9
$\begingroup$

TL;DR The system promotes popular questions, but popular might not need to align with characteristics that are desirable. Especially complex and difficult questions might be very desirable, but difficult to showcase in a popular way. If we would want to changes this, then how can we do this?


I saw among the questions with bounties a question that interested me:

How is EMSE derived for causal trees in Athey and Imbens (PNAS 2016)?

But I had a hard time to follow it (a bunch of formula's that don't make immediately sense to me, to be fair in light of my further "complaint", these type of questions might be improved). It requires either previous knowledge or some amount of research effort to understand the topic.

The question had an answer, but I don't understand it enough at first glance in order to give it an up-vote. (and while working on my second glance, I think that I will only get it during my third or higher try).

I imagine that this is the case for many other questions and many other people. Many interesting questions don't get attention, because we have a voting system that punishes them.

I know for my personal situation that some of the answers that I value myself as very good, are not very well received. And at the same time some answers that I value myself as very bad (or not much special) are very well received.

The consequence is that these type of posts that might be very good (although difficult to understand by most people), but receive much less attention (measured by up-votes, which is a system that has a strong positive feedback loop as popularity leads to more votes and popularity is acquired by votes).

The consequence of the up-vote system is that topics that are less popular or known, will receive less up-votes (obviously). It is questionable whether we want 'popular' votes to dominate the presence on the website. Those popular questions are often the simple questions (or the ones that got a boost via the 'hot network questions') and maybe, the network should turn towards a direction of improving the quality of more sophisticated questions and answers? (maybe the stackexchange software neural network that selects these hot topics should use a bit more randomness like adding an extra dropout layer? Or use a different way to evaluate the 'success' of it's actions.)

I am not sure whether it is possible to turn the direction of the website, and whether this is even desirable. Maybe answering simple questions is the formula that works well. And, it can be an interesting endeavor to make slightly simple questions even more simple with good answers (that's the area where I am placed, I can only hit a few of the difficult topics). But, the popularity of the simpler questions is overshadowing the more difficult questions, and I have a feeling that this doesn't value those questions in the right way. Popularity is not the sole measure of relevance, and certainly not of quality. If we want to improve quality, then maybe we should have more ways beyond the voting system that relies entirely on (self-reinforcing) popularity.

$\endgroup$
1
  • 2
    $\begingroup$ The primary mechanism for this is going to be adding bounties. Note further that while our chat is not for asking questions, most of our highest rep users check it, so you could mention in chat that you are placing a bounty on a particular, and difficult, question to bring more attention to it. Those actions would drive more attention. That's about all we can do. $\endgroup$ Jan 19 at 12:23

1 Answer 1

6
$\begingroup$

I have been here for not quite long but I do have a taste of what HNQ does. So, count me in among the league of those users who feel exasperated when an HNQ bloats a mediocre question and it gets unnecessary number of upvotes.

But before investing in too much, let's take a stroll. HNQ has been the centerpiece of debate among many of our sister sites and Meta since time immemorial, mostly for the reason you articulated. (See Dealing with low-quality questions that end up on Hot Network Questions, Prevent questions on the Hot List from being upvoted by casual visitors (only rep is from association bonus), What is the Goal of "Hot Network Questions"?, What is the goal of having questions from Math.SE in the SE Network Hot Questions List?, Do Physics Hot Network Questions help or harm the site? etc.)

Few points are noteworthy:

$\bullet$ The intention of HNQ is to advocate the field among the wider community to popularize the field. To evoke interest among the non-users and lure those who aren't yet in the community. (And entertainment.)

$\bullet$ In pursuit of those, we witnessed not every time "popular" questions are synonymous to well-researched sophisticated conceptual/computational query that practitioners, veterans and students of that field would like at the outset.

$\bullet$ There is the assertion that casual users who join only after viewing the HNQ would flood the posts with upvotes and so there a proposal was raised to prevent that. But it didn't go anywhere. IMO, it can be a reasonable issue to ponder about but would not jump to the conclusion that this is the sole issue.

$\bullet$ Not every time HNQ posts are low-quality. Sometimes even though the original query is mediocre, the answers are are broadly appreciable (eg. Ben's post on possibility theory).

Crux is that while HNQ can be flawed, it's not outright useless. How it can be fixed and any other issue are, however, not the relevant queries to be addressed here - those are for another debate.

Coming to the main query: how can we improve attention to the not so popular and yet the more useful, more conceptual and more intricate posts that are lurking in the deep vault of CV?

Here are my takes:

$\bullet$ Advocate our site among your peers! Seriously, I urge you and all comrades of this community to advocate and disseminate the site among your colleagues, your social media influence sphere. Invite the professionals you are acquainted with. Invite your professors here. Draw in more professionals who can appreciate the genuine posts.

CV regularly tweets posts (different from HNQ) in its handle. Share those among your followers. We do have an envious number of veteran practitioners and big guns here. But we want way more.

$\bullet$ Award those unsung posts by putting bounty. And why not? What are these reps worth for if not for uplifting other good posts? If you have come across any post that you feel the user has toiled and sewed meticulously to concoct a well-researched post, then reward it! Questions but not answered yet? Put bounty. Bring it to the center stage. Well-researched formal answers, properly cited answers, nitpicking yet formal derivations that were written years ago but only received paltry votes? Reward them! By doing so, you are necessarily infusing them with new lives. The posts would be revived again to be viewed by the present community.

$\bullet$ This is more of a speculative lament: Adopt practices that potentially can make well-researched posts to HNQ. Unfortunately, I haven't yet delved into what makes a post to HNQ apart from some quick votes. If it is so, then so be it. Quickly upvote those good piece of questions that seem to be very well-researched and conceptually rich and intricate.

But imo, the goal should be to rope in more veteran practitioners, professors and other advanced people who aren't in our site yet. After all, we are here to make a knowledgeable repository of statistics and that should include those advanced posts that otherwise get snubbed thanks to handful few users who could appreciate the worth or have the knowledge firsthand to respond. Increasing the knowledgeable base here would certainly be beneficial.

This is, again, my humble opinion only. Perhaps let me stay here for a year or two to acquire more experience to provide a better commentary but for now these are the options that I can conjure up.

$\endgroup$

You must log in to answer this question.

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