-16

This post is strongly inspired by a blog post by Jeff Atwood. In fact, most of the words are Jeff's. I've just changed the object a bit. Any eloquence is his. The arguments — well, some are his, too.


We've observed a particular pattern of questions emerging on several Stack Exchange sites.

How do you test if something is hidden with jQuery?
How to “add existing frameworks” in Xcode 4?
How do you search for backdoors from the previous IT person?
How do you find what process is holding a file open in Windows?
How do you know if you are watering a plant too much vs. too little?
How often do you need to turn the compost heap?

All these questions are effectively asking someone to do your job.

I love steaks so much, that I would like to get a green card and live in the States just to enjoy good steaks whenever I want.

When I try to cook a steak at home (I live in Italy), I almost ever get it burned outside and almost totally raw inside. Actually only a thin layer is cooked (or I should say carbonized).

If I try to lower the temperature of the grill pan, I get an extremely dried meat that resembles cork.

So, which are the basics for properly cooking a steak (say, to a medium-well grade)?

The question owner tries to describe something they can't quite do, in hopes that the greater community will “buzz in” to tell them how to do their job, as if they didn't have one of their own that pays. The best proposals gets upvote, and potentially an accepted answer checkmark. It's helpful, right?

Our engine is great at these kinds of questions, and they tend to do well:

  • “How” is the most popular first word in titles on Stack Overflow, by a 6:1 margin
  • “How” is the most popular first word in titles on Server Fault, by a 4:1 margin
  • “How” is the most popular first word in titles on Cooking, by a 1.5:1 margin
  • “How” is the 3rd most popular first word in titles on Theoretical Computer Science*

Of course, how-to-do-my-job questions aren't a new phenomenon. Jeff alluded to them in Should we require minimum reputation to continue asking questions? But after years of observing these how-to-do-my-job questions grow and spread to multiple sites with similar effects, I do not believe that the slight benefit of these questions outweighs the many negatives.

1. How-to-do-my-job questions aren't answerable

Consider Stack Exchange's first rule of questions not to ask:

You should only ask practical, answerable questions based on actual problems that you face. Chatty, open-ended questions diminish the usefulness of our site and push other questions off the front page.

Vague requests for a method to perform a badly-described task is not what I'd call a practical, answerable question.

Unless the asker has demonstrated a genuine reason they need to perform this task, documented that they’ve invested substantial effort in finding it, and given us something concrete that provides us with a reasonable chance of actually guessing the answer — it’s simply Not a Real Question. At best it is a work order.

2. How-to-do-my-job questions don't help others

Because these questions are based on specific, situation-dependent, half-understood descriptions, it is unlikely anyone else will be able to find them through a web search. I have a difficult time imagining how you’d construct a web search, either on Google or via Stack Exchange’s built-in search, to find something that you can’t fully understand. What’s even worse is that these questions, by their very nature, will contain a bunch of broad, speculative “maybe you can try …” catch-all terms that are likely to trip up future visitors who end up there by accident.

Consider the example of parsing HTML with regular expressions, a task apparently so difficult to perform that our programming site contains 171 duplicate how-to questions about it:

The goal of Stack Exchange is not to construct un-findable single-serving questions that only help one person, but that’s exactly what how-to-do-my-job questions tend to do.

3. How-to-do-my-job questions are unfair

If we allow thinly wrapped “do my job for me” questions, we are effectively turning into Rent-a-coder, only without the pay. No need to expend hard-earned dollars, or even appeal to charity: just post a work order, slap “How do I” at the beginning and a question mark at the end, and we’ll do the rest of the hard work necessary to figure it out for you? That’s a dangerous precedent to set. It is disproportionate and unfair to the experts on the site.

Also, an expert in the topic should be able to have at least some confidence that the answer he’s writing answers the question. Take that away, and you’re left with questions that don’t know what they want, and answerers throwing guesses at it hoping one will stick. “Does it work if you add a comma?” Nope, try again! “Surely #>+++++++4+[>++++++<-]>++++.-----.+++++.*\ must work!” Sorry, go fish!

