16
$\begingroup$

Recently I came across with a question that was asking for book recommendation in an expertise area related to statistic. Then, I was checking for duplicates and I realized among all book recommendation questions some of them were community wiki (e.g. 2,3,4) and some were not (e.g.,5.6.7).

Logically, among the examples given above, the group of question classified as community wiki were the ones more upvoted, edited, etc, according to Cross Validated SE rules stated here.

Although there are specific rules to be followed regarding community wiki classification, there are also flexibility which I understand these question fits. For flexibility I mean the quotation bellow:

Moderators can also choose to convert posts into community wiki mode if they feel it is appropriate for the question or answer. Once a post is made community wiki through any of the above ways, that mode cannot be removed (except through moderator intervention in extreme circumstances).

Finally, once these type of questions have very similar objectives, do you think it is a good idea to standardize all of them to have community wiki status?

On the other hand, I know book recommendation questions are considered off topic in many SE sites (e.g.: Arqade, Ask Ubuntu, Engineering, Programmers, Server Fault, StackOverflow, SuperUser, Travel, among others). We have a different theme here in CV, but maybe their arguments are valid for us.

$\endgroup$
  • 6
    $\begingroup$ +1 Thanks for bringing this up and expressing it so thoughtfully. Moderators do not manage to see every post--I don't recall seeing any of the three that are not CW--and so we have to rely on community flags to suggest that some posts ought to be CW. SE has sent out mixed signals about CW status; recently they have been discouraging its use. $\endgroup$ – whuber Apr 17 '13 at 21:49
  • 1
    $\begingroup$ Your examples of threads that were not CW were still not CW until yesterday - that's four years later! I flagged them all for mod attention suggesting to convert to CW, and they were all converted to CW with my flags being marked as "helpful". So I believe this shows that the current (2017) guidelines are still that such threads should be CW. $\endgroup$ – amoeba Apr 18 '17 at 19:56
13
$\begingroup$

To answer the title question: Yes, all such questions should be community wiki.


Proposal

I propose the creation of an exception rule for questions about book recommendations, similar to the one created for What is your favorite “data analysis” cartoon? and Famous statistician quotes.

To ensure a certain quality, the exception rule can only be applied if the following conditions are met (needs discussion)

  • An answer has to provide a at least one paragraph long founded justification why the book is recommended. Answers simply naming the title or stating unfounded reasons ("I liked it") have to be deleted.
  • Beware of duplicates. Identify them, then merge or delete them without mercy. If possible, merge related questions into a bigger one by broadening the scope of the question to avoid one question for every pair X,Y,Z where the question pattern is I have expertise in y for x years, which book do you recommend for z
  • All this type of questions needs to be community wiki to prevent the reputation-only-motivation to answer this type of questions

Reasoning

Recommendations from experts matter

No doubt, book recommendations from experts in the particular area are far more valuable than the ones provided by John Smith (no offense to all John Smiths) or the number of the times the book has been bought.

To get such recommendations, one needs ...

  • a site, where experts do contribute
  • a way to somehow measure the degree of expertise to assign a subjective degree of belief to the recommendations of the recommendation provider

Accidentically, stats.SE has both. On the other hand, the review system of a big name book seller, although it might be useful in general, loses more and more of its value if the area of interest gets more and more narrow. Less reviews, less ratings of reviews, etc..

Misuse of community wiki

We all know: The quality of this site heavily depends on the users learning, accepting and living the rules which have lead to the quality. If the number of users who do not care about the quality in general (!) gets too large, the site is doomed, especially when users with a good amount of reputation do violate the rules in order to farm reputation on their own, hence being a bad example for new users.

IMHO, the current sense of quality in the users with a good amount of reputation on stats.SE is great. On the other hand, stackoverflow (again, only my opinion) is already a big pile of mess with gems hidden in it. I am still able to find the gems, but only due to the search- and "sort by vote"-function. I cannot count the duplicates or the number of times I have stumbeld upon high-level users violating the rules discussed in meta, so I guess that one can get the deputy badge within a day without ease.

