Proposal: Reputation per tag

As Olin previously stated:

I think this is the core of our disagreement. From a website perspective, sure, we want content. But the entire reason we have this site in the first place is because SE didn’t care enough about its users.

A more accurate statement about what I think our goals should be:

We should always care, whether or not good content is provided, but we encourage good content and discourage bad content.

This site is first and foremost about the users, not the content, and that’s why we are here. SE cares more about the content, legal matters, and money. We should put users first.

And that’s why a user’s motivations do matter, because any content they create will be influenced by their motives. If someone cares about users, they do their best to help users, whether or not rep is on the line. If someone cares about reputation, they follow a script in order to get that.

3 Likes

We aren’t doing HNQ now, if ever. We haven’t talked about it.

This is addressed generally, not just to you: let’s be careful about our assumptions. We’re not building an SE clone. “SE has X” does not automatically mean “Codidact has X”. Maybe it will, if we think it’s a good idea. Maybe we think it’s a terrible idea. Often it’ll be somewhere in between; we like the concept but approach it differently.

3 Likes

Even then, it’s really not that large in the scheme of things. I’ve gotten a bump from that a few times, but those blips were still small in the overall scheme of things. Anomalies will always occur, but as long as the relative noise level remains small compared to the signal, it’s just something you live with.

This probably matters even less in your scheme where you show both the total number of answers and the resulting vote aggregate. If someone has only 10 answer then their vote total isn’t all that meaningful yet. If someone has 200 answers, then one or two anomalies aren’t going to make much difference to the total votes anyway.

I can go along with this.

2 Likes

I agree with the idea, but I think this is a false dichotomy. I would say that having good content is what makes us good for the users. We need both. But I think we have to focus on good content so that we can be helpful to our users.

I’m convinced that rep is one of the reasons that SE has been so successful. The entire concept of gamification is a good one! It provides the motivation for people to use their valuable time to create good content and help people they will never even meet. It makes it fun and exciting to check back regularly, just to see if you’ve gotten something new.

If people want to chase a big number, great! Keep the good content coming! If they like badges, woohoo! The more you help, the more you get! If you don’t need these things to help you contribute, wonderful! You can ignore them!

(Even here, on this forum, we get badges for doing new things. And it’s fun :slight_smile: )

To bring it back on-topic, I think per-tag rep may be a very good idea. It seems similar to, but more visible, than the Tag Badge process on SE. I, like Olin, could get behind the summary given by @manassehkatz.

2 Likes

These are two orthogonal things, and the last statement is a bit misleading. SE has de-emphasized contributors. This seems to have been largely driven by excessive favoring of ordinary users. They see the users as providing clicks, and therefore revenue. At least that’s my perception without be privy to any of their internal communication.

OK so far.

Not necessarily. Building a repository of answers is also certainly a part of it.

No. It doesn’t work that way. Even in your terms, that script would say “write high quality answers”. The result, which either way is all you truly care about, ends up being the same.

I answered a lot of questions on EE.SE (see https://electronics.stackexchange.com/users/4512/olin-lathrop?tab=topactivity), not because I cared about some dweeb’s silly-ass problem, but because:

  1. I like teaching.
  2. I wanted to be seen as an expert.
  3. We couldn't have Andy or Spehro or Stevenvh have the top rep on the site, now could we? I mean, geesh, that would subvert world order and end the universe as we know it.

Now think about how those aims are addressed. It’s done by writing a lot of good answers. You simply can’t get a lot of rep without doing exactly what you want a contributor to do. Of the 5612 answer I wrote, even the top one only accounted for 1.3% of the total 291,000 rep. And, that’s a pretty good answer even if I say so myself. What exactly do you think is undeserved, or should have been different?

Maybe you can game the system and get a few 100 rep somehow, although I’m not sure how to actually do that. But, that’s just a small blip in the scheme of things. Even then, why bother? Writing a good answer or two is the quickest way to gain rep and a lot less trouble than trying to figure out the system and exploit some loophole.

Take a look at some of those answers and you will see that the “script” (your words, not mine) was nothing but “Write a good answer that the OP and others will think is awesome”. And, it’s not just me. Take a look at answers from a few of the other top users on the site. They all got there by doing exactly what we want them to do.

