3

This issue has been addressed scores of times. How can we improve our voting system? touches on some of the concerns which were exemplified today in two ELL questions (votes at the moment of this writing are in parens):

The voting model serves SE very well, it seems to me, in most of the areas of interest served by its sites. In the case of ELL, though, our objective is supposed to be the creation of a database of useful Q's and A's on English, with the emphasis on material which benefits EFL students. It is surely helpful to a learner to understand that "girlfriend" does not always mean "lover", but it seems to me that an interesting question about voice and the copula is of wider use and benefit to such a student than a mildly provocative discussion of the reluctance of men to refer to their male friends as "boyfriend." (Not to mention that the male repressed homosexuality evidenced by this reluctance is a better fit for cogsci.stackexchange.com than for ELL!)

As was pointed out in commentary, the second question above certainly has value to a learner, but its high popularity put me in mind of previous (and better) examples. The point here is not just the high vote count for the second, but the disparity between the two vote counts.

Isn't there anything we can do to prevent the occasional elevation of questions whose popularity is based as much on their subject matter as on their value to the learner? For instance, can we provide moderators with some form of "enhanced up/down vote" which could be applied judiciously in such cases?

(That is a seat-of-the-pants idea and may not pass muster, but this seems to me an important and obvious flaw in the "popularity" model that we could well consider addressing.)

| |
  • Are you saying that a pragmatic question that helps learners not embarrass themselves because they mistake "girlfriend" to mean "lesbian partner" is foolishness and that somehow a terminology discussion about predicate nominative versus participial adjective is more valuable? I'm guessing the community begs to differ. The first question is more interesting to me, but that's only because I know the answer to the second question with certainty. – ColleenV Oct 25 '16 at 22:02
  • No no no. I'm not saying that there is no value in the girlfriend question. I'm saying that the voting model skews the relative values of pairs like the one in my example. I should edit "foolishness" above, it's not accurate. – P. E. Dant Reinstate Monica Oct 25 '16 at 22:21
  • As far as I know, it's everything to do with how a question is considered "hot", as in "Hot Network Question" worthy. Here is a likely scenario of the "girlfriend" being raised up to be a HNQ: the OP posted the question, a user posted an answer (which, IMO, is not particularly good, as it's so short and doesn't explain much), another user posted another answer. (All of these happened within 15 minutes, according to the timestamps.) One of these answers got upvoted. The HNQ algorithm saw this question get lots of attention, and very quickly, so it made the question a HNQ. – Damkerng T. Oct 25 '16 at 22:31
  • @DamkerngT. The HNQ algo is a black box to me, obviously, but that makes sense. – P. E. Dant Reinstate Monica Oct 25 '16 at 22:34
  • If I may butt in, my question actually had a fair number of views, over a thousand, and two good solid answers. I can't complain really. – Mari-Lou A Oct 25 '16 at 22:49
  • 2
    But sex will always win on the popularity stakes, i.e. The Hot Network Questions, and you don't need to be particularly good at English to have an opinion, or to post an answer. Interesting to see that the most upvoted answer is one of the shortest and is the only one that contains a dictionary citation. – Mari-Lou A Oct 25 '16 at 22:50
  • 3
    Oh well. You've brought up an issue that's been regular users' concern for a long time. Questions that make it to the Hot Network Questions list usually gain a lot of additional upvotes that generally comes off as an inflation is votes and reputation and hurts the image of other highly upvoted good content. Over time, we've discussed a possible solution in meta.SE in various posts, but the only solution I have thought of right now is to ignore them and move on. A post with a score of ten to twenty tends to be of higher quality than posts with scores more than forty. – M.A.R. Oct 26 '16 at 6:29
  • @Mari-LouA - I don't think this is just a matter of "sex sells." A lot of higher-ranking questions are simply more original and less mundane than the common fare here. – J.R. Oct 27 '16 at 2:06
  • 1
    @J.R. the problem I have with the girlfriend question is not its slightly prurient character, which I hope no one is denying, but "how" it was formulated. It is POB, it shows no research, it lacks context. The only context provided is "classmate" and "my girlfriend" and... that's it. But I am aghast that the question currently has 34 upvotes, according to SE philosophy this is a good question. I don't think it's a particularly good question, I think it's a fun question, and a good distraction worthy of three or five minutes. I even like trying my hand on fun questions myself. [cont'd] – Mari-Lou A Oct 27 '16 at 4:51
  • [cont'd] However, I believe that P.E.Dant was spurred to write a meta question because at one point I exclaimed "two downvotes"? On a question that had, at the time, received four upvotes. According to the SE tooltip that meant my question, "he is buried" does not show any research effort, it is unclear or not useful which is laughable. – Mari-Lou A Oct 27 '16 at 4:51
  • @Mari-LouA - Oh, well, that's just politics. You're going to get that sometimes – people have their "reasons". – J.R. Oct 27 '16 at 8:03
5

Isn't there anything we can do to prevent the occasional elevation of questions whose popularity is based as much on their subject matter as on their value to the learner? For instance, can we provide moderators with some form of "enhanced up/down vote" which could be applied judiciously in such cases?

