31
\$\begingroup\$

I've noticed many challenges getting to the top of Hot Network Questions because they are really easy and get many answers that feed the activity-rewarding hotness formula. In turn, the visibility attracts more answers and more votes and keeps the cycle going. Can we do something about this?

Some are poor questions, with downvotes for significant issues that are flooded out by upvotes when the challenge reaches the top of the front page. Some are simply a race to be the first to post in your language. This is not to say that all easy or catalog-style questions are bad, but that bad ones evade community judgment.

The result is that we advertise a skewed selection of our questions to the SE community and to those browsing our top-voted recent questions. It creates the impression that our community is about simple, short, one-liner challenges where the goal is just to collect answers in as many languages as possible. In turn, that's what new members are likely to look for and post.

Moreover, the rep incentives for easy questions are totally wrong. Simple golfs that you write in five minutes should give you less rep, not astronomically more. Likewise for writing challenges.

The effect of HNQ on communities has been well-discussed on the main meta, and there doesn't seem to be any technical change coming. Is there anything we can do on our end to improve this situation?

\$\endgroup\$
16
\$\begingroup\$

I think Trichoplax answered this pretty well, so I'll just address one aspect:

Some are poor questions, with downvotes for significant issues that are flooded out by upvotes when the challenge reaches the top of the front page.

If the question has significant issues, close it. I'm not saying don't also downvote it, but downvotes are too easy to overcome by the flood, as you mention. Closing it (or "putting it on hold" as we now say) should take it off the HNQ completely until any issues are resolved and it gets reopened. "Significant issues" is basically the entire reason for putting something on hold.

It's not impossible that it gets reopened before you think the issues are resolved, but at least it can only be reopened by people who have some rep in the community, rather than random visitors from the network who can upvote.

\$\endgroup\$
  • 13
    \$\begingroup\$ I have seen a question which hit HNQ before it was closed for an unfixable lack-of-objectivity problem get reopened with a vote from someone who wouldn't have had enough rep to vote to reopen if not for their answer on that same question. The system more or less works in most cases, but HNQ amplifies its imperfections. \$\endgroup\$ – Peter Taylor Oct 11 '15 at 7:14
8
\$\begingroup\$

Making downvotes count for more

Downvotes could be weighted more heavily in the formula (at present one downvote is worth the same as one upvote). I would guess that this is unlikely to be done unless it is agreed to do it network-wide. This only helps with the problem of poor quality questions reaching HNQ. It does nothing for the many questions that reach HNQ without a single downvote, as they are good questions just not time consuming to answer. Whether such questions should be the main advert for our site is likely to divide opinion, and including downvotes in the discussion makes it harder to judge the split of that opinion. Hopefully answers here can address those two issues independently.

Excluding specific tags

It may be possible to stop certain tags from appearing in the HNQ list. There is already a list of words that are not permitted in titles in HNQ. Questions with a title containing such a word simply do not appear. However, the forbidden words are network-wide, whereas excluding a tag would be site specific. I'm not sure if this would be a problem, but if the tag we want to exclude never appears on other sites then it should be easier.

An example of a tag that makes a question likely to reach HNQ is . However, to me that tag seems like one which it makes sense to keep on HNQ since the high number of answers does not indicate low quality, nor attract users that would skew our voting and erode our scope definition.

Opting out of Hot Network Questions

Each site has a manually adjustable weighting that is currently used to adjust for the traffic level of a site (so that HNQ doesn't fill up with only Stack Overflow questions). Since this is site specific it might be easier to get this changed. It won't correct for any of the perceived biases but it could be used to simply opt out of appearing on HNQ as a site, if the community decided that was best. We'd still need to convince the Community Managers that this is also not detrimental to other sites in the network (there may be an argument that the high variety of sites in the HNQ side bar is healthy for SE as a network).

Is there really a problem?

Although it's good to ask the question, and good to see possible things that could change, I think it makes sense to gather meaningful data before making any decisions. I'd like to see suggestions of what statistics would best help decide whether there is a problem.

It's informative to look through the various discussions that have taken place on Mother Meta. There are many different opinions expressed and many of them sound plausible even though they contradict each other.

What I'd like to see is a HNQ history. The ability to call up statistics on questions based on the maximum hotness score they achieved, the length of time they spent in the HNQ list, and the voting data showing the proportions of votes that came from users at different rep levels (including votes from passing users who had never gained rep on the site beforehand). Even better would be the ability to see what proportion of users new to the site went on to gain rep there later.

I don't know how much of this could be made available without starting to unintentionally open routes to mining data on specific users, so this might need to be a mod only reporting tool (or even a community manager only tool) but I think its existence would be beneficial to everyone, regardless of who happens to be using it.

Easy and quick are not synonyms

It's possible to have a question requiring very little skill or innovation, but still taking a long time to write an answer for. Equally it's possible to have a question which attracts answers very quickly, but mainly from ingenious users who happen to be able to solve tough problems at high pace.

A question being in HNQ suggests that it is quick to answer. Whether it is also easy to answer will vary from question to question, and what the balance is would be interesting to try and measure. In the absence of definitive stats, I'd be interested to hear if anyone has examples from either side: HNQ questions that were easy and/or answerable by visitors, and HNQ questions that were hard and/or answerable mainly by existing users, or even only by a select few existing users.

\$\endgroup\$
  • 1
    \$\begingroup\$ An example -- Print out all IPv6 addresses is (at the time of this writing) a HNQ. Of all the answers, there are two or three from new/visiting users, while the rest either have hundreds of rep or have been here years (or both). \$\endgroup\$ – AdmBorkBork Oct 9 '15 at 13:21
  • 1
    \$\begingroup\$ Alternatively, Don't google "google" absolutely exploded on HNQ, and that's a good example of something that's easily answerable by visitors. \$\endgroup\$ – AdmBorkBork Oct 9 '15 at 13:23
  • 4
    \$\begingroup\$ Also note that we get a lot of traffic from other sites in the SE network, presumably due in no small part to our presence on HNQ. If we opt out of appearing on HNQ, we may lose out on a lot of traffic, which could be quite detrimental to the health of the site. Besides, we get some decent users who first heard about us from HNQ--like me! :D \$\endgroup\$ – Alex A. Oct 9 '15 at 20:45
  • 1
    \$\begingroup\$ @AlexA. I also arrived via HNQ - so it's not all good ;) \$\endgroup\$ – trichoplax Oct 10 '15 at 8:53

You must log in to answer this question.

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