5 Likes

I appreciate the discussion. Several people made good points.

However, I feel that the opposing points being made - while all good - just argue for different end goals.
I have created a post for figuring out those end goals.

I feel that until that is resolved, either side will just continue to make good arguments for their particular end goal. There will not be a meaningful conclusion here (we could vote, but it would just be what the majority feels).

1 Like

Sure there is, and there are (anecdotally) quite a few users on SO with tens of thousands of rep gained this way, and probably more than one with 100k.

Just answer dupes with copy-pasted answers instead of voting to close, and watch the votes roll in. Sure, the answers aren’t quite as good as they would be if they were just in one place, the canonical dupe target, with focused, non-duplicated edit and comment attention. But who cares? There are upvotes to get!

1 Like

Imagine a hundred questions with identical scores. If they are split equally into 10 tags instead of one single tag, the number next to each person’t name will be 1/10th of what it would be if they all had the same tag.

That’s how it would encourage broader tags.

…I said a simple average was just a demonstration.
Simple multiplication, (Average * amount of tags) would already completely mitigate that.

As for possible reputation system, what do you think about implementing a maximum for single question - lets say you get 10 points for first vote, 9 for second and so on, with this you get 55 points for upvotes and 15 for acceptance, this gives 70 total for one question, no more. Votes still will be shown, but no points gain.

With this people who are more active (more correct answers) will be renked higher that those with very few answers but with hundreds of upvotes.

What you think about this?

Question to all: what problem are you trying to solve? Why do we need a reputation system (a) at all or (b) per tag? Before jumping to “reputation is the answer”, we need to know what the problem is.

I deduce the following problems from past discussion:

  • People want to be rewarded for their contributions.
  • People want to compete with each other, and a numeric score does that clearly.
  • People want to be able to tell which other people are the relative experts (a) at all or (b) in these tags.

Did I miss any?

I think there are better ways to do those things, but before saying more – before, IMO, any of us should say more about implementation – I want to see if there are other problems motivating these requests.

7 Likes
  • People want to feel like their questions and answers matter, perhaps regardless of their amount of time on the site.

  • People want to feel like they matter, as well

3 Likes

I’m not sure I understand what you mean. On a given question, for a given user’s rep display, multiply the average of all their relevant tag scores by the total number of tags they have a score (or a non-zero score) for? Or something? I guess that would eventually settle down, but it seems like the number of tags would be pretty volatile for folks with comparatively little site experience, making it difficult for them to get any sort of idea what their rep would be the next time they saw a question with the same tags.

I want to see if there are other problems motivating these requests.

I wouldn’t call motivating things only problems, there are some advantages in my opinion that reputation/score system will bring (still not sure which a/b is better, maybe other). What you have mentioned is quite broad so some of the below is somehow connected with yours:

  • helps to maintain better quality of the posts - some of users will put more work to gain more reputation,
  • reputation is higly connected with voting, without this votes lose some value.
  • people may want to know from who they get the answer (doesn’t mean that ‘expert’ cannot be wrong or newbie cannot right).
  • high reputation also brings responsibility (at least as I see it) to keep quality of the posts,
  • people develop and want to see it somehow - most of users are not experts when they start, but during their parcitipation, work and so on they become ones. People collect diplomas/certificates/prizes, this is nothing other.
  • it can be fun
1 Like

Regarding (a)
That is diligence.
But did we also have this level of diligence when we decided to remove reputation?
What problem did we actually solve with that?
And is that worth losing people who want to write highly rated answers?

After all, we started out from “SE is a good system, but the greedy company behind makes it suck”.
I don’t want to engage in whataboutism here, but I feel that before we ask for a good reason for keeping the working system the way it is, we need a good reason to change it.

To me, [only having selfless people] seems more like a reason to start a religion than a website.

Regarding (b)
That was just an improvement to the rep system, which, now that I think about it, might not apply everywhere. On French Language, rep probably just means “this user is good at French”.

But on SO, somebody who’s excellent at C# might have no idea about Python; somebody who’s good with databases might give bad advice about version control systems. And it also would be a pity to split this into all those subgroups, because there’s a lot of overlap between things.

3 Likes

No, multiply times the number of tags on the question.

