also make a plugin for those who want to have a ballroom system on the site. Evaluation is easy. Much more work on the history of fees. It must be transparent. And this change, which will have huge consequences.

I have a Q & A site based on WeCenter. He uses the ball system. Until I started using it, I had no idea how it would be otherwise. Many attempts at deception, scandals, etc. And at the moment, can there be more rewards based on reading? Hmm …

Medals, awards - discrete. Karma, scores (maybe this is the reason) is an ongoing process. Reading is analogous to a continuous process (it is there). Can try to take advantage of this better?


(Sam Saffron) #33

I am having trouble figuring out what you are advocating though? :thinking: a metric that somehow captures how useful a member is and keeps on going up? I can’t even think of how we would synthesise something here that would not lead to many problems.

Let’s try

Reputation = Number of Likes Given * X + Number of Posts made * Y + Time read * Z

Where do you even go plugging in numbers there? How do you not lead to problems where users just post randomly to increase the “reputation” or fake “read” to add more reputation?

I do not believe this can be done though… and I don’t see any concrete proposal here except for “it would be awesome if”.


(Adrianbblk) #34

Agree, I do that on reddit


(Mittineague) #35

I’ve thought a little about how I might go about creating a “valuable member” metric by consolidating various other numbers. eg. read time, number of visits, post count, number of Likes received etc. into a single number. But when it came to weighting I struggled to think of how an algorithm might look. multiply, add, divide, and by what numbers?

It feels like it would not be a one size-fits-all thing. What I might consider to be indicative of value and how much value I think various factors have is very unlikely to be the same as everyone. Too many possible ways to combine them and too dependent on what context.

That said, I see no reason why a community should not create a custom badge query based on what they think is important to their community.


#36

The thread here isn’t regarding concrete proposals, it’s regarding “thoughts” about reputation and karma. I posted my thoughts.

I admit I don’t have a specific proposal that is better than what has already been suggested, but I also don’t agree with the prevailing consensus that “reputation systems are bad”, that kills this idea before it can get any legs. In this thread in particular, @codinghorror’s response that its advocates are conflating Discourse’s usage with Q&A systems doesn’t seem like a fair assessment to me, hence my TL;DR.

I agree there are problems, but there are also problems with the badge system. Any system of gamification will be a target of gaming; it’s inevitable. What I can’t understand is why reputation is so adamantly shot down while Discourse’s badges are presented as an example of gamification done correctly, as if badges don’t also have “many problems”.

My suggestion here isn’t for a specific implementation, it’s simply for a reevaluation of the idea’s merit. After all, any suggested implementation is going to be shot down so long as reputation systems’ benefits are so widely ignored because of the problems.


(Evgeny) #38

Just an attempt, an experiment, more visually show. Maybe it can be useful for thinking.

We will change it still 100 times.

  1. trust_level (there is a repetition with a level of trust. It already exists on the page.)

  2. badge_id; ( Licensed)

  3. recent_time_read;

I do not like what I did. Because I want to show more progress than the final result (badges).

I agree, so that time reading is an important indicator. But I think we can not use (time_read) only (recent_time_read), because new users will never catch up with the old timers on the forum.


(Erlend Sogge Heggen) #39

I stand by what I’ve said about this in prior discussions:

I’ve yet to see a large scale reputation system, on the internet or elsewhere, that successfully incentivises healthy community dynamics – which is probably why they always face ridicule in pop entertainment.

And the very real-life applications of such systems are downright terrifying.

More often than not such systems only help to amplify innate human flaws, such as us/them thinking.

The fact that our badge system could still use some improvements is not a good reason to implement a new system with a very bad track record. I tend to agree with you when it comes to our “First X” badges. I’ve always been of the opinion that Discourse should focus much more on human-awarded badges rather than algorithmic ones.


#40

Yep. Totally agree.

A simple illustration is someone who seeks to win the support of a large population of users by expressing a known popular belief. That’s a pretty awful dynamic and tends to polarise the debate. It happens in political discussion all the time and it simply doesn’t move the argument on and any feature that encourages too much should be deprecated. A ‘like’ button is enough! :slight_smile:

Perhaps this kind of thing works better in a ‘fact based’ universe where there are generally more definitive ‘right/wrong’ answers, but in a space where things are more grey and there are lots of unknowns, less obvious ‘right answers’, you can see it being less helpful and possibly detrimental.

I think in summary what I’m trying to say is:

If we are trying to get to some ‘truth’, we should be promoting all points of view, not just encouraging the popular ones. In many cases it turns out the truth is not the popular opinion (flat/round earth?) - and in any case people will learn more by being exposed to a variety of different arguments. Rewarding popularity is too simplistic.


(Evgeny) #41

I fully agree with this.

And we use the karma, we do not want. And if more clearly to represent progress?

  location = data.user.location;
  bio_excerpt = data.user.bio_excerpt;
  signature_url = data.user.custom_fields.signature_url;
  date_of_birth = data.user.date_of_birth;
   
  if (location) {  my_loc = '25';...
  if (bio_excerpt) { my_bio = '25';...
  if (signature_url) { my_sig = '25';...
  if (date_of_birth) { my_bir = '25';...

<div class="progress-bar"><span class="progress-bar-fill" style="width:'+ progress +'%"></span></div>

  1. Filled profile.
  2. Reaching reputation.

But this has the final result. The profile will be filled. The level of trust will be obtained. What else can we use in the Discourse for progress throughout the work. Visual progress.

Maybe many just want it. Offering karma, but in fact, want to see more visual progress in visiting the forum?


(Dave McClure) #42

I like this idea. What would the ideal human-awarded badge system would look like?

I’m thinking that it should include the abiliity to define badges that can be awarded by other members or through some other means that scales beyond the moderators… maybe through some anonymous nominating/voting process?


(Erlend Sogge Heggen) #43

I laid out one such system 3 years ago:

Feel free to continue that discussion there.


(Jeff Atwood) #44

You have shown no evidence of said problems with badges other than “I don’t like them”.

For known reputation system problems, see:

http://xoxco.com/news/i-love-my-chicken-wire-mommy.html

The main issue is that you aren’t distinguishing between systems of opinion and systems of facts and data. These are radically different things and require very different approaches.


#45

That article is a great example of when point systems go awry, but hardly evidence that point systems themselves are inherently a net negative. A poor implementation doesn’t negate the value of a concept. “This point system was terrible, therefore all point systems must be terrible.” isn’t something that I am convinced of.

Similarly, with Discourse’s badges, I don’t think that their problems negate all of their value. As for those problems, my issue isn’t simply that “I don’t like them”, but as I posted earlier:

There are other problems such as profile clutter, or members contributing non-constructively, but really none of that is ever too damaging because you guys have done a great thing by offering the option to disable badges system-wide, which is awesome.

I am though. To one of my earlier points, system of opinions can also benefit from a reputation metric specifically because they are reflecting something more fluid like social standing.

I’m not necessarily looking for a measure of truth, or that a user is an authority on knowledge. I’m looking for an expression of group dynamics that makes conversations more than just their words. That’s the potential value I see for Discourse compared to Q&A software.

To that end, I am grateful for Discourse’s plugin system. :slight_smile: Those of us who want this have reasonable means to do it. I’d love to change peoples’ mind on this, and give reputation a better uh… reputation… :man_shrugging: by showing better possibilities.


(Theportablegeek) #46

Considering the discussion here I’m surprised that nobody has weighed in over here


#47

I hadn’t seen that one. Thanks. :+1: