Here's a still admittedly somewhat lengthy attempt to summarize my more detailed thoughts:
I agree with some of AndyT's concerns, but questions which are difficult to research should be answered, in order to achieve Stack Exchange's goal of disseminating information.
Proving a negative is harder than proving a positive. It is at least hypothetically possible for somebody to try and do research and have absolutely nothing to show for it, or at least nothing worth showing, which leaves them with little to no recourse for reopening the question, even if it is worth answering. This is most particularly the case for when people who do cursory research with a search engine and find no reference websites with their search, and other forms of research can be harder.
I doubt delimiting strict expectations will do much to help us. It would be one thing if the clarification could raise compliance with our standards, but we have enough difficulty getting people to do much of anything without asking even more of them than what we presently enforce, and any amount of shown research shows that they did the least they could do for the people bothered by that. We would probably only see more questions closed and fewer questions reopened, because it will require more work to keep them open or mend them.
If the main purpose of the research requirement is to reduce the number of answers that solely rest on an appeal to authority from a readily accessible resource, on the basis that questions which can be answered that way do not merit our personalized assistance, then one reference usually suffices to highlight the uselessness of answers that just duplicate co-equal resources, since the veracity of information of one co-equal authority like can not be trusted more than another like it without further proof or explanation.
Unfortunately, in cases where no productive results are found on a search engine, the questioner providing us with their search terms probably does little to nothing to help us, even under unlikely optimal circumstances. They will not always reliably produce the same results that the questioners and other visitors may find, even in the best case scenario, so we gain little to no insight regarding the nature of the problem. It also requires us to manually double check anyway, and if we are going to bother doing that, then I trust we are capable of devising our own and evaluate if answering the question is needful for ourselves.
I would like to add that there may be ways to improve the phrasing of our current closure reason, but I doubt delimiting stricter standards as suggested will help for these reasons.
Look, I know that strictly speaking the present research standards endorse checking multiple resources, but we have never actually gone so far as to enforce it to my personal recollection, and it may just be too high of a threshold to be reasonable, so I do not think we should start.
Firstly, I have difficulty seeing much of a practical benefit, if the of our research standard goal is simply to bar questions that are too simple. If a user checks even just one authoritative reference work, then I think it does enough to solve the problem of an overly simple question because who's to say,for example that Merriam-Webster's 11th Collegiate Dictionary defines a word better than The Oxford Dictionary of English 3rd edition, or vice versa, without a more thorough examination of the circumstances and facts that can be used as a basis to explain which one did a better job? Moreover, if all of the dictionaries are in perfect agreement, then checking one is more or less effectively the same as checking them all anyway. My opinion is that checking even just one reference work should be enough, but we often have difficulty persuading new users to even do just that and I think the community would be elated if we found a way to encourage even just that. Referencing even just one authority does enough to prevent merely making an appeal to an equivalent authority.
Moreover, we often have difficulty persuading people to edit questions to add any any amount of research, and I think we would be very happy if they did. We should keep in mind that if the research standard is too high that we may start losing questions that could have been answered very well to other resources.
However, more importantly, I doubt making the standards clearer and stricter solve the problem you are experiencing anyway. I think the root of the problem may be that they are too clear, and too strict for us, because they inadvertently cause a problem. Questions are just closed whenever they lack research, without any effort on our part to determine if research would have much practical benefit. Now do not get me wrong: I am of the opinion that a question with research is better than a question without it, but consider this:
Showing an absence of evidence is much harder than proving a positive. How should people even show just one reference to a question which doesn't have an answer, let alone three? Do they give us a link to a nonexistent webpage? Do they read an entire book that might be expected to have the answer so they report that it does not?
Originally the research standards were established as a preventative measure for questions that are too easy to answer with "commonly available resources", but the way it is now, where we needn't necessarily check for ourselves if such is the case may also inadvertently bar questions which are too hard, except I do not even think that was ever the intention of it. That does not mean that there are not some questions which are too hard to answer, but we have separate guidelines for those which are handled by different closure reasons specifically designed to ensure that we only close questions that are too hard to answer when answering them is not productive, which are Too Broad, Primarily Opinion Based, and Unclear What You Are Asking.