To fight against questions with many tags showing significantly lower rep.

Sorry, I don’t like this much. It defeats the reward for writing a really good answer.

2 Likes

This is backwards. SE did get a lot of things right. Most of the issues with SE are with management and their attitude towards the people that contribute the actual value to the site. The SE system as a whole works pretty well. The default is therefore do what SE did unless there is a good reason not to.

So the questions really are Why would you NOT want rep? What problem are you trying to solve by taking rep away?

The ‘problem users’ that I’ve had to deal with in my time were problems because they want either rep (/more upvotes) or power - mostly the former (albeit possibly as a way to get the latter). Getting rid of rep and introducing a better voting system should get rid of most of this.

Ultra high-rep users have a tendency to be very quick to answer questions (because answering first gives you the rep first and prevents others from answering) - there’s nothing inherently ‘wrong’ with this but it does lead to fundamental issues on the thinking of how the site works - my best answers have been the ones where I’ve had to do the learning in order to teach, so I get as much use out of the question as the questionner got out of my answer. If there’s already an answer by a high-rep user, I’m much more disinclined to answer and so, learn less, defeating the point of a community for learning.

In addition, the above leads to extra competition, which tends to lead to ‘putting people down’ and ‘beating the other person’ instead of ‘lifting people up’ and ‘encouraging the other person’.

It makes the other important stuff less noticeable (e.g. editing, flagging, reviewing, cultivating the site, using comments to clarify and improve questions etc.)

A number by itself doesn’t say whether someone got that number by answering lots of easy questions (i.e. is quick at answering questions) or by answering lots of harder questions (i.e. is an expert, at least in some sense). It also doesn’t say what they’re good at. In addition, there are some amazingly wonderful people who don’t have the same rep, yet can be just as knowledgeable and good at answering questions as someone with loads of rep, perhaps because they answer stuff on an obscure tag.

If you can loose rep for being wrong, why risk being wrong? Yet being wrong is a part of life, something everyone needs to learn to deal with and an essential part of the learning process. Being wrong is bad for getting the right answer. Being wrong is good for learning.

The ‘HNQ effect’ where people would write controversial things and clickbait for the sole purpose of getting upvotes and rep. I like neither of these.

the “Someone downvoted me, so I lost reputation, so they’re unwelcoming!” misconception. (Perhaps conflating votes with rep a bit here, but votes and rep are explicitly tied, so they’re hard to separate) Whoever had the idea of a system called ‘reputation’ and a way to loose reputation may not have thought about the consequences that ‘loosing reputation’ in real life is a very bad thing, so people are naturally going to consider this a bad thing online as well. Perhaps even the choice of word used matters?

The above may lead to fewer people using downvotes as much as they should.

It places the emphasis on something other than what we want the emphasis to be on and heavily related to this, it provides an extrinsic motivation. I’d like a psychologist to confirm this but I’m led to believe that if anything, extrinsic motivation doesn’t generally last long and often decreases intrinsic motivation.

7 Likes

That was true for the first few days, if I recall correctly. (I wasn’t here on day 1 either.) But if all we want is an SE clone without the company, that exists now – there’s a long list of clone sites/software listed somewhere on this forum and also on Meta.SE.

We aim to do better. We’re starting from scratch, and that means we can go back to core principles: what serves communities best? This has led to a lot of innovations and there will be more.

Once reputation is granted it will be very hard to take it away. In the absence of good reasons to have it, we chose to leave it out. We do need to provide suitable feedback to users in a way that encourages participation; I just don’t think a single rep number is it. Let’s look at what users have done. For example, if somebody answers my complicated question on a niche topic, I don’t really care that the person has 50k rep from answering totally unrelated questions; that tells me nothing about this answer. Knowing that this person has answered dozens of other questions on these tags and those answers have been well-received, however, tells me something. When reviewing an edit, seeing that someone has 20k rep means nothing, but seeing that this person has suggest 97 edits, 96 of which were accepted, tells me something. Etc.

I want us to develop these ideas – for version 1.2 or so. Meanwhile, rep should stay out of MVP to avoid later confusion and loss aversion.

Does the fact that you can’t see my “rep” here on this forum affect how you read my posts? Demonstrably it hasn’t kept me from posting.

9 Likes