4. How-to-do-my-job questions aren't educational

I understand that it's sometimes fun to puzzle out how someone may do their job. I also appreciate that it takes a lot of expertise and deep domain knowledge to take a vague, badly-described problem description and nail the exact solution. But I would also argue that these questions aren’t educational in any way, because there’s no way to learn by doing. A particular community member, by virtue of their experience in the field, just happens to be able to take the limited information you remembered and fill in enough of the blanks to guess a correct solution.

I urge you to click on the questions tab yourself and take a long, hard look at the artifacts these how-to-do-my-job questions are producing. After a couple of years I am convinced that how-to-do-my-job quetions do not meet our goal of making the Internet better.

Quite the opposite, in fact.

* And when the next in line is “complexity”, there are perhaps deeper problems to contend with.


It's amazing how few words I had to change from Let's Play The Guessing Game.

28
  • 6
    How do you propose to solve this problem? Banning the word "how" won't stop them being asked. Apr 1, 2012 at 1:42
  • 6
    No no no, we need to ban the horrific grammar miscarriage that is the "How to do it" question title.
    – Ben Brocka
    Apr 1, 2012 at 2:24
  • 5
    The 'How to “add existing frameworks” in Xcode 4?' example you list would seem to work against your arguments, because it is a specific, answerable question and the directions provided there could be of interest to many others searching for how to do this. If we're just being asked to do someone's job and the answer wouldn't help others, the "too localized" close reason seems to work just fine. Banning a common word would seem to be overkill. Apr 1, 2012 at 2:48
  • 1
    +1 - I actually agree with this suggestion. My only problem is that it isn't inclusive enough. I would probably include "Is there a better", "What is the best", and "Is this the right way" (on SO) in the mix somehow. Though if I had it my way I'd disallow all of the w's. Luckily though... I'm not in charge.
    – M.Babcock
    Apr 1, 2012 at 3:09
  • 9
    Wait a minute, is this an April Fools' Day prank? If so, well played. ಠ_ಠ
    – user149432
    Apr 1, 2012 at 3:14
  • 1
    How do I create a numbered list in my questions? Apr 1, 2012 at 8:18
  • 1
    @MarkTrapp Gee, ya think? I'm surprised how this was taken as a troll, and not a parody. I do have a serious point, however, but I realize after the fact that it's incomprehensible if you haven't read the blog post by Jeff that I'm parodying. Apr 2, 2012 at 22:07
  • 2
    @Gilles The funny thing is, what people strongly object to (and why people think you're trolling) isn't your argument, it's that you attempted to make the case to ban the word "how". That's not at all similar to banning "how to do my job" or "story identification" questions. The parody would be apt if, for example, Jeff made the case to ban the word "story" in titles, or there was a how tag.
    – user149432
    Apr 2, 2012 at 22:42
  • 3
    It's not a very good parody. Sorry. You can go re-read the data I cited on the blog post if you like. May 13, 2012 at 4:35
  • 2
    @JeffAtwood Data? The only data in your blog post is the number of questions and number of views in the tag, which is irrelevant to your arguments. None of your arguments is substantiated with data. Of your four arguments, only #2 (“don't help others”) is a matter of fact rather than opinion, and the data (number of deleted “me too” answers) do not justify your claim. May 13, 2012 at 14:53
  • 1
    @JeffAtwood You are very much Pee-Wee Herman here. It all boils down to your not liking a type of question, and you don't stop to think what other communities might think of these questions. The fact is, SF communities (and more generally, reader communities, and I think moviegoer communities) do not think that story identification questions are impractical, unhelpful, unfair or uneducational. We consider this kind of question to be a normal, expected part of a place where questions about books/movies are asked. May 13, 2012 at 14:57
  • 3
    @gilles whether the communities like them or not is irrelevant; if they want to do it on our engine, they need to consider needs other than their own selfish ones. There are a zillion discussion forums where such "questions" can be asked. Take it somewhere else, or shut the sites down entirely; that's fine with me too. Our engine does not work for all topics, the purpose here is to generate artifacts that are useful to the outside world. May 13, 2012 at 16:13
  • 2
    @gilles these are the most popular [or 2nd, 3rd most popular] tags on the site. You don't see any anomaly there with them being nearly dead last in overall views, yet hugely "popular" on the site as measured by # of questions? I do. It means the community is optimizing for the wrong thing. It's fine for such "recommend me an X" and "identify me a Y" topics to exist in the world, just not on the SE engine. May 13, 2012 at 18:01
  • 3
    and for the record I think virtually none of current questions on scifi.stackexchange.com make the Internet better by the definitions we use at Stack Exchange. To me browsing that page is like watching a car wreck in slow motion. May 13, 2012 at 18:52
  • 3
    @JeffAtwood If you want to look at it that way, there's way more on-topic, open crap on SO (and clearly trumps scifi in numbers due to its volume), and that's not making the internet any better either... May 13, 2012 at 21:24

