8

9.42% - That's the percentage of questions on EL&U.SE that are marked as duplicates. These are other SE sites' duplicate percentages:

  • Stack Overflow: 2.4%

  • Super User: 3.26%

  • Server Fault: 1.66%

A reason for EL&U.SE having so many duplicates is that most questions don't have searchable titles.

An example scenario of why non-search-friendly titles are bad:

I always try my best to find out if my question has already been asked before asking it. I usually can't find an existing one, so I ask my own question. Later on, a notice appears on my question:

Is or are - "The most preferable length of the internship is 3 months but 2 is also acceptable"

Closed as duplicate of

"If your car has 4 wheels, it's drivable. 3 wheels (is/are) a safety hazard." Which conjugation of "to be" is right?

How would I have possibly known about the question that mine was closed as a duplicate of?

Question titles on EL&U.SE either ask about the correct usage of word(s) in a specific context, like this:

  • "1 or 2 friends {was/were} eating {a sandwich/sandwiches} at the park. Which is correct?"

  • "The rock was uplifted vs. the rock had been uplifted?"

  • "What's the prepositional phrase in this sentence: The lights were bright in the room on Monday?"

Or their titles are stupidly vague, like this:

  • "What is this part of speech called?"

  • "Why is this word used here?"

  • "Is this sentence grammatically correct?

When someone does research prior to asking a question, it's highly unlikely that they'll find what they're looking for. And they'll probably write their own question's title with the same issues, later being closed as a duplicate of another question with some obscure title.

Idea:

Edit question titles from this...

Why is it "the tiger is endangered" and not "a tiger is endangered" when referring to all tigers?

..to this...

"When should I use a definite article with a generic noun?"

Should we title our questions to optimize for searchability?

  • 6
    "The reason for EL&U.SE having so many duplicates is that most questions don't have searchable titles." ← This is plausible as a reason for some of the dupes, but I'm not sure it's significant enough to count as the reason. I can imagine lots of other reasons people ask dupes, such as not searching in the first place. When I look at the questions on the front page of the site, certainly a large number of them appear to have no research effort whatsoever. I do think it's important to ask questions with specific, searchable titles, but I'm not sure that will solve the dupe problem on EL&U. – snailboat Jun 25 '18 at 0:01
  • 1
    In AskUbuntu.com I've changed a question title from "Why do I get this error?" to "Why does Xxxxx occur when I do Yyyy?" for greater search-ability. I would guess anyone on this site with sufficient reputation could do likewise. I would encourage you to do similar when you have sufficient reputation. – WinEunuuchs2Unix Jun 28 '18 at 2:17
4

Should we give our questions titles with greater searchability?

Yes.

There are at least three stages.

  • As the OP avoid obvious faults in the title:

    Never cause people to click to see what the question is, IE: no "What is this called?", at least provide a shape, approximate size or some other relevant information.

    Include relevant parts of the text in the title - this will vary greatly; from question to question and site to site - as the OP simply try to do your best.

    Be certain to include the subject in your title and any unique words from the text which differentiate your question from another whenever possible.

    A longish title is far better than one which is too short. If it takes two sentences, preferably just a comma, that's OK; a few sentences or paragraph are almost certainly not acceptable.

  • As a subsequent reader, please correct gross errors in the title or request changes using comments.

    Misspellings of keywords in the title should be quickly corrected. There's no purpose in indexing misspellings.

    Making a wordy title shorter is helpful, but don't go past point form and reduce it to a collection of keywords. Use appropriate style guides for the particular SE site and subject matter.

    If you don't know what is meant but clearly understand the subject a comment is appropriate. Deleting them afterwards is often appreciated too.

  • As an expert, or ones whom reads after the dust has settled, a constructive comment or edit should be welcomed.

    Be cautious about changing the meaning or offering no improvement - either is fair grounds for a rejection flag. Read twice if necessary.

  • As an archaeological reader if there's an improvement to be made your help is welcomed.

    Sometimes new technology and words or terminology are developed - adding these to the title is very helpful as it prevents new questions duplicating matters long settled.

    Sometimes lengthy reflection or a better understanding of things can lead to an enhanced ability to refine the title. If the OP hasn't revisited the site in years it's your job to correct it.

Use of popular keywords and ones which differentiate one question from another enhance both the human user and search engine, saving time and effort.

Bad titles (and poorly chosen first couple of sentences) make humans uncertain if the result is relevant and if another obviously relevant result appears nearby it's certain to be chosen over something that is vague.

Yes, we want the best possible titles, and first couple of sentences.

11

This is not an answer but an explanation of the difficulty of making the titles more general.

Here's a very recent question: Two "and" in a sentence

It's essentially asking how to distribute adjectives over 'and'. That is, if you have A C, A D, B C, and B D as things, can you say "A and B C and D" to get the four pairs more succinctly by multiplication, so to speak.

The problem is that by the time you're able to articulate/specify the question in generality (using variables, knowing the term 'distribute', etc), you probably have enough thought on how to answer the original very specific question.

I've seen this 'distribution' question more than once. Searching specifically through tags I found "Using and twice for four items". Surely there are a few more similar ones, but without knowing what to look for it's hard.

Another difficulty is removal of stop words in the SE search. Eliminating the high frequency terms in indexing text is a huge resource saver in document search techniques. Indexing 'the' and 'and' and 'a' can take up half the space of an index if they're not excluded from the process. So most document indexes ignore these highly frequent terms. Unfortunately for ELU, those terms are usually the grammatical ones people care to ask questions about.

6

When I edited your question recently, I didn't give an explanation in the edit summary; I'm sorry about that. Here's an explanation of why I prefer more specific titles (with the exact sentence in question) rather than more general titles like "Plural noun with singular verb".

  1. A number of broad topics have tags that can be used to narrow a search. For example, you used the tag "grammatical-number", and I edited your question to add the tag "verb-agreement". These tags communicate a fair amount of information in their own right, so I feel like the title should be used to provide some details about the specific sentence/example that you are asking about. You can look through lists of the other questions that have these tags (and try different filters like most frequent or highest upvoted) and see if any of them seem to relate to your question.

  2. There are a lot of complicated parts to subject-verb agreement and so often it's impossible to write a short answer that adequately covers a general topic like plural nouns used with singular verbs. Your recent question deals with plural noun phrases like "5 votes": these don't necessarily behave the same as other kinds of noun phrases that could be considered "plural" in some way, like "the United States" or "six out of ten students" or "data" or "blue-green algae". When I'm looking for a specific question to use as a duplicate target, I find vague or very generalized titles make it more difficult to find a relevant question because I don't know if the question is about the right type of "plural noun" (or whatever) before clicking on it.

So I would say your title should certainly describe the issue that you're facing, but I think you want to avoid using a title that is too similar to the titles of many other questions that don't necessarily have the same answer.

1

Titles are an integral part of a question. It may be a poor idea to alter them later; and when written, the original poster may not know how to make the question searchable in ways that would be useful to others.

In general, making answers (as well as questions) searchable is a major task that the site has not yet addressed. Search in the search box for "who whom": you get 1,436 results, the second of which is irrelevant, as many others are no doubt also.

But search Wikipedia for "who whom" and you'll get an immediate relevant hit with subsections; search just for "who" and you get a disambiguation page.

Better indexing would prevent the many duplicates (many questions are in fact duplicates and have been answered more than once, in greater or lesser detail) and remove the frustration of regular users who try to identify suitable answers, decide whether to provide brief help and vote to close, etc. It would also reduce the questions on meta that complain about questions closed as duplicates.

You must log in to answer this question.

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