Yes, we've had this debate before. Unfortunately, there's no definite consensus to be found about their community wiki status:

  • Good list, bad list (from July 2013) has the top voted answer (with a meager +11/-2) saying that book recommendation questions need to be actively moderated by the community and should be made community wiki.

  • The faq post Are resource recommendations allowed? (from August 2013) implements the "Good list, bad list" top scored answer by setting rules for what a good res. rec. answer must contain, and that all answers prior to that policy should be moved to one single community wiki answer that gets an additional notice. No mention of making the question or its policy-conforming answers community wiki is made.

  • The question Do we need/want an overarching books question? (from December 2013, about the great book list) has no answers of positive score. The great book list itself is closed and has a single large community wiki answer that does not conform to the answer policy, but also lacks the additional notice mentioned in the faq post.

So, already the three main questions I can find on the book recommendation questions disagree on how to exactly handle these questions, or have just largely been ignored. It gets even more confusing when looking into practice:

  • We currently have 762 questions tagged with . Many of them are closed as duplicate of the big book list, others as primarily opinion-based, but only 174 of them have been made community wiki. No pattern as to which ones are made wiki and which ones are not is evident to me.1

  • No pattern as to which ones are closed as duplicates of the book list, which ones are closed as primarily opinion-based and which ones are left open is readily apparent, although the ones to be closed as duplicates of the book list tend to be rather unspecific.

  • There are 1.063 answers to questions, of which 544 are community wiki. This indicates that it's mostly the closed questions which are not made community wiki, because the percentage of answers that are wiki is much higher than that of the questions. But again, no clear policy is apparent. Also, I have been unable to find the "additional notice" from the faq post on any old answers.

  • The amount of answers conforming to the policy set in the faq post is surely debatable, but I'd say at least some of these answers are too short to be considered descriptive evaluations of the books in question.

So, it appears no clear policy exists, or at least, none is enforced in practice. The topic of this meta discussion is therefore:

Should resource recommendation questions be on-topic, on-topic but community wiki, or off-topic?2

1To search for wiki/non-wiki questions, add wiki:yes or wiki:no to your query. Combine with is:q or is:a to get only questions or only answers.

2I have purposefully tried to refrain from making an argument for either case in the question, so that people can vote on the respective answers with their argumentation as they come in.

  • $\begingroup$ Note: Particularly for the case of on-topicness, an additional meta discussion on how to curate the questions and enforce the policy is likely to follow. $\endgroup$ – ACuriousMind Sep 9 '15 at 22:10
  • 2
    $\begingroup$ I'd clarify one thing: the master book list is a unique special case, and nothing about it should be used as an example for how we handle resource recommendation questions in general. Technically it's not actually a resource-recommendation question, but it has the tag because of its close association with resource recommendation questions. $\endgroup$ – David Z Sep 9 '15 at 22:27
  • $\begingroup$ Related post on Cross Validated: stats.meta.stackexchange.com/q/1593 $\endgroup$ – Qmechanic Apr 24 '17 at 7:08

Resource recommendation questions should be on-topic, but community wiki (and with the same expectation for the content of the answers as currently set in the post). The post should be altered to reflect that both old and new questions should be turned into community wiki posts, as indicated in the top-voted answer of "Good list, bad list".

Resource recommendation questions cost no effort to ask. There is not much research one can do for them (since disallowing certain recommendations just because asker has already looked at them would be very strange, considering questions should not merely be about helping the asker, but anyone finding the question). One can't write them particularly well, since they're always going to follow the same scheme. Upvotes will indicate the popularity of the topic that is being asked about, but nothing else. For this, one should not get or lose reputation.

Resource recommendation answers are easy to write - pick your favourite book on the subject and describe why you like it, what it contains, and why it is appropriate for the topic being asked about. Expertise in the topic would be ideal, but is in no way required for recommending a book. Upvotes and downvotes indicate not "factual correctness" of the answer nor the effort poured into it, but how much the userbase likes or dislikes the particular book being recommended. One should not gain or lose reputation based on how much other people like one particular book.

Nevertheless, resource recommendation questions provide a valuable...resource for people trying to find books or other sources on specific topics. Unlike reviews at book stores or the like, Physics.SE provides the unique possibility to source those recommendations from a community of physicists. This is why they should not be entirely off-topic.

If we would take a principled stance against making any questions CW by policy (which this SE blog post also linked by DavidZ seems to indicate), then resource recommendation questions should be off-topic, since for the reasons above one should not gain or lose reputation from them. However, this policy, though talked about in the blog, is not implemented network-wide: At least MathOverflow still habitually makes questions community wiki which are broad or opinion-based, but nevertheless considered worthwhile by the community.

