Is the Critique rule the right rule?

Question

Stack exchange has a strong history of answering questions like "Why is this code broken? How do I fix this? What is the best way to write this algorithm?"

It is strange to me that this is not allowed in the writing community on this site; which is not to say I don't see why there is such a rule. Anyone can submit bad writing, at higher flow rates and in ways that are harder to categorize, than perhaps code questions are. But, if it is the case questions on specific works are going to always be off limits, it does seem largely at odds with the purpose of the rest of the exchanges.

I thought I'd re-challenge this restriction on this site as in my mind it is the main thing that would prevent me from voting for this stack to leave beta status.

In short, I think the the main reason this stack doesn't "fit" SE and the largest problem to solve for the long term health of the site.

---

If the rule were going to change, what would need to be in place to support it? Does SE have the tools to divert critiques into a classification that doesn't drown out the professional development/skill level questions we do a relatively decent job supporting today?

Answer 1

Not strictly an answer, but some thoughts.

"Why is this code broken? How do I fix this? What is the best way to write this algorithm?"

"Why is this code broken?" and "How do I fix this?" are (usually) two ways of saying the same thing, so I'll lump them together. Assuming an otherwise clearly written question, it'll be pretty clear what the person asking the question is after. However, it's worth noting that e.g. Stack Overflow has a specific off-topic close reason for "debugging help" questions "without a clear problem statement", as well as one for "problem[s] that can no longer be reproduced or [] simple typographical error[s]".

Many critique questions are largely similar, except with natural language instead of code. "How can I improve this text?" is not really an answerable question; there are far too many ways that any non-trivial chunk of text can, in someone's eyes, be "improved". On the flip side, "here is an example of my writing that I feel is lacking in X; what techniques can I use to emphasize the X?" can be answered with experience, and to my reading is already on topic under the "own writing as examples" exception to the critique questions ban. Even something like "why doesn't this text evoke any emotions?" can in principle be rephrased into "what techniques can I use on a text like this to evoke emotions in the reader?", at which point it becomes a question about technique rather than a question about the specific block of text while still giving the OP what they are likely to be after.

An unqualified "what is the best way to write this algorithm?" would almost certainly get closed over on Software Engineering (which would likely be the most appropriate site for that type of question) as either unclear or too broad. However, if one qualifies it with what the most important factor is (run time? memory usage? how either grows in the size of the input? etc.), and asks for algorithms to solve a given problem, then it'd probably be acceptable, but at that point, we're back to the distinction between solving a generally applicable problem and critiquing something specific already written.

The closest thing to critique requests that we can find on Stack Exchange is probably Code Review, and as you'll notice if you browse their site, they often get multiple very different answers to a single question, but each answer tends to point out one or more issues with the code in question and propose fixes. If we want to discuss allowing critique questions, I suspect that we'll need to take a few pages out of their book. Notice that they pretty much spend their entire "What topics can I ask about here?" page just to outline what is required for a question to be on-topic with them. Code critiques also have one major benefit over writing critiques, in that it's (a) immediately obvious if the answer works at all (does the code compile?), and (b) one can tell somewhat objectively whether the code still does what OP describe as its original purpose. An answer that meets both of those criteria is likely providing worthwhile feedback on the code, even if it is feedback that OP might perhaps not enjoy, while an answer that fails to meet one or both of those criteria is clearly not a useful answer. I can't quite tell what similar criteria would be for natural-language text critique questions, given that the bar for "what works" and "does the same thing" are both higher and fuzzier.

in my mind it is the main thing that would prevent me from voting for this stack to leave beta status.

Unfortunately in a sense, that's not a vote the community can make. It's even out of the hands of the moderators. Even though all our other stats are "Excellent", it has been repeatedly mentioned that 10 questions per day is a strict limit to graduation under the current scheme. Although there has been some discussion about changing what graduation means, my understanding at this point is that the questions per day metric remains extremely important for The Powers That Be.

What I really think that we shouldn't do is to sacrifice quality for sheer number of questions. It would probably be easy to get to 10 q/d if we considerably lowered our standards for what constitutes a good question, but I suspect that doing so would also drive away a lot of users. In my opinion, that's not the proper trade-off to do.

Answer 2

I think that the main barrier is this: Generic critique questions are, inherently, poll questions.

