Are books special?

When I see a question like this, I naturally transform it into a more general question about resources. Printed books aren't necessarily the best way to learn about programming concepts; web pages may work much better. Some people like video tutorials; I personally have found they're usually not very good and that letting someone else set the pace doesn't work well. However, fundamentally I don't see a big enough difference between "recommending a book" and "recommending a Youtube video" to give them different policy treatment.

Beyond even that, I would classify resource requests the same way even if they aren't educational. Looking for a library to handle certain types of programming task, is currently deemed off topic on Stack Overflow - with all the same rationale as the questions about books etc. But in my view, a place that sees value in recommending one should see value in recommending the other.

With appropriate focus, sure, why not

The scope of Software.CD has already been established to be broader in some regards than Stack Overflow; in particular, we accept design and architecture questions that the Stack Exchange network would definitely see as a better fit for softwareengineering.SE, cs.SE, langdev.SE etc. I feel like if we're including those sorts of things, then "softwarerecs.SE for programmers" (i.e., recommendations for libraries, linters, IDEs etc.) fits naturally, and therefore so do the book requests.

posted about 1 year ago

CC BY-SA 4.0

1y ago

Karl Knechtel‭

2175 reputation 39 54 280 85

I respect Alexei for opening a discussion on this, and Lundin for providing hard-won real world experience. I would like to put in a "pro", on the grounds that "I read this book and I'm a better programmer for it" is information that directly supports the codidact mission of helping each other learn. It's related to the old observation that bad IT people have walls full of certifications and good ones have shelves full of books. A fair question is how to maintain adequate quality and I'm short of concrete suggestions at the moment. Lundin's attempt to limit recommendations to things the person posting has actually read seems like a reasonable minimum.

posted about 2 years ago

CC BY-SA 4.0

Fred Wamsley‭

290 reputation 3 4 30 0

Are we talking about questions asking for book recommendations, or for a way to share/compile book recommendations? A question doesn't have to be the vehicle for the latter.

A question like "what are the best C++ books?" is likely to yield results that are haphazard and hard to manage. It would probably be more productive to focus on questions of the form "I want to learn more about X and have the following background and constraints; what is a good resource for self-study?". Notice what I did there: it's not a general-recommendation question, it's not meant to be the place for a whole topic, and it calls for context. (This is the approach that the "(something) Recommendations" sites on SE use.) This means we might end up with a bunch of questions about the same general topic, but approached from different angles. It seems like, with proper tagging, that could work, especially if these questions are in their own category.

But if the goal is to share knowledge, not ask a question (what I gather happened on SO, though I wasn't there), the community could instead use articles -- either general resources like Languages & Linguistics is doing, or a blog where a single author writes an article and there can be multiple articles on the same general topic. It sounds like you're not looking for the general-resources approach of L&L, but a blog approach could still work. You can set rules for blog posts, and community voting would provide signal.

posted about 2 years ago

CC BY-SA 4.0

Monica Cellio‭ staff

2510 reputation 14 35 345 57

As someone who spent a lot of time trying to get this to work on Stack Overflow, I would advise against it.

Some background story of my merry adventures with book lists:

The story starts around here: Delete the list of random books? Back in 2015 I stumbled upon a list of random books not necessarily even about programming. It was such a clearly bad and off-topic post that I thought it would be no big deal to delete it. I quickly got quite significant community consensus of around +75 score to back me up. Several moderators had correctly tried to get rid of it over the years too, only to get some deletion rollback. SO staff overruled moderators at several occasions. In the end, I could only conclude that the list of random books was important because it had lots of traffic, even though it was blatantly off-topic.

A couple of years later, a moderator decided to merge all post in a similar book list regarding C programming specifically. This list was far more on-topic and I had initially supported it and contributed to it. But after the list was merged, it was no longer clear who was recommending what, which recommendations that had been up-voted and so forth. It was no longer a list of recommended books, but just a list of books. It was also highly subjective and poorly maintained.

Taking it to meta again, I posted The C book list has gone haywire. What to do with it? Again community consensus was reached, the post was deleted and that ought to have been the end of it.

Then some year later, staff (again) does a rollback undeletion, unaware of the previous consensus to delete it. Just because a high rep user has urged them to undelete it. Again, I had to explain that it was still a mess, a list of random books neither sensible nor actively maintained by the community. The user who had insisted to undelete it said that they would organize active maintenance of the post and they have since then done a little bit of that, although not nearly enough.