Now it could be argued that those questions are P.O.B. for lack of reference, but I think the guidelines for P.O.B. are more sensitive and fine tuned than that, since other forms of corroboration can be used in lieu of direct reference. The way the research standard is set up now seems like we may only answer questions if an answer exists in an easily found reference work that somehow fails to explain the matter sufficiently. I think there are many more questions of interest that we should be addressing than that.
Fixing that problem may actually require experienced and privileged users to exercise more responsible discretion with the vote to close options, and actually check to see if a commonly available resource actually even exists more often, rather than just assuming that it does not whenever the questioner fails to provide research pursuant to Jeff Atwood's guidance in Basic Questions Are Not So Basic.
The initial closure of the Emoji/Emoticon comparison question would have been averted if we did that for example, and that's not the only time it has been a thorn in my side, as somebody who has not as of yet even asked a question, and usually tries to write lengthy in-depth answers. Sometimes I think we need to concern ourselves more with answer potential than question quality if we want this to be the best website it can be. (Which is not to say that question quality should be ignored, but again, I doubt that raising standards even higher than they are now will help much with this particular problem.)
There is another half of the proposal, regarding sharing search terms. Unfortunately, irrespective of what our research standards are I do not think that will help us very much, because even if we know all of that, it might not help us very much. The problem is that they are inconsistent.
One way they can be inconsistent is from search engine to search engine, since they are all competing to deliver better results than the rest.
Even on the same search engine, if they are personalized according to location or search history. Google, which is the most popular search engine right now, does personalize search results somewhat. The C.N.B.C. tech article We sat in on an internal Google meeting where they talked about changing the search algorithm—here’s what we learned, written by Jillian D’Onfro and published on September 17th 2018 states the following:
Other times, ideas for algorithm changes come from broad company directives or priorities. For example, some employees have long argued that Google search results should be more personalized, Nayak said. Right now, there is very little search personalization and what exists is focused on a user’s location or immediate context from a prior search. (If you Googled something related to baseball followed by “The Giants,” the results wouldn’t surface the football team, for example.)
Even if we take this statement at face value it has significant consequences. For the purposes of verifying research, it is problematic enough that somebody in the United States of America might not get the same results as somebody in the United Kingdom for dialectical questions. It also seems possible that we might get different results for people who study punctuation more as opposed to people who study syntax. Ellipsis is often used as a syntactic term which refers to the practice of dropping words out of sentences, but then there is the ellipsis as a punctuation mark that represents the omission of words from quotation.
Even if we take steps to reduce or eliminate the amount of personalization in our search engine by changing Google's user settings or using Duckduckgo instead, the results are also transient. They change with time as search engine optimization improves and websites pop in or out of existence. Over time, some questions which are likely to be general reference today might not be general reference a year from now, or questions which need answering today might be general reference a year from now. In the latter case, I believe it is better if we get to the question first: General reference often offers sparse explanation and rests on the authority of the source, and we can give questions much more thorough treatment and share our evidence if we get to them first, so every time we are beaten to the punch represents a wasted opportunity to optimize the utility of our answers not only through quality, but through earn greater exposure and offer better timeliness than more slowly updated references.
Another thing we should keep in mind is that one of the founding principles of Stack Exchange is to make the internet a better place by making it easier to find information through search engines, so we should probably keep an eye out for when we can serve that purpose.
Since search engine results can not reliably communicate to us what the questioner did or did not see, they don't prove to us that the questioner couldn't find anything useful. Our audit of those search terms also won't be able to prove that they could have easily found something useful, or what they are likely to have seen. I will admit that it is perhaps somewhat more useful than nothing at all, but again, I fail to see much practical benefit because it does not offer much more information for achieving our practical goal.
The only thing it really does is give us a starting point for an audit, but if we are going to go through the effort of doing an audit, we can probably do nearly as well through some cursory guesswork, and decide how to handle the question from there.
