Often I notice that the search function seems a bit off.
Don't get me wrong, I've found lots of great questions and answers. It just appears, to me, that it heavily leans towards getting questions answered at the cost of leaning somewhat away from finding answers to questions. Which makes complete sense.
Possibly the greatest thing about SO is the great wealth of information left behind, but it can be rather difficult to find some of that information. This is especially noticeable to me when duplicates arise, yet the keywords in the title and question don't yield relevant results. IMHO, this is because people are trying to find their answers but don't.
I don't know what formulates the results, the only solutions I've personally thought of were a very clunky second search box or some sort of advanced search with a different style of search possible for finding archived results.
Seriously feel that there needs to be some work done there to make good answers more accessible.
The only time this has happened to me I feel is only a mediocre example. I have a question "Hard Copy vs Soft Copy" over at programmers. I looked around for a while but didn't find the question that later people said mine was a duplicate of: "Is it common to print out code on paper?" In this instance it was more my fault because even optimally, I wouldn't expect the search to find "Hard Copy" and "print" similar enough, and also I should have more logically just searched for "print". I've seen much better examples but they come and go and I forget them.
I feel strongly about this issue though because I am sure it is affecting many users, especially the newer ones. Is there a way to look at only closed questions? Some of the patterns are somewhat obvious in them.
I received a request from the system to detail why this question is different from A new search engine for Stack Exchange
First, this question refers to an entirely different search implementation from 2011 and the supposed duplicate does not.
Second, the other question is basically an RFC. Not asking for specific alterations. It is possible my critique of the old search could be a relevant answer but bears little resemblance to the question.
There ARE some references that strongly imply, though not fully spelled out, that questions and answers are differentiated these days. I have no knowledge of current SO search semantics but that is not surprising a decade later. A better answer to this question could likely be proposed.
Rules, audience, topics and etc were different then. Really this question is not a duplicate but does serve as an example Q&A that implyies a need for archiving without removal.
For anyone jumping in:
- Old SO search promoted questions with no or few answers, and questions with unaccepted answers, rather highly. Mining for answers to check duplicates was sometimes a hassle.
- Lucene replaced that, if not another implementation
- Apparently Elasticsearch has replaced Lucene
Much of the above is now only historical.
These days it is somewhat like playing Jeopardy when you phrase a question; the content does not matter so much as the formation. Reframing into our current scenario is something like:
How can the benefits and tradeoffs of enabling search to distinguish between a request for questions and a request for answers be weighed?
Nothing from the referenced potential duplicate comes even in the ballpark of answering to the utility of discerning the two or whether the implementation of such a feature is worthwhile.
Finally the referenced question came much later and would therefor be a duplicate of this question, were they actually similar at all