-2

Question: Answerers may be under time pressure to answer first, reducing the quality of initial answers. How can this be mitigated?

While the primary objective of this site may be to provide a repository of information on server topics, there is undoubtedly an element of competition (which is part of the appeal). A consequence of this is that multiple posters often post answers within a few moments of each other shortly after the posting of an easy question.

Since easy questions or those that are more subjective tend to receive higher hit counts, and can offer significant numbers of votes, there is considerable incentive to have the first valid post (since the first valid post is likely to earn an up vote and the default sorting scheme orders by votes, it gives the first poster an advantage).

Unfortunately, a consequence of this is that the initial answer may not be up to a poster’s usual standards, and may require subsequent editing (which may not occur if the question garners a sufficient number of votes).

One possible solution to this is to not display answers for the first few (say 10) minutes, and to treat all answers posted within that time frame as being equal. This allows answerers to know that they have at least a few minutes to answer the question appropriately instead of simply providing a superficial answer. Adding to the same idea, if answers are randomly ordered (for all non-negative score answers, at least for the default sorting scheme) during the first few (say 30) minutes – it may prevent answers from being unfairly advantaged.

closed as off topic by MDMarra, Chris S, mdpc, Ward, John Gardeniers Oct 24 '12 at 22:58

Questions on Meta Server Fault are expected to relate to Server Fault or the software that powers the Stack Exchange network within the scope defined by the community. Consider editing the question or leaving comments for improvement if you believe the question can be reworded to fit within the scope. Read more about reopening questions here. If this question can be reworded to fit the rules in the help center, please edit the question.

  • This has been covered many times on mSO, most popularly in Fastest Gun in the West Problem. I don't think SE will change their position on this, but you're welcome to take it up with them. – jscott Jan 8 '12 at 11:50
  • @jscott: Thanks for your response - I didn't think to search mSO, and that didn't seem to show up mSF when I looked for similar questions. That question definitely makes for interesting reading - and appears to cover pretty much every angle possible. (I do rather like 'Evan Miller/Reddit, etc' approach, though). I likewise agree with the point that a quick answer is exactly what the original poster is sometimes after (although, I usually use comments for that, even though that may not be ideal). – cyberx86 Jan 9 '12 at 4:41
3

While there is some merit to the idea on a really fast moving site like SO, here on SF we're slow enough that for most questions the long tail pays better than the immediate post. There are questions here where there will be three answers on it within two minutes of posting, which is the exact case you're talking about here.

However, in my experience these questions are a small minority of questions that arrive on SF. We actually get more 'does not belong here' questions than we do the sorts of questions where fastest-gun is a big problem. What's more, many of the fastest-gun questions are actually duplicates of other historical fastest-gun questions so dropping a book-sized answer is not that productive.

There are some competing interests when it comes to questions, answers, and voting:

  1. The questioner just wants their question answered. Faster good-enough questions may be all they need or want.
  2. The answerer wants the accolade of their peers (votes). They want each question given equal consideration by voters.
  3. The electorate hates stupidity and wrongness. They'll upvote the good stuff and occasionally downvote the bad stuff in order to drown out the bad stuff.
  4. The long tail just wants the best answer. They'll upvote the good stuff for years after the questioner has wandered off.

Point 1: The Questioner
There is a reason that some low rep users have a waiting period between when they ask a question and when it can be marked accepted. This is so they don't post a question, come back five minutes later, read the three answers, accept one, and leave. It's to incentivise answerers to keep working.

Secondly, getting answers is the whole point of these sites. Many, many users will just wander away if their question doesn't get any attention within 10 minutes of posting. A steady stream of answers is a good thing. Fast upvotes tell questioners which answers are the good ones. Feedback is good, since it encourages repeat visits.

Point 2: The Answerers
Answerers want votes and accepted checkmarks. Accepted checkmarks are great since they bring a low-voted answer to the top of the list for the long tail to consider and vote on. Does each question get an equal weighing by voters? No, in most cases the top three vote-getters are all that are read.

Point 3: The Electorate
There are fewer of these around than there used to be, but there are SF users who read questions so they can vote on them. You can tell who they are by looking at the Electorate and Vox Populi badges. For the rest, they'll read questions for which they have sufficient clue to possibly drop an answer on it and end up voting for the answers that are already there.