  • 1
    $\begingroup$ To be more specific: are you suggesting that the 'dealing with old recommendations posts' section of the faq post be altered to recommend that they be turned into CW? (I would recommend that you did!) $\endgroup$ – Emilio Pisanty Sep 10 '15 at 12:09
  • $\begingroup$ @EmilioPisanty: Not only that, but also that the faq states that new res. rec. questions should be CW (it currently doesn't, unless I can't read). The problem is that 'Good list, bad list' says that (and that some (I think the majority, but not all) new res. rec. questions have indeed been turned CW), while the policy in the faq post doesn't say that. $\endgroup$ – ACuriousMind Sep 10 '15 at 12:37
  • 1
    $\begingroup$ As in: I'd +1 this again if you explicitly said that in the post. And if I could vote twice, that too. $\endgroup$ – Emilio Pisanty Sep 10 '15 at 13:11
  • 3
    $\begingroup$ This is the one answer choice I very strongly disagree with. Personally, I'd kind of rather recommendation questions be off topic (with that information going in the tag wikis). The community seems to want them on topic, which I'm okay with as long as they're not CW. But I'm absolutely dead-set against on-topic-but-CW. (I especially disapprove of using the logic that another site does it, therefore we should do it.) $\endgroup$ – David Z Sep 10 '15 at 13:20
  • 1
    $\begingroup$ @DavidZ This answer provides an eloquent argument for why, in general, resource recommendation questions should not accrue reputation. Would you care to argue in more detail why you seem to think of CW as the devil incarnate? (OK, maybe not that much.) $\endgroup$ – Emilio Pisanty Sep 10 '15 at 13:42
  • 1
    $\begingroup$ It's outlined in my answer, but the gist is that I do not agree with the reasoning for why people don't deserve reputation for these questions. Asking a question which helps many people deserves the reward of reputation, IMO. More generally, I don't believe there's any such thing as a useful question which is undeserving of reputation. $\endgroup$ – David Z Sep 10 '15 at 13:47
  • 2
    $\begingroup$ @DavidZ It's not really the question that matters - it's the answers. It's less bad than before but it's still the fastest-gun-to-mention-Feynman that will get the most votes. Voting is (and should) be driven by the quality of the resources referenced, and only secondarily by the quality of the reference itself. Such answers shouldn't need their posters to voluntarily CW them - it's implicit in the format. If the asker's rep also goes, then that's an acceptable casualty, I feel. $\endgroup$ – Emilio Pisanty Sep 11 '15 at 11:11
  • $\begingroup$ @Emilio I consider recommending a good resource to be itself deserving of reputation, assuming that the recommendation has some thought in it (which is part of the logic behind forbidding "link-only" recommendations). It's on par with writing a good answer to any other question. So I don't support forcing CW mode on the answers. (Of course, the fact that we're having this discussion in the first place I would offer as evidence of why we shouldn't have recommendation questions at all, but given that they're going to be allowed, these are my arguments for why they should at least not be CW.) $\endgroup$ – David Z Sep 11 '15 at 11:15

First of all, I think there is a clear precedence order among these posts. I may as well give the full ordering as I see it:

  1. The help center
  2. The canonical (generally, top) answer in an post on this meta
  3. The canonical answer in an faq post on Meta Stack Exchange
  4. If there is not an FAQ post on the topic, answers in non-FAQ posts on this meta, with answers to later questions overriding those to earlier questions
  5. As in #4, but for answers on Meta Stack Exchange

Recommendations and discussions on the Stack Exchange blog, and any other sources that come directly from the SE team, don't really set policy, but they should be taken into account when interpreting policies or deciding how to handle situations that aren't explicitly covered by an FAQ post. If I had to place them in the list I'd put them roughly between steps 3 and 4, but in a fuzzy way that indicates their influence on all five levels.

Out of the questions you linked, Are resource recommendations allowed? is the only one with the tag, and thus (as far as I'm concerned) it sets the site policy on the matter, since there's nothing relevant in the help center. It says nothing about making resource recommendation questions community wiki, therefore I don't make them community wiki. If we want to change that, we should build a consensus as a community (as we might do with this question) and then edit (or, in more extreme cases, replace) the relevant post.

But I don't think we should change that. The reason I hold that view is fairly well summarized by this SE blog post (also see this more recent one which continues the same argument). In short, there should be no such thing as a question which is good enough to be on the site, but not good enough to earn reputation for its asker. We want to host good questions and only good questions, and it makes sense to reward people for asking good questions. That is what reputation is for. Take away reputation, and you take away the incentive to keep question quality high.

