I find this question a little more pertinent now, given Stack Overflow Careers and Joel's assertion that reputation is a good indicator of programming ability.

The majority of my reputation comes from either insanely easy questions or meta-ish questions. Sure we have the meta thing policed, but what bugs me is the huge difference in "plus rep" you can receive for providing pretty simple responses in popular technologies (such as C#) against excellent answers in more niche ones. I've seen awesome posts in niche techs that have like + 30 tops, whereas there are others that are kind of "meh" in a popular technology with +100 or more.

Is there anyway we could adjust the voting system to address this? Bounties help a little, but I was thinking more along the lines of being able to +x an answer or even gift reputation to others for specific answers that really helped.


I appreciate that answering the question in the title is the obvious thing to do but I'm much more interested in discussing systems to address this issue that don't also open up other cans of worms.

For example, if we had a feature where I was allowed to give an extra 100 reputation to an answer per day because I have greater than X reputation where would such a feature fall down?

Or are we agreed on the premise that such a system is impossible, and the current one is the one that "sucks least"?

  • Can you give an example of a "kinda 'meh'" post with a +100 score? – mmyers Nov 5 '09 at 20:21
  • 3
    @mmyers: Look at any of my answers with more than 100 upvotes. I'm not saying they're bad, but there are plenty of answers I'm far more proud of. – Jon Skeet Nov 5 '09 at 20:25
  • For example, stackoverflow.com/questions/284797/… (to beat a dead horse (well, pony (token Lisp joke here))). – Super Long Names are Hilarious Nov 5 '09 at 20:33
  • I'm actually referencing +100 rep which is only 10 upvotes. stackoverflow.com/questions/178456/… is an example of a relatively "meh" post. It's okay, it doesn't answer the specific question. It does show a trick but doesn't explain the trade off of the trick as this can make it harder to debug if they're too far nested (and if you use this trick "as default" then they will get VERY nested). – Quibblesome Nov 5 '09 at 20:35
  • Actually I think the "hello world" language one is the best post on that thread as it explicitly points out how pointless the question actually is. The ending line is also hilarious too. :) – Quibblesome Nov 5 '09 at 20:37
  • 1
    I can't edit, but please consider revising your title and clarifying the question. My reading suggests that you are really asking, "How do we make sure the hard questions get as much rep as the easy questions?" – Pollyanna Nov 5 '09 at 20:50
  • @quibblesome - It's funny, but not 296 upvotes of funny. And also, the post explicitly disallows interpreted languages (which may have come after Dr. Skeet's answer, so I won't hold him responsible for it). And also I wanted to cite a specific Skeetian example, rather than just a random "meh" post (which your example certainly is). – Super Long Names are Hilarious Nov 5 '09 at 20:51
  • @quibblesome: your edit deleted the original question. The answers look funny now. – snicker Nov 5 '09 at 21:07
  • Keep in mind that the title now reflects the original question. The old title was simply a teaser. Anyone who read the question and understood what was being asked will see that the new title is a better representation of the question. – Pollyanna Nov 5 '09 at 21:13

No. There is some measure of ability (you'll have a hard time getting a high rep without it), but it's also a lot to do with how well you can express yourself in writing and even more to do with have much time you spend on StackOverflow.

| improve this answer | |
  • 3
    Exactly. High rep (say > 10k) generally implies skill, but skill does not imply high rep at all. – mmyers Nov 5 '09 at 20:23
  • 1
    "how well you can express yourself in writing" - Most places would rather hire a mediocre programmer that can readily express themselves over a better programmer with poor communication skills. So, in this case, the "better programmer" is the one that can not only program, but explain and express their work to those who lack subject matter expertise. – Pollyanna Nov 5 '09 at 20:45
  • 1
    +1 on the time. I don't see how a high rep will help out on the careers site either...it just shows that someone spends a chunk of their day on SO instead of working. – dlux Nov 5 '09 at 20:52
  • @DLux It's hard to really gauge how good/bad spending time on stackoverflow is. For example I love the #c# IRC channel on Freenode because in "downtime" I can look at it and pick up really useful information about techs that the people are talking about that I don't know. Later on a problem will appear that this tech solves and i'll already have the answer because I was "wasting time" on IRC. It's really hard to measure how productive/unproductive this is in the long run. – – Quibblesome Nov 5 '09 at 21:03
  • Joel, do you have enough rep to edit the posts that came in before I changed the title? Like the way Adam has done, re-add the old question before the response. Otherwise it looks like people are answering a different question than before. – Quibblesome Nov 5 '09 at 21:10
  • Yes, but even if I didn't the ♦ is all the rep I need ;) . A better questions is will I? People who browse meta generally know how to check a revision history. Any individual user who want to edit their own answer to match up with the edited question can also do so. – Joel Coehoorn Nov 5 '09 at 21:15

Totally agree with what Joel Coehorn says. Time spend is the major driver behind my reputation :)

In addition, high rep depends on how (for lack of a better word) "hot" your answers are. An answer to a simple C# question can get you tons of upvotes. Because it's simple, many people feel confident that your answer is right. And many people follow the C# tag.