Another year later, not much had happened. I posted this, arguing that if we were to keep the book list, we had to severely step up the quality concerns. I made some suggestions for a quality criteria, including a proposal to add shockingly pedantic requirements such as "you shouldn't recommend a book unless you have actually read it". This post didn't get much attention and so the whole thing remained in "status quo", where it remains to this day.

Many years later and these two horrible book lists are still there. Even though they are explicitly off-topic for SO and of very poor quality, recommending things at a whim, including bad books. Making the Internet a worse place, one recommendation at a time.

Lessons learnt

Since book recommendations are subjective by their nature, people's opinions about how to maintain the book list are also similarly subjective.

Including:

• What scope should the book list have?
• How should the book list be maintained and by whom?
• What criteria should there be for adding a book?
• What criteria should there be for removing a book, if any at all?
• Should anti-recommendations be allowed or not?

Codidact perspective

Ok so Codidact isn't SO and at least we don't have to worry about some company goons undeleting things that were bad but rendered precious site traffic.

We do also have a little bit experience from other Codidact sites allowing recommendations: Outdoors - Gear Recommendations and Photography & Video - Gear Recommendations. Ignoring imported SE posts, these have mostly been well-received, or at least not heavily down-voted.

Overall, I think the spirit of Codidact is to have a higher tolerance about posts that address important topics even though those topics may be on the opinion-based side of things. So maybe recommendation-like questions work better here than on the big SE sites. There do exist smaller SE sites dealing exclusively with recommendations and those seem to be doing ok too.

A need for trusted recommendations still exists

With crappy Internet/Youtube tutorials becoming increasingly common, it can't be denied that a canon of actually good learning material is more important than ever. And as shown by the previously mentioned book recommendation lists, there's no shortage of really bad books either.

Just yesterday, I went up against some C programming book on SO, verifying that yes indeed, you are right to question this book, because it is wrong. It's valuable to mankind to have a place where confused students can verify that their books or teachers are not a source of learning, but a source of confusion.

However, and this is probably the biggest argument against it, I don't think Codidact has enough domain experts to maintain quality of any such book recommendation lists. On SO you might have some 10-20 leading experts in any given topic, proof-reading what you write. Meaning that the site can actually do things like confidently questioning the correctness of authoritative sources like programming books or compilers.

Although even with expertise in place, SO has repeatedly failed to provide useful book recommendation lists... There are some which supposedly are ok(?) like their C++ list, but I haven't read enough C++ books to tell.

My conclusion

There may be a place for programming book recommendations here, but unless we can definitely guarantee quality, we shouldn't attempt it.

posted about 2 years ago

CC BY-SA 4.0

1y ago

Lundin‭

8948 reputation 36 122 1027 166

Disclaimer: I am new here, I have not understood the overall policy of the site (yet), I can offer no moderator perspective, more a user and regular (unpaid) documentation contributor perspective.

The current answers mix a lot of different approaches together and the question was not specific in which approach should be pursued. Maybe instead of saying no, the goal should be to find a solution which is managable, not "authoritative" list with random books of questionable quality.

I would like to respond to the initial question how I understand it: Questions asking for book recommendations on a specific topic (e.g. getting started with python, clean code etc.) or even if a more complex topic can't really be answered or should be answered with book (or online resource) recommendations. It is not about "A question about a specific book" or comparing 2 books. Also, I am not sure about the collaborative community maintained authoritative answer. Why not use the ususal answer format and get the best answers upvoted?

I would be pro questions asking for recommendations and answers with (ideally) annotated book recommendations (where the recommendation is explained) because:

(1) A large flaw of Stack Overflow is IMHO offering too much fish and not enough teaching how to catch fish. If you ask for a book recommendation or how to get started as beginner or how to increase knowledge on specific topic or how to debug and troubleshoot, the question will (usually) be closed. But trivial RTFM or copy-paste beginner questions are heavily upvoted.

To be a "better" site with helping people learn and grow and figure things out for themselves (instead of only giving them the answers, ready to copy-paste), it is my wish the site offers more help for "how to learn something" and "figure things out themselves" and maybe not less but less prominent "how to do xyz".