4 Answers 4

11

TL;DR:

  1. Word bans don't work. Banning "How" won't make these questions go away.
  2. Half of the examples you provide are not indicative of the problem you claim exists. Indeed, the last one about compost turning isn't even phrased as "how do I"; it's phrased as "how often". It asks for a time interval, not a process. Thus showing that banning the word "how" would make formulating such a question more difficult.
  3. The concern with helping others is focused in the wrong direction. You should focus on having titles be more descriptive of the guts of problem (which makes it easier to find when someone else comes along), not the task. Ask about regex parsing of HTML, not your specific task that you plan to do with regex HTML parsing. Banning the word "How" will not do this, because people will still just focus the title on the task they're trying to do. They simply won't use the word "How" to do it.

Personally, I'm of the opinion that word-bans don't work. Your reasoning is basically, "a lot of questions with word X in them are bad. Let's stop people from using word X."

This would only be reasonable if the word itself were the cause. But it isn't; at best, it's an indicator of a problem. And it's not even a reliable indicator at that.

It's like being burned and then thinking that the best way to avoid fires would be to avoid anything orange. Not only will that require avoiding a lot of non-fire things, but there are non-orange fires that you can still encounter and be burned by. So not only does it elicit a lot of false-positives, it also doesn't actually solve the problem.

Remember: banning "Problem" from question titles didn't make bad questions stop. As far as we can tell, it didn't decrease the number of bad questions. At best, it removed a word that was superfluous, making titles shorter. In some cases, it caused people to mangle the word like "pro-blem" or such nonsense in titles.

And in the worst case, it stops people from legitimate uses of the word: you cannot post a question on SO where you mention the "Halting Problem", "Traveling Salesman Problem", or many other computer science "Problems" in the title.

So on the one hand, people who ask bad questions will... continue ask bad questions. They'll just use different words or, worst-case, just employ name-mangling. And on the other hand, people with legitimate uses of the word are prevented from using it. Again: not only is the problem itself not addressed, it creates a new problem for legitimate users.

I understand the impetus to use automatic filters and such in an attempt to prevent low-quality questions from being asked. But outside of the worst-of-the-worst, that's generally not going to be possible. And even that requires fairly complicated heuristics to accurately detect; a word ban is far too coarse-grained of a solution.

It's using a shotgun to do surgery: you'll likely still miss the target and the patient will probably die.


Having looked back at the argument in greater detail, I have to say, it's not very well reasoned. Your examples don't seem to fit the argument. Consider:

1. How-to-do-my-job questions aren't answerable

Well, let's use the examples you give:

  • How do you test if something is hidden with jQuery?: This is a practical question that has an answer. It's based on problems the user faces (having DOM elements that he wants to hide). It isn't chatty or open-ended; there's not much wiggle room for answers. They must use JQuery and the JQuery they use must show whether a DOM element is hidden or not. What's there to chat about?

    As for demonstrating a reason for it, it's obvious. You're writing JavaScript using JQuery. You have a DOM element. You want to know if it's visible. It may not be rocket science, but it is a very clear and obvious reason why someone would want to. Anyone who wants to learn how to use JQuery to detect element visibility will benefit from the question. The problem you seem to have is that it's minor, something that anyone could look up in documentation or something.

    But that has nothing to do with the word "how" or its use in this question.