As a practical matter, no, there isn’t anything ELL can do along these lines. That kind of change would have to be applied to the entire Stack Exchange network—and I can assure you that there are other SE sites would not welcome that change, no matter how much ELL thought it was a good idea.

Moreover, it’s not at all in keeping with the voting model that Stack Exchange is built upon, and doesn’t mesh with the SE Theory of Moderation. Voting is a matter of prime importance on Stack Exchange, and is ultimately the most important way in which Stack Exchange sites are community-run. The job of moderators is primarily exception handling, either by handling flags or by raising community discussions on subjects that might not be apparent to those who aren’t following questions the way moderators do.

Which is why the moderators aren’t in any way empowered to influence voting above and beyond their status as community members. The final privilege with respect to voting, the downvote, is received at 125 reputation—every member with that much reputation is intended to have equal voice in the voting process. In fact, moderators cannot even see who voted or how (barring the semi-automated systems that may bring suspected sock puppets or other voting abuses to moderator attention).

| |
  • 2
    I concur with the sentiment of this answer – particularly the last two paragraphs. In fact, I think it's rather presumptuous for someone to suggest that the community is in the habit of "voting wrong," and therefore moderators need an intervention mechanism. If 20 people think a question is worth an upvote, then that question should get 20 upvotes – even if a few snoots regard the question as "not particularly helpful." – J.R. Oct 27 '16 at 2:16
4

Things that factor in to my up-votes:

  1. Was this question or answer useful and/or interesting to me?
  2. Was it easy to understand and read?
  3. Can I imagine this question being helpful to other visitors to the site?
  4. Is the author engaged with their post? Are they refining their post and taking feedback into account? I always up-vote that when I see it, because it's behavior I want to encourage.
  5. Did the author put a lot of effort and detail into their question? I also up-vote that because I want to encourage it.

Nowhere in that list do I think about whether this question as valuable as other questions I've up-voted, or more valuable than questions I didn't up-vote, or how my vote will change the score of this question relative to other questions on the site.

I don't want to disparage the "He is buried" question. I think it is a good quality question, but the language is far more advanced than most of the questions on ELL. If you look at the list of top scored questions our highest voted questions are not about sex (but there is one about a female friend that is not a girlfriend) so I don't really think 'lesbian' in the title mattered all that much. Other than StoneyB's canonical post on the perfect, the most popular questions are mostly focused on meaning and practical usage.

The "does girlfriend always mean romantic girlfriend" question is easier to read and deals with a practical conversation problem that many folks have already encountered (or can imagine they might encounter). It doesn't surprise me that more of our community is interested in/finds more value in understanding something said in a conversation over understanding the nuance in a written phrase.

We don't know where the "final" score of these two questions are going to end up. Maybe the girlfriend question stays at 25 after its initial burst of popularity and the 'is buried' question turns out to be more interesting to people over time. Regardless, score doesn't measure inherent value, which is really a philosophy question.

Value is relative to the person assessing it and their current situation. Someone trying to understand the Richard III: Does it matter where he is buried? article will value the "He is buried" question, and someone who has a peer that uses "girlfriend" ambiguously would value the other question. The assessment of value will probably change the day after they cast their vote when their context changes.

Popularity gets a question more attention and if it has mass appeal the votes can pile on rather quickly. Not much you can do about that except maybe doing some more aggressive rate-limiting on the hot network questions after they've gotten a certain amount of attention. Neither of the examples in your question has any more inherent value over the other though, and raw score without normalization across a correlated group can't really measure much more than popularity.