Want an easy way to encourage voting on question? Drop a slightly wrong answer on it. It'll cause voters to upvote the others, and occasionally downvote the bad one. If there is one thing our electorate hates, it's seeing crap have equal weight with a good answer.

Point 4: The Long Tail
Questions earn rep long after the OP has left the site. I've earned many Enlightened or Good Answer badges from questions over a year old. A couple of my answers have earned more than the accepted answer on a question.

When answering questions, even fastest-gun style questions, remember you're answering for Google as much as you're answering for the questioner. If you do a good enough job of answering, it might even make it onto the canonical answers list and used as a Duplicate-Of close-target; which will earn rep over the long tail.


Your proposal overbalances in favor of Answerers. They are the bulk of our active community, but StackExchange as a whole has deliberately tipped things in favor of questioners and the long tail in this regard.

  • Thanks for the response. In retrospect I believe my own experience on the site does agree with yours. I have found myself expecting there to be multiple answers by the time I post mine, but invariably this is not true. Questioner: agree - although, I usually post 'short' quick answers as a comment (possibly because my academic background cares more about the why/how than just the solution) - your explanation does highlight some harm in that though. [...continued] – cyberx86 Jan 9 '12 at 5:13
  • Answerers: agree, although, I hope that there is the thrill of solving a puzzle in there somewhere (and maybe even helping someone else) - although those are harder to quantify. Electorate: paritially agree - only the top ~300 voters (out of 69k users!) have more than 250 votes. It is usually easy to distinguish utter refuse from the rest, but to judge correctness on two plausible answers does require some knowledge of the topic. [...continued] – cyberx86 Jan 9 '12 at 5:14
  • I think it is that most people just don't read random questions (and answers) - they read a question for a purpose (and great example with the 'contrast' - common marketing technique). Long Tail: partially agree - I like the thourough answers (both when looking for a solution and posting one) as it provides an opportunity to look into the material in more depth and learn something new. As for the canonical answers, I have found that are sometimes duplicated to a comm. wiki. Excellent points overall - thanks for sharing your perspective. – cyberx86 Jan 9 '12 at 5:14
3

This is called "The Fastest Gun In The West" and was discussed on meta.stackoverflow about 2 years ago I think.

Answers used to default by sorting by votes > age. Now they sort by votes > random.

This means that in the beginning, when there are 0 votes, no one answer floats to the top, because the top answer tends to get an un-even (and often un-deserved) number of upvotes.

Easy questions tend to get 2-3 answers within about 60 seconds of eachother, which means that none of those answerers have any systematic advantage over others, and it comes down to who wrote the better answer.

I've seen plenty of quick, first answers that were crap being outvoted by better answers that came along hours or even days later.

The real problem is that not enough people vote...

  • 2
    "not enough people vote..." I'm doing my best! – Ward Jan 8 '12 at 20:19
  • 1
    You are the exception... – Mark Henderson Jan 8 '12 at 20:29
  • 3
    "not enough people vote..." Sometimes there just isn't a reason to vote. There are some really great answers but the majority are mediocre at best - mine included. – John Gardeniers Jan 8 '12 at 20:54
  • @Mark Henderson: Thanks for the feedback. The random arrangement doesn't fully address the problem - only a small percent of users vote (a real problem as you point out) - which means that there is likely a (small) time delay between votes. This results in the first upvoted answer rising to the top, negating the random sort, and causing a positive feedback loop. It also somewhat disadvantages answers that arive say 15 minutes later (although, as per sysadmin1138 answer, these will likely rise over a longer period of time). – cyberx86 Jan 9 '12 at 5:14
  • @John Gardeniers: I think that part of the issue here is that people assign different meanings to their 'up vote'. To one person it may mean that the answer is right, to another it might mean that the answer is outstanding, etc. If the meaning was standardized (or arguably, if you could cast one up vote for 'correct' and two for 'exceptional') I think it would result in more votes. – cyberx86 Jan 9 '12 at 5:14
  • And sometimes you get answers that are both correct and outstanding, like this: serverfault.com/a/347786/7709 – Mark Henderson Jan 9 '12 at 5:23
  • @cyberx86, you are perfectly correct and there's nothing wrong with that. Nevertheless, well written questions and answers do get more votes. – John Gardeniers Jan 9 '12 at 21:16

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