  • How to “add existing frameworks” in Xcode 4?: Again, it's a practical question for any user of Xcode who doesn't already know how to do it. It has an answer. It's a problem a (novice) user faces. There's no wiggle room for chat or any open-endedness.

    The reason is again obvious: you're using Xcode and you want to add a framework to your project. Again, not rocket science, and something that is readily available in the documentation.

    But again, that has nothing to do with the word "how" or its use in this question.

  • How do you search for backdoors from the previous IT person?: This is a very open ended question, as there are many avenues for slipping backdoors into systems. As such, you are correct: it's not a good question.

  • How do you find what process is holding a file open in Windows?: This is a practical question for a problem many users face. I don't really know why it's on Server Fault and not Super User, but there it is. Where this question becomes borderline is in the specificity. There are many tools that could be used, and all of them work. The question doesn't outline additional criteria to narrow these down to exactly one solution.

    That being said, not all questions have exactly one answer, and we allow plenty of those to exist. There will always be different ways of doing some things. So while this isn't as closed as some, it's a lot less open-ended than most of the material we close.

  • How do you know if you are watering a plant too much vs. too little?: Way, way open-ended. Not all plants are the same, not all plants react the same to over-watering, etc. Not a good question.

  • How often do you need to turn the compost heap?: I am not in any position to categorize this question. I don't know enough about composting and heaps to know if this question is narrowly or broadly scoped.

    However, I do know English, and there is a big difference in sentence structure between "How do you do X" and "How often do you...". Indeed, you have given a prime example of why simply title-banning "How" is a bad idea.

    This question is not a "how-to-do-my-job question". It's a question that simply uses the word "How" as the first word. It's a question that asks for the frequency of performing a task to achieve optimal results.

So, out of the 6 main example questions that you led off with:

  • 2 were clearly answerable.
  • 1 was borderline answerable.
  • 2 were clearly not answerable.
  • 1 was not even an example of the problem.

So at best, your title-ban of the word "How" would have gotten 2 false-positives and one case completely outside of the problem you state. That's a 50% failure ratio, and that's on your hand-picked set of questions. And it could be as high as 66% failure if you take the borderline one.

Note: The above discussion did note a fundamental difference between "a question that is in line with the rules" and "a good question," the later being a purely subjective measurement. Questions can be within the rules and still be considered not very good.

2. How-to-do-my-job questions don't help others

Your big example here is this:

Consider the example of parsing HTML with regular expressions, a task apparently so difficult to perform that our programming site contains 171 duplicate how-to questions about it:

This is a fair point. Those questions really all are talking about the same thing with different uses, but they don't know it. However, consider this.

Let's say I'm the person who asks, "How do I filter all HTML tags except a certain whitelist?" I look at SO, and I search for HTML tag filtering. Maybe I'm clever and put [C#] in there. What do I see? Well, the first questions are either newer than the one we're talking about (and thus haven't been asked yet. You can't blame me for duplicates created by others), and the vast majority are useless to me, even if you ignore the timing of when this question was asked.

In short, I won't find anyone else's answer that can help me. Why? Because both I and they did not provide a proper title for their question.

There are many reasons why someone would want to try to parse HTML with regex. The problem this creates is that everyone puts in their title the task that they're trying to solve. I'm doing tag filtering, he's extracting tags, that guy's matching string values, etc. The task we're trying to perform is what people put in their titles.

This problem has nothing to do with "How" questions. This problem is bigger than "How" questions (even if we assumed that every question with "How" in the title were one of these). The problem is people making bad titles. "Filtering all HTML tags except a certain whitelist?" is no better than "How do I filter all HTML tags except a certain whitelist?"

Restating a bad title doesn't make the title better. That's why title-banning "problem" didn't make titles and questions better.

In short: you're focusing your effort on the wrong part of the problem. You're looking at an (unreliable) indicator of a bigger, more significant problem.

3. How-to-do-my-job questions are unfair

There doesn't seem to be much evidence for any of this. As previously pointed out, at least 50% of the questions that your proposal would have been "stopped" were just fine. They weren't overly broad and they would help other people.