Leading to all the usual problem with polls and bikeshedding:

• They aren't about solving a problem -- "Wow, that's a really fantastic piece!" is a perfectly valid answer.
• They don't necessarily agree on what the problem is -- "This opening is boring, here's how to fix that" and "This opening doesn't give me enough detail, here's how to fix that" are both valid answers, even if the answers are 100% in contradiction.
  • This means answers are impossible to compare. Each is a solution to a different perceived problem.
• A single poster may have multiple suggestions -- does that mean multiple answers? A single answer with five different points? How do I vote if I love two of the suggestions and hate three? If two answers offer the same suggestion, is one of them better than the other if it has another couple of suggestions as well?
• Which answer does OP accept? If each answer is one poster's genuine opinion, how is any one of them "more" correct than any other?

---

Not only do all these issues undermine Stack Exchanges entire system of voting and visibility -- they also don't provide the (mixed) benefit that poll/bikeshedding questions do on Stack Overflow. On SO, a popular bikeshedding question can attract engagement and activity, from the site and from search traffic. But critique questions are, by nature, extremely localized. They matter to the person writing the piece, but it's hard to find any circumstances under which anybody else will find those responses helpful to something they're dealing with.

• There's almost no reasonable search likely to lead writers to these questions.
• There's almost nothing to differentiate one critique question from another.
• There's no way for a writer to look for subsets of critique questions that are helpful to them right now -- they can go reading everything, but there's no better resolution than that.

So another hypothetical way to help, would be to come up with at least some taxonomy for critique questions. Critique questions asking "is this a good opening" are somewhat more focused than "is this a good snippet"; you could read a bunch to learn about how to open strongly. You could read critiques of romance scenes, SF-nal infodumping; establishing character in family drama; blank verse poetry.

This is hypothetically possible, but I'm very skeptical of it.

• I don't think we've got anywhere near the volume of users and writing we'd need to create a varied body of critique questions.
• Picking a single snippet to be critiqued with little context is incredibly difficult. Many users (and most newbies) will not do it well.

---

In conclusion, you write:

But, if it is the case questions on specific works are going to always be off limits, it does seem largely at odds with the purpose of the rest of the exchanges.

I disagree. SE is all about Q&A -- that's what it's built around, and that's what it excels at. You can't post code to Stack Overflow asking "Here's some code, is it good? -- you have to narrow yourself to a clear, answerable question.

A Writing.SE for critiques is a very understandable goal, but it's also a very difficult one. Critiques to Writing is like Code Review to Stack Overflow -- asking for general feedback, rather than Q&A. And Code Review is a much more obscure site, more difficult to understand and make use of.

The problem is that writing doesn't lend itself to concrete problems, the way that code does. (Ah, if only Word would tell us, "This story does not compile; excessive POV shifts in Chapter 3"!)

We basically need to make the choice: focus on what concrete problems writing does yield up, or abandon the focus on Q&A.

Neither choice is wrong, but I think there's no lack of writing spaces online that don't have the Q&A focus to begin with. If we're going to be good at anything, it'll be questions that can be framed as Q&A, not the critiques.

Answer 3

To solve some of the problems noted by Michael and Standback (that posted before me); I'd suggest we allow critiques, but limit answers to ONE topic in response. Answer the question,

what is ONE thing wrong with this piece?

In comments one might say "nothing is wrong, I love it", but that is not a helpful answer to the question. Limit answers to ONE topic of critique. If I see more than one major error, then sure, I will post multiple answers, but they get separate votes depending on how many people agree with that ONE topic. And multiple answers are already allowed under SE rules.

I would also limit the amount of text to be critiqued to 250 words; or two screenfuls of the question writing text box (I think that is the same on every system, I don't know). 250 words is about a page of a book; that is enough to critique.

As far as other sites, World Building has TONS of questions that are effectively "what to write" (about science, biology, evolution, sociology, psychology, etc), and "critique this idea" and "I need help with my plot", and other questions with no "right" answer; like how would technology develop if there were no iron in the Earth's crust; or for a what to write Question; something like "My human-like aliens are an all female race, how do they reproduce?"