TL;DR book recommendations should be on topic (subject to guidelines) and not community wiki

Now, we do have some inconsistencies in how this is handled in practice, as you pointed out. (The following is to be considered descriptive and not part of the recommendation that constitutes the first part of my post)

...No pattern as to which ones are made wiki and which ones are not is evident to me.

Probably has more to do with which moderator took action on the post than anything else. We haven't gotten together to hash out an agreement about whether those posts should be wikified or not. Actually until recently I didn't know that anyone was still making questions CW.

No pattern as to which ones are closed as duplicates of the book list, which ones are closed as primarily opinion-based and which ones are left open is readily apparent, although the ones to be closed as duplicates of the book list tend to be rather unspecific.

I personally close questions as duplicates of the book list when they're asking for books that would also be listed under more than one of the "standard" book recommendation questions.

There are 1.063 answers to resource-recommendation questions, of which 544 are community wiki.... Also, I have been unable to find the "additional notice" from the faq post on any old answers.

Answers are an entirely different issue. Posters still have the option to wikify their own answers if they want, to indicate that those answers should reflect contributions from multiple people.

Also, the recommendation post notice is for questions, not answers. If you meant questions, a lot of old questions don't have it because we don't go back and systematically add the post notice to all of them. If you see one, flag it as needing the recommendation notice and we'll add it.

The amount of answers conforming to the policy set in the faq post is surely debatable, but I'd say at least some of these answers are too short to be considered descriptive evaluations of the books in question.

Yep, probably. I'd say flag them, probably as NAA but a custom flag works too.

  • $\begingroup$ To the descriptive part: 1. The "additional notice" I was taking about is the notice for "old recommendation posts" at the bottom on the faq answer, which should be tagged onto the non-policy-conforming answers after they've been made CW. This is a minor issue, though. 2. I don't think many of the answers have been made CW by their authors, they are mostly CW because the question was CW. $\endgroup$ – ACuriousMind Sep 9 '15 at 23:47

I think the main confusion is between "full communiti wiki", driving to only one collective answer, and partial community wiki: each resource goes to an answer, so people can still vote to order them, but all the resource descriptions can be edited. Here I would like to propose to use the second method:

  • question is community wiki
  • answers separate, one by resource
  • answers which are no community wiki should generically not be upvoted.

"Resource" can be actually a collective resource: website, all the books of an author, all the work in a research branch, etc... the glanularity should be specified in the question.


We close list-based question as too broad. We also close questions asking for opinion as primarily opinion-based. Resource recommendations are a hybrid of these two types of questions: they're posted to get one's opinion about which of many options to pick from.

I'd also argue that there isn't any physics content to a resource recommendation question. While the subject matter may be about physics, the question asked is about what someone should read and not about the physics itself.

As such, we should make resource recommendation questions completely off topic.

As an aside, while it might prove to be useful to include a "top few" key books for each subject within the relevant tag wikis, I think it to be impractical to define, develop & maintain that.


I don't see the 'disagreement' you talk about. The Good list, bad list is general and its conclusions are supported and made specific by the faq post Are resource recommendations allowed?. On the other hand, Do we need/want an overarching books question? asks about one specific question and has no bearing on general posts.

We currently have 762 questions tagged with [...] but only 174 of them have been made community wiki.

This is because re-processing old threads is a labour-intensive manual process that cannot be done in intense bursts, and we simply haven't done all of them. As you correctly point out, the policy does not indicate whether old questions should be made community wiki. In general, they should, to conform with current practice, and because of the reasons you outlined. (Some cases, though, are not so clear. Should this answer be forbidden from earning rep?)

For old questions, the procedure is:

  • Check that the existing answers conform to the policy.

  • Consolidate the ones that don't into a single CW answer with the banner from the policy.

  • Overall spelling / formatting checks.

  • Flag for the banner on the question.

Ideally, the moderator will also turn the question into CW unless there's some specific reason not to. If it's already been processed but it's not CW, flag it for a moderator to CWify it.

This bumps the old post to the front page, so it should be done sparingly (i.e. no more than five at a time or so) to avoid flooding the front page. There's currently some three hundred open non-CW resource recommendations questions, so there's still plenty of room for keen-eyed eager volunteers to reprocess old posts.

  • $\begingroup$ The disagreement is: Good list, bad list says that all book questions should be made community wiki in its top voted answer. The faq post, however, makes no mention of it, and new res. rec. posts are not all processed the same: Some of them just get the res. rec. notice but remain non-CW, while others get turned into CW when the res. rec. notice is added. This mainly depends on the moderator processing the question (DavidZ does not apply CW, while Qmechanic does). $\endgroup$ – ACuriousMind Sep 10 '15 at 12:34

You must log in to answer this question.

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