What is meta? ×
Meta Stack Exchange is where users like you discuss bugs, features, and support issues that affect the software powering all 127 Stack Exchange communities.

I think the reputation points schemes of stackexchange is something very important that drives us.

However I have the feeling that sorting people (instead of just characterize the range of point) could create voting strategies that are not necessarily good for the community.

With a site like stackoverflow it is not really a problem as there are a lot of users but with new sites such as the ones arising from Area 51 I think this can go against the common will ...

My idea is that reputation should only be seen as a range and not a precise number of point (and this number should be hidden).

EDIT: I see I get a lot of down votes, is my question annoying (hence refering to some dark side of man) ? badly written ? useless ? please comment !

share|improve this question
    
Do you mean the sorting of user lists by reputation? Just to clarify. –  Pëkka Aug 11 '10 at 8:52
    
It is more than that, I think the user themselves should not see the exact number of point they have, they should see the range they belong to... –  robin girard Aug 11 '10 at 9:04
    
you mean like displaying a "user level" instead of his rep \approx experience? –  Tobias Kienzler Aug 11 '10 at 9:11
    
bump @Pekka (@robin you need to use an @name to notify someone of your reply) –  Tobias Kienzler Aug 11 '10 at 9:12
    
Cheers @Tobias. This is an interesting idea, but I don't see what the advantage is of showing a range ("10k-20k") instead of the exact number? If anything, (external SO/MSO...) reputation should maybe be hidden entirely on SE. That might be something worth discussing... –  Pëkka Aug 11 '10 at 9:14
    
"not necessarily" - what? –  nb69307 Aug 11 '10 at 9:21
    
@Tobias Not if the person you are responding to is the questioner. –  nb69307 Aug 11 '10 at 9:21
    
@Neil Butterworth: he updated it "not necessarily good for the community" –  Tobias Kienzler Aug 11 '10 at 10:50
    
You mean that users are specifically downvoting people to avoid them reaching certain benchmarks? –  Ocaasi Aug 11 '10 at 11:38
    
@Ocaasi I don't think people are that bad :) maybe its just me that feel bad being ranked like at school... I should see a psy for that :) but I guess it creates incentives for who you vote up for (i.e people that are "far enough" are more likely to be voted up, I'd love to do statistic to see if I get confirmation:) ) –  robin girard Aug 11 '10 at 12:06
    
@Robin, Meta is a little different from Stack Overflow. Because it is the venue for feature requests and discussions, downvotes indicate that it could be a bad question or (more likely) that people disagree. This simply means that the downvoters do not believe your idea should be implemented. –  devinb Aug 11 '10 at 12:36
    
@devinb thanks ! I did not know. –  robin girard Aug 11 '10 at 12:39
add comment

3 Answers

up vote 2 down vote accepted

This is an interesting idea. Two things you haven't clarified enough (IMO), is what exactly this would accomplish, and how large the ranges would be.

Still a few items to note

Skinner Box

Like it or not, we are all behaviourally driven to continue doing things that create an immediate positive reaction. When I post a strong answer, I am able to immediately (hopefully) see my reputation tick up by ten. That makes me feel warm and fuzzy. However, if I am represented by a point range instead of a point total, I lose that immediate reinforcement.

An analogy (in support of you) could be made (validly) to exp in RPGs, where it ticks up gradually, building to a "level" increase. However, I don't feel that model is necessarily appropriate to Stack Overflow, where individual reputation points are much more important. The EXP comparison falls because in RPGs your level is simply a sum of all the experience you've gained, whereas on Stack Overflow reputation is much more tied to it's source. "I earned X rep from Y question."

While this might be a case of variable interval reinforcement. I think it would produce a lot frustration, because my incredible posts would rarely be the ones that happened to tick me up to the next level. It would be one of the throwaways. This serves to not reinforce quality posts, but rather arbitrary posts.

While this might be a positive step in helping other users (readers) focus on the correct variables, it will have a huge negative effect on the answering users, and they are the core of the site.

share|improve this answer
add comment

There is something interesting in the idea of larger ranges. Currently there is the very narrow (1 point) and the very broad (ability benchmarks like 1k, 2k, 10k). Using levels rather than points causes people to think more generally about their behavior. It might cause an increase in focus on "things that matter" rather than points.

Then again, the point system, properly designed is supposed to make points reflect things that matter. Caring about whether or not your answer will get voted up typically equates to writing a good answer. Since the points are aligned with broader site goals, there is little harm to showing the exact score.

One possible caveat is that users are focusing on the score, and on writing good questions, but somehow ignoring "other" aspects, perhaps the types of things that cannot be quantified, like, "being nice" or "welcoming new users". I consider this a quantitative bias which typically occurs whenever some concrete metric is introduced. The analogy to grading in schools might be that if teachers relaxed grading schemes, then students would question deep truths and seek great insight if they weren't chasing to hit 89.5 on their next exam. Then again, if teachers didn't give grades, students might not show up.

The SO solution has been to simply point-ify or badge-ify everything, closing any leaks in the extrinsic motivation vessel wherever they appear. It doesn't seem like your proposal would fit into that so well, since suddenly points wouldn't matter immediately. Instant feedback is one of the hallmarks of user-interface and game design; psychologists have long known that the proximity between a reward and a behavior is highly correlated with the reward's effectiveness at conditioning the behavior. In other words, the rat won't push the button if the cocaine only comes out two weeks after the push.

In sum, there's always a temptation to relax or do away with overly specific schemes of extrinsic motivation. Yet, since SO was designed to capture coders and gamers, two communities which live and die by 'the detail', a broader-ranged reputation probably wouldn't satisfy their carnal thirst for numeric validation. I was a philosophy/polisci major, so I don't value anything not amenable to Hegelian synthesis. We all have our tendencies.

share|improve this answer
add comment

I've seen no evidence of counterproductive voting strategies. If you have evidence of such, that would be interesting, but it'd have to be pretty strong.

Certainly I've not experienced any apparent downward pressure breaking various thresholds (1000, 1500, 2k, and just yesterday 3k) since I became much more active on SO this year. In fact, I rep-capped somewhere around 2,940 the day before I hit 3k, which is the last big threshold until 10k (which I don't expect to hit for a long time, if ever).

Besides, we're developers. We like precision. :)

P.S. On meta.SO, downvotes on proposals amount to a vote against the proposal itself, not necessarily a statement that the proposal shouldn't have been brought up. I've downvoted this proposal even though I think having the discussion is interesting.

share|improve this answer
    
I haven't seen any evidence neither, but who can see that ? we don't have access to the voting affectations... (hopefully). But I like the statistical problem that arise from that :) –  robin girard Aug 11 '10 at 12:37
    
@robin: Well, there are some statistical analyses we might do with the data we have available. Like looking for an odd "bunching up" around practical and psychological thresholds. –  Nicholas Knight Aug 11 '10 at 12:52
add comment

You must log in to answer this question.

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