But in the job market, the opposite is true. The "colder" your knowledge is, the harder it is to find, and the higher your market value. How many people answer SAP questions here? Yet SAP positions are among the higher paid in the industry.

I wouldn't be surprised if high reputation is negatively correlated with salary.

| improve this answer | |
  • 1
    I always thought SAP jobs paid well cause SAP sucks and nobody wants to work with it all day long. – Quibblesome Nov 5 '09 at 20:48
  • It is, to some degree, a self limiting employment pool. Unlike actuaries where the pool is artificially limited... – Pollyanna Nov 5 '09 at 20:52

Is reputation a good measure of knowledge?

Obviously, no.

How do we make sure the hard questions get as much reputation as the easy questions?

I don't think it matters. Reputation and question difficulty are unrelated. The system as currently implemented allows bounties for those really hard questions, but other than that there is no need to award harder questions with more reputation.

They either get answered, or they don't, and that's the point of Stack Overflow - good answers to reasonable questions. Reputation is just the carrot - it turns Stack Overflow into a game with a leaderboard, and encourages people to play.

Once you make reputation dependent on a subjective measure, such as "difficulty" then the control loop (the carrot egging the programmer on) becomes unstable without very, very, very careful adjustment and damping.

I believe adding more subjectivity to a scoring system will not only result in a worse outcome, but will actively drive some programmers away who currently enjoy being 'graded' using relatively objective measures - you are either right or wrong. A lot of programmers hated English not because they had difficulty with the language, but because you couldn't turn in a paper and know in advance if it is objectively 'correct' or not.

There's still a ton of subjectivity in the current system, but we already have plenty of complainers about spurious downvotes they don't feel are deserved. Do we care to face the onslaught of, "The question I answered should have been rated difficult because it was hard for me!"

| improve this answer | |
  • 1
    /me likes the link. – Quibblesome Nov 5 '09 at 20:42
  • It's an interesting question. You can be good at stackoverflow, but can you be really good at it if you don't already have a large corpus of knowledge instantly available? These guys don't just answer questions, and they don't just answer them well. They often know the answer before they've finished reading the question, and they answer it well within seconds. – Pollyanna Nov 5 '09 at 20:47
  • But if you want to demonstrate this, feel free to upvote my answer significantly above the other answers, since it was obviously very hard for me to answer. I mean, I even sweated a little bit, almost! – Pollyanna Nov 5 '09 at 21:08
  • Thanks for the update. As per Stackoverflow's original purpose you are spot on, it works for that. I think the issue is the new feature of allowing potential employers to view profiles, they're not using the site for Q&A and therefore risk taking "rep" out of context. I might go as far as to suggest that when employers are re-directed from careers rep should be masked and possibly replaced with a measure that shows how active they are on the site instead. – Quibblesome Nov 5 '09 at 21:16
  • 1
    It's actually quite hilarious to me. Less than a year ago I asked stackoverflow.com/questions/424727/… and the response was swift and unanimous - no one would ever put their rep on their resume. Yet here we are, and now people are facing the same problems that question brought up - what does rep really mean, and does it have any place being near a resume? – Pollyanna Nov 5 '09 at 21:21

Given equivalent behavior and areas of interest, rep can be a relative measure of ability, I think. For example, if you had two users who answered questions in roughly the same tags and had roughly the same number of answers, but whose reputation level were an order of magnitude different (10K vs 1K, say), I think you could infer that the higher rep user had more ability than the lower rep user. I doubt that you could say that they were 10x better, though.

| improve this answer | |
  • Stackoverflow is not unlike the game show Jeopardy in a few ways. Knowledge is important, but the person with the best trigger finger wins, all else being equal. – Pollyanna Nov 5 '09 at 20:56
  • oOo now that is interesting. Perhaps we could add a "relative karma" score that relates to the tags you mainly post in and the other people that post in those same tags. Programmer hunters (as per stackoverflow careers) would love such a view as well. – Quibblesome Nov 5 '09 at 20:58

The answer is... it can be.

There are many users that have gained a fairly high reputation by only asking questions... and fairly simple ones at that.

The best way to tell the difference is the questions asked to answers given ratio, the lower the better. Other factors are involved, such as the number of total upvotes on their answers, and esotericity of their questions (higher is better).

| improve this answer | |

It has a lot to with luck.

Can you get your well reasoned and crafted answer in there with at least one up-vote before anyone else?

If you can then your answer appears first on the page and is more likely to gain more up-votes. There have been discussions on this - search for "Fastest Gun in the West" and now answers of the same score are randomised rather than being presented in chronological order, but once an answer starts getting votes it's much like a snowball rolling down hill and gains more.

NOTE This answer applies to the question as originally titled - I'm going to have to think some more about the updated version

| improve this answer | |

Highly technical questions usually get low views because not many people are interested in them, especially if the user choose a good title.

StackExchange could choose to value up-votes more when a question has low views.

An up-vote could count +15 rep when 25% of the people who viewed the question choose to up-vote the question.

| improve this answer | |

You must log in to answer this question.

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