So I suspect, that stackoverflow, as the flagship of the SE-sites with the most traffic, requires stricter rules to be sustainable.

Summary

So: While I agree with most of Robert's arguments, I do not think that such a harsh treatment of book-recommendations questions on stats.SE is necessary .... yet. If the quality of the site starts to decrease due to heavy misuse we might go back and enforce all these rules. Until then, we should not delete or prevent content which provides actual value.

$\endgroup$
11
$\begingroup$

I don't agree that book recommendations are a good use for this site — but let's set that aside for a moment to avoid using Community Wiki as a way to make these somehow okay.

Community Wiki was not designed to solicit crowd-sourced lists of stuff. The problem with Community Wiki is it's often used to to allow questions that otherwise would not be allowed on the site. That's why we largely deprecated the feature — Future of Community Wiki.

The problem with most of these "let's create a list" questions is that it becomes really hard to begrudge anyone their addition, and every suggestion of a random favorite book/tool/application/website/blog/etc is no better than any other, until all semblance of expertise is lost.

In this question (for example): What is the best introductory Bayesian statistics textbook, 26 answers citing the "best textbook" (really?), and the voting has a strong correlation to the age of the answer. More views = more votes. That's not exactly what we're supposed to do here.

There will always be users who prefer to just ask the easy "getting started primers" — It's just easier to cast a broad net and pick through the rubble of random answers to see what suits them later. But let's start with the premise that, with overly-broad, generalized questions, users will only be guessing how to help them, specifically.

This site will work better if you help these users flesh out a question about what specific problem they are encountering and what they are having trouble understanding in their day to day work. Specific, long-tailed questions will work much better than "you guys talk and I'll pick out the good stuff later."

It's not like these polls of the community should never be asked anywhere (discussion forums are great for this stuff), but what makes a Stack Exchange site "work" is its laser-sharp focus on specific, long-tailed questions at can be answered canonically. We stick to the practical, answerable questions based on actual problems that we face in our day-to-day work. That's a pretty big ask, I know. But specifically forgoing these types of overly-generalized questions is how we set ourselves apart from every other forum on the topic. It's a tough sell, but it's that practical, low-noise approach to Q&A that drives experts to these sites.

And let's not (mis)use the Community Wiki feature to fix these questions.

$\endgroup$
  • $\begingroup$ @Robert, as the asker of one of these book recommendation questions stats.stackexchange.com/questions/56385/…, was this an appropriate use of CV? I tried to give as much detail about my specific situation as possible, but I admit that I may have been in error. $\endgroup$ – Patrick S. Forscher Apr 18 '13 at 3:00
4
$\begingroup$

There's a third approach, which is at least worth presenting: such posts could be treated just like others—recommendations of a statistical method, say—, rather than like polls. A great answer might deal with several notable textbooks; discussing what they cover, & omit; their statistical prerequistes; their mathematical level; their strengths & weaknesses in exposition. A poor answer might just name-check a popular book—and would be deleted if not improved.

$\endgroup$
  • 1
    $\begingroup$ +1. I very much support this attitude, and I have a problem with our liberal CW usage in general (stats.meta.stackexchange.com/questions/2953). $\endgroup$ – amoeba Apr 20 '17 at 9:45
  • $\begingroup$ @amoeba: Well, it'd be more work, more friction. And there's nothing to stop people voting for the book they like the most rather than the most useful answer. $\endgroup$ – Scortchi Apr 20 '17 at 9:51
  • $\begingroup$ That would be great, but I see people already can post such types of answer, but it is not common. I also think what you propose partially overlaps to what @steffen says in his first bullet (in proposal). To me, the next step to what we already have would be to make these types of questions (shopping lists, polls, etc) to be off-topic. It would solve the overuse of CW, and would automatically demand from OPs to ask a more elaborated and specific question (eg: stats.stackexchange.com/questions/116725/…). $\endgroup$ – Andre Silva Apr 20 '17 at 14:00

You must log in to answer this question.

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