6

I'd like to present for comment a process by which we might productively handle requests for lists of reference material on the History stack using the current StackExchange software.

Requests for references are a common type of question people would like to ask on the History stack. This shouldn't be surprising, as the study of History is all about reading and analyzing references. Unfortunately, the StackExchange system is not well designed for this type of question, so these often get closed, as much as many users here would like to answer those questions.

Some stacks allow questions like this, but an examination of the results show that the content doesn't end up being very useful (low votes, seemingly capricious acceptances). What is needed is a way within the system to set up such questions in a way that does encourage the production of useful content.


Under the proposed Reference List Question process, when a question that qualifies as a request for a list of references or books is submitted by a user, the following will happen:

  1. Someone1 notices the nature of the question, and adds the tag .
  2. Someone creates a community wiki answer.
  3. Someone adds a comment to the question giving the questioner the option of either accepting the community wiki answer, or having the question held until they can rephrase it to no longer qualify as a list of references. The comment should reference this meta post for a fuller explanation to the original poster.
  4. Users who wish to answer at this point may do so (with normal non-wiki answers).
  5. Any references listed in a new answer will also be added to the accepted wiki answer by someone (preferably the person who posted the answer), with no commentary.

1 -"Someone", in true StackExchange fashion, means any user who happens to notice that it needs doing and has the requisite reputation to perform this action. Users who notice and don't have the needed reputation may flag the item to request it get done.


Advantages of this approach:

  • The compilation wiki answer puts the complete list of mentioned references at the top of the answers (regardless of voting activity). Being a wiki, no user will get reputation credit for it, as is right for a compilation of everyone's answers.
  • The other answers, all sharing the property of not being accepted, will be automatically sorted via the voting process.
  • Users who post answers will get reputation credit for the quality of their answers as per normal.
  • A random web surfer hitting our page should find first the question, second a complete list of references, and following that answers with references and explanations, sorted by quality (votes).

In short, I believe it sets up a system where the existing SE engine will provide the proper incentive for good useful answers.

Drawbacks of this approach:

  • It requires an extra step from the person asking the question. Often such people are brand-new users who are likely to have more trouble than most comprehending jargon like "accept button" and "wiki answer".
  • It requires extra moderation work from alert and knowledgeable users.
  • It doesn't solve the fundamental issue of such questions being solicitations of personal opinion.
  • Its a big experiment. As far as I know, no stack has done something like this. It seems to me it ought to work, but nobody knows how it will perform in practice.
  • This seems like too much work for very little benefit. Most reference requests I happened to see on the site are either extremely broad, or quite vague, or both. Do you have any example questions that you think would have produced good answers, if we had followed your approach instead of closing them? – yannis Feb 10 '15 at 12:22
  • Also, there's a special lock ("Wiki Answer") that only allows for a single answer we might use. You can see it in action here (that question has 87 deleted answers - most were deleted after they were incorporated into the wiki answer). However, the lock is only available to diamond moderators and it only makes sense after everything valuable in the answers has been moved into the main one. – yannis Feb 10 '15 at 12:30
2

I believe that we should experiment to determine whether reference requests fit History.se or whether the current opinion that they be rejected is correct.

This sounds like an excellent and well thought experiment, that we should conduct, and then review in 6 months in order to finalise policy.

  • I make this answer, noting, that I oppose reference requests. – Samuel Russell Feb 10 '15 at 2:16
1

I would be 100% for this proposal if reputation could be taken out of the equation. As is, I worry that SE's reputation system would pervert incentives too much.

What Does OP Want From a Recommendation Thread?

A list of only the best books. We all know that there are more good history books than there are days left in our lives, so we only want to read what are truly the best.

What Does an Answerer Do If There is No Reputation To Be Gained?

Recommend a book that they absolutely love. This matches up well with with OP's wants.

What Does an Answerer Do If There is Reputation To Be Gained?

Unfortunately, I think the system breaks down here because we are basically setting up a Keynesian Beauty Contest. There's going to be a mad race to be "first" when someone asks for recommendations about the Roman Empire, because answering "Edward Gibbon" might net you 400 easy points. But even after all of the big names are taken, some users will be incentivized to search Amazon for a book on Rome that hasn't made the list yet, and they'll add it hoping to get a few easy upvotes.

Maybe this is contradictory of me because I think the reputation system, though not perfect, generally creates healthy incentives for SE. I think that's because normally, the quality of our answers themselves is being judged--wanting an extra upvote pushes you to do extra research, extra proofing, extra thinking. But for book recommendations, the referent of our answers is being judged. You can't write a better answer (Edward Gibbon is already as good as he is going to get), so wanting an extra upvote pushes you to post additional recommendations. So I both worry that signal will get lost in the noise, and that some of these threads will become de facto pinned to the top of "active questions."

The only way to take reputation out of the equation would be to force all OPs to make their questions community wikis (which forces all answers to be community wikis) or to host these questions in Meta. (I know the latter isn't practical, but I'd actually like it because then we could freely downvote books we don't like without costing others rep.)

It's possible that community pressures would keep everyone in check, or maybe I'm too concerned that some users would try to game the system. I'm not opposed to experimenting with the proposal, but these are my initial concerns.

  • As an example for you, I know from past experience that whenever I recommend Zinn for anything, I'm pretty much guaranteed lots of downvotes. So I have seen this exact process you are worried about. However, that doesn't really stop me from recommending Zinn. So either the -2's aren't enough disincentive for me when I think I'm right, or I'm just stupid. :-) – T.E.D. Feb 11 '15 at 0:03
  • @T.E.D.: Haha, I can imagine you'd need to be a glutton for punishment to recommend Zinn on this site. – two sheds Feb 11 '15 at 1:01

You must log in to answer this question.

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