(2) IMHO, some questions should not be asked, especially if they would be self-answered by taking a good book or online tutorial or online course and learning for an hour or even a week. But, as a beginner, it is very difficult to figure out what a good resource is. Help make it easier to rely on and strengthen good learning resources (be it in books, online, videos or whatever) instead of effectively replacing them. And offer something only where these fall short.

Example: Not programming related example for trivial beginner question and docs drowned out by forums in search:

• search for "exit vim". The documentation is drowned out by the forums in the search results. In the documentation, this question is answered. If the explanation in the docs is not sufficient, I would attempt to improve the documentation, not ask and heavily upvote a question about it on Stack Overflow. The same goes for many Git questions, JavaScript etc.
• Another example: The Apache web server documentation online is free and excellent. Yet many people insist on asking trivial questions where the answers are literally copy-pasted from the docs.

Book recommendations will not entirely solve this. But they are a start and a commitment to fostering learning.

posted about 2 years ago

CC BY-SA 4.0

sypets‭

29 reputation 0 1 2 0

Here is a slightly different suggestion, that may avoid the problems with open ended book recommendations.

Book lists? No

I'm not entirely against opinion based questions, but I am against opinions that are not attached to any meaning. It doesn't help me to know that one book has more votes than another. Some of the most misleading books in existence are very popular. Similarly, some excellent books are very popular but irrelevant to what I need to learn. I don't need to know how many people the book is useful to, I need to know what type of people the book is useful to: their prior experience, specific aims and requirements. Otherwise I have no idea if a good book is good for me personally.

In a book list question, even an answer that only contains one book is no use - the votes tell me that some people who know about software like that book. But people working in software may have very different ideas of what makes a book good. If they work in a field I know nothing about, the book may be too advanced for me. If they have fond memories of the book from when they first started out, the book may be irrelevant to my current understanding.

Specific book questions? Maybe

A question about a specific book can have answers that give reasons why that book is useful, and specific areas it is relevant to or not. Then votes won't be for the book, but for the answer that best sums up who this book is good for.

A book is not good or bad, it's fit for a specific purpose or not. The question should ask about a specific book, and not ask "is this book good?", but instead explain in detail why the question author is considering the book: what they have heard about it, what they are looking for, what is their background, what is their aim?

New post Category

Even having said this, I'd still question whether it is a good idea to allow book questions in amongst software questions. Is it worth considering a separate Category? If so, would that Category be for book questions, or for questions about general educational/reference materials?

I'd like to see a different term introduced, to replace "book recommendations". Although accurate, it also brings expectations of judgement of the book as good or bad, rather than as suitable for a specific set of needs. Maybe "book suitability question" would work - or perhaps someone can come up with a catchier Category name that gets the same point across - possibly not restricted to just books.

An advantage of a separate Category is that it can have a separate tag set, which might help people searching for specific fields, tasks, backgrounds or experience levels. I'd expect some of these tags to be problematic to introduce to the main Q&A tag set. Moving from language A, or moving to language B, might be useful tags in a book Category, but would probably just be noise if people start using them in the Q&A Category.

Strict focus

If this Category is introduced, I'd want questions to be closed unless sufficiently detailed. Not just mentioning a field and a few words describing experience level, but specific examples of what kinds of work has already been done and what needs to be learned.

This would not be a book review site. It would be for specific questions each from one person seeking a solution to match their individual circumstances - much like the Q&A Category.

One book per question

There will understandably be people who want to ask a question comparing 2 similar books, or asking for helping narrowing down from a shortlist of 5.

I would recommend putting in writing at the top of the Category page that it is strictly 1 book per question. This way the answers focus on the question author's requirements and current understanding, rather than tending towards spotting the differences between 2 books, where those differences may not be relevant. This also avoids the problem of voters rewarding the answer that favours their preferred book (voting for the book) rather than the answer that best addresses the requirements of the question author.

Potentially multiple questions per book

Personally, I would not necessarily consider a question about the same book to be a duplicate if it describes different requirements or a different current understanding/background.

posted about 2 years ago

CC BY-SA 4.0

trichoplax‭

551 reputation 0 15 55 20

To be a "better" site with helping people learn and grow and figure things out for themselves (instead of only giving them the answers, ready to copy-paste), it is my wish the site offers more help for "how to learn something" and "figure things out themselves" and maybe not less but less prominent "how to do xyz".

That would be a site I'd be inclined to visit often.

posted almost 2 years ago

CC BY-SA 4.0

Fred Wamsley‭

290 reputation 3 4 30 0