I don't see this as a problem with the voting system. I think that it's a statistics problem. Everyone assigns numbers more meaning than they actually have. If you want to solve it, SE should stop displaying a number and just show a color gradient (and then watch the flood of questions asking "How much better is 'puce' than 'lilac'?" roll in).

| |
  • The gradient idea is as amusing as the projection of a flood is accurate. But you are not a garden variety voter, and your very rational reasons for voting are, I'm pretty sure, much more considered than those of many voters. As interesting as a gradient scale might be a logarithmic scale rather than linear one. Also, I'm willing to bet that if the question had been posed as I edited it—the identical question, but phrased without the word "lesbian" and since rolled back by Mr Ashkan— it wouldn't have made so much as a ripple. There is information in that conjecture, if it's accurate. – P. E. Dant Reinstate Monica Oct 25 '16 at 23:33
  • -1 for the question "He is buried" ... but it's written for a different audience than most of our community. I disagree hugely. It makes it sound as if it is elitist. As if learners are incapable of understanding that question or any of the answers posted. – Mari-Lou A Oct 25 '16 at 23:48
  • The interested community was the Hot Questions Network, there are many SE meta posts complaining about this phenomenon on SE. Again, I repeat, I am very pleased by the number of views the question attracted, 1,203 so far, and the the two good answers it received. I was perfectly satisfied until I read your answer. – Mari-Lou A Oct 25 '16 at 23:52
  • @Mari-LouA It's not a bad thing to write a question more interesting to our fluent/advanced learners. I even said that it wasn't my intention to disparage the question. I up-voted your question because I thought it was interesting and that the discussion of "is buried" compared with "has been buried" or "was buried" would be helpful. Diversity is good. I think we could use more "advanced" type questions on ELL. Obviously I didn't communicate well because it wasn't my intention to make you less satisfied with your question. – ColleenV Oct 26 '16 at 3:23
  • @Mari-LouA I think you interpret "written for a different audience than most of our community" inaccurately as pejorative, but still, I might have phrased that differently myself (only, of course, after having been advised to reconsider my phrasing by helpful commentary.) – P. E. Dant Reinstate Monica Oct 26 '16 at 4:07
  • It's not your upvotes that are the main concern here. Not even my upvotes, as a rash ELL user. And NOT EVEN any ELL user's upvotes. The main problem is people coming from other sites that know nothing of our voting conventions and just pile up using their assoc. bonus because a popular question is interesting. And the problem rises because then the score is not an indicator of quality like it should be, but an indicator of popularity. – M.A.R. Oct 26 '16 at 17:21
  • @M.A.R. I don't agree that "quality" is more important than "interestingness". Looking over our top voted questions, I don't really see anything too terrible going on with the questions that have risen to the top. I ended up here because a HNQ caught my eye - maybe some of y'all don't think that's a good thing :) - but drawing new folks into the community is good for ELL and a question getting a few too many votes isn't that high a price to pay IMO. Score can't be a measure of quality. There are no consequences for UVing a question that is 'low quality' nor any normalization. – ColleenV Oct 26 '16 at 19:12
  • When someone begins to say "I don't want to sound sexist, but..." or "I'm not a racist, but..." you can be sure what follows is a sexist or a racist comment. So beginning with I don't want to disparage the "He is buried" question ... but ... really means that you don't think much of it. Which is fine, I don't expect everyone to "like" every English language question I post. – Mari-Lou A Oct 27 '16 at 5:06
  • Nevertheless, the views it attracted is remarkably high for this site, and all merit must go to Colin Fine's clear, and well argued answer which has earned a healthy 14 upvotes. I believe those upvotes reflect the ELL community. I am sure that the majority of those upvotes came from learners whilst others from native speakers who are geeky about grammar. – Mari-Lou A Oct 27 '16 at 5:07
  • Colleen I never said 'quality' is more important. I said that votes are ideally indicators of quality, not popularity. – M.A.R. Oct 27 '16 at 7:33
  • @Mari-LouA I didn't write "but...". I wrote. "I don't want to disparage the "He is buried" question. I think it is a good quality question..." I up-voted it and would like to see more questions like it! I don't understand why I'm not able to communicate that to you. – ColleenV Oct 27 '16 at 11:06
  • @M.A.R. Why would score ideally be a measure of quality over popularity/interest if quality weren't more important? – ColleenV Oct 27 '16 at 11:07
  • Per definition. Votes are meant to indicate 'usefulness', and that usually translates to quality rather than popularity. – M.A.R. Oct 27 '16 at 11:12
  • @M.A.R. Usefulness does not equal quality. If there is a question about the same sentence that confused me in a story that I'm currently reading, that question is really useful to me even though it may not link the source or include a lot of context. I'm going to up-vote it because I want to see an answer to it and I think that's OK. – ColleenV Oct 27 '16 at 11:23
  • I never said they're equal. In your example, if the question you want to upvote is txtspk mumbo jumbo no one can understand, you will not upvote it, will you? When you want to upvote it, it doesn't matter to you whether it's a popular question or not. Yes, usefulness is not the same as quality. But the deciding factor in deciding usefulness is quality. – M.A.R. Oct 27 '16 at 11:43

You must log in to answer this question.

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