If World Building manages to field these broad, no-right-answer questions, so can we. And WB is nothing like Code Review. Many of these questions attract several disparate answers, but the OP picks whichever one they like best. We don't require a "right" answer to a critique; or alternatively, presuming the writer wants their writing to be popular, the highest voted answer is the right answer to a critique.

In my opinion, if we are here to help writers become better at this craft, prohibiting critique in any form cripples that mission. And it isn't solved by making the writer pick what THEY think is wrong and ask how to make that aspect better: Beginning writers don't necessarily know what is wrong with their writing, they want to know if anything is wrong with it.

---

Reasons why this approach works where others do not.

1. Length is limited. The truth is, many writers would like to cut and paste their whole book in for a "review". By limiting them to a page, they present a passage, some prose, a conversation, etc, something interested people can read and respond to in a reasonable length.
2. If necessary, we can establish a policy of one review per day; posts could be flagged/closed for violating it.
3. By limiting the answers to one topic of fault, votes indicate whether that singular topic is something others agree upon. That can be useful information to the OP. For example, if many people agree the characters are too cliché, then they should take it to heart. If many agree that the descriptive language is too convoluted to follow, they should take that to heart. But if one answer has 8 votes and three others get zero, those three others can likely be dismissed as "personal preference" answers that most do not support.
4. Many amateurs don't know even the basic terms of writing. We shouldn't hold that against them! This lets us review amateur work and suggest a course of action (or study), perhaps with links supporting that to other questions or outside sources that can provide the author instruction. They can learn the lingo as they need it.

Answer 4

Why we must do this here, or not at all.

I think it's worth posting The StackExchange position on "Review Stacks" which has been used to close Adi's request; as it is an answer of sorts to this question.

There have been a few attempts in the last week or two to suggest a new form of stack-exchange specifically for writing reviews. All of them have been deleted/closed.

There's a standard answer posted whenever suggests such a thing on area 51.

On Proposals Soliciting reviews, recommendations, comparisions, etc

Software Recommendations" was created in a time when the sheer size of our software developer audience gave them the collateral to give this a shot. "Hardware Recommendations" later expanded on that idea and it's struggling critically with nearly half the questions either deleted or closed. Since then, proposals have pushed out even further into website recommendations, book recommendations, travel recommendations, idea exchanges, essay reviews, textbook errata, job scams registries, software comparisons, suggest-a-song —

I don't have any regrets with the sites we tried, but hardware and software search is about as far as we can take this format; we don't wish to keep expanding this model any further.

I can appreciate that you have [other] questions you would love to ask on a Stack Exchange site, but gathering up questions from a community where they're not wanted doesn't seem to make them work any better here. Honestly, where we should be spending our energy is in figuring out how to make the best of these questions work on our existing sites, but that's a big boat to steer and we need a lot of software support and a buy-in from the communities that host these subjects.

This does not provide a positive outlook for the suggestion that we allow critiques on this site as apparently they are a source of struggle and lead to many, many unanswered, deleted & locked questions. It also does not preclude us from having some form of review on this site. What it does ultimately say is that we either find a way to have constructive critiques live here, as a part of this site; or abandon the idea of it living in the StackExchange environment.

I don't want to extrapolate too far from the content I'm linking/quoting, but I think we should acknowledge the following:

• The need for these types of things, does not exist here solely. There is a site-wide desire for these types of things. Point: Common Problem.
• Most communities have not figured out how to make them successful, excluding, perhaps "Software Recommendations". Point: Hard to Fix
• The owners of the web site have concerns about the software structures that are available to support this type of exchange of ideas. Point: System May Prevent Solution
• Communities that exist need to be the point of innovative beginnings if expansion were to ever occur. Point: Even though it's hard, maybe impossible given current state, it's still valuable for us to consider this.

Answer 5

Code fixes, identifications and answers dependent on physical laws are all pretty concrete and objective correctness can be judged. Writing is very subjective, genre writing is even more subjective; without completely objective standards to adhere to there is no "right" against which to measure answers given. This is a major problem on English Language and Usage when sentence interpretation/improvement questions get answered, it would be much worse when considering large parts of pieces of narrative fiction.

Also purely from a "good use of our members' time" standpoint the amount of writing that would be needed from OPs to give respondents enough context to adequately critique a piece is huge. Reading enough material to create a good answer will take almost as long as writing a reasonably detailed critique as well.