Even better, none of the programming/computer-related questions you cite are from after 2010. So if this truly were a problem, where people would just fill out post a work order, slap “How do I” at the beginning and a question mark at the end, we'd already be swimming in it. But we aren't.

Also, such a statement basically suggests that SO patrons are idiots. That we'll answer anything that gets put before us. But SO users have not exactly been shown to be idle when it comes to shutting down bad, broad questions. We're not exactly a conservative lot when it comes to downvotes and closures.

As to this:

Also, an expert in the topic should be able to have at least some confidence that the answer he’s writing answers the question.

Again, 50+% of the examples you provide have this quality. Indeed, for this question, I'd say 66% do.

4. How-to-do-my-job questions aren't educational

This one falls prey to the 50+% of provided examples problem. Again, when half of what you provide isn't indicative of the problem, then it probably isn't a problem.

22

The reason "how" is so popular is because, in the English language, that's the standard way to begin a question asking for the manner by which to solve a problem. You're effectively asking people to ignore the rules of the language we use to communicate and avoid asking well-formed questions because some of them happen to be bad.

What's next, banning "the" because most closed questions have it? Capital letters? You're treating an unrelated symptom of the problem.

To deal with the actual problem and not create a new Stack Exchanglish language, you've already highlighted the solution: close the questions as Not a Real Question. That's the purpose of this line in the FAQ:

Your questions should be reasonably scoped. If you can imagine an entire book that answers your question, you’re asking too much.

A person who hasn't narrowed down the scope to a fun size unit of work for the answerers—i.e., someone asking for strangers to do their job instead of for help solving a specific problem—isn't asking something that's reasonably scoped.

11

While I can certainly agree that some of those questions are bad, I don't think banning the word "how" is a good idea. In fact, I didn't have to look too far to find an counter-example, because I have one such counter-example in my own profile. Its title is "How do I remove a git submodule?". The question has more than three hundred votes, the accepted answer has more than four hundred, and the page was viewed more than thirty thousand times. It is currently the first result for the Google query "remove git submodule". It was answerable, it helped others (judging from the 300+ upvotes), and people having the same problem can find it easily.

Even if it can be argued that such good questions can be rephrased so that the word "how" is not present, it can also be argued that the bad questions can simply be asked without the word "how". Bad questions tend to be poorly formulated anyway.

test if something is hidden jQuery
Xcode 4: “add exsiting frameworks”
search backdoors form the previous IT dude?
Windows: find proc holding open file
water a plant too much vs. too little?
frequency of turning the compost heep

I don't want to have to rephrase my posts over and over again until I hit the right combination of words. That's not what I want to spend my time doing. If I have a "how" question to ask, I want to ask it without having to workaround word filters. I have better things to do than battle machines for the right words.

If you want to get the community to lay a heavier hand on "how-to-do-my-job" questions based on the actual question and not on the words used, I'm all for it. I'm not for putting word filters into place especially if those filters filter interrogative words.

3
  • These questions are not popular for the right reasons, though, see: blog.stackoverflow.com/2012/01/the-trouble-with-popularity Apr 1, 2012 at 2:07
  • 6
    @Gilles I'm confused. Are you really arguing that questions with the word "how" on them are automatically popular for the wrong reasons? Apr 1, 2012 at 2:11
  • 1
    @Gilles: Well, the question is short, succinct, to the point, clearly specified, properly-scoped, answerable, and provides useful information. You can argue that popularity in general is not indicative to the quality of the question. But the objective merits of this particular question are a lot harder to argue against. Apr 1, 2012 at 18:41
6

Is this also not what the up/down votes on questions are for? As the tool tips indicate:

  • Upvote: "This question shows research effort; it is useful and clear."
  • Downvote: "This question does not show any research effort; it is unclear or not useful."

I believe the types of "how" questions that you're apparently frustrated over should all fall under the category of not having any research effort, and should simply be down-voted - and/or given the close votes. And to the voters: Please follow the site's advice and leave feedback as to why / how the question could have been improved. This will help educate the user base - though the level of potential success here is debatable.

You must log in to answer this question.

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