We search by question, not answer
People generally search for questions that may have the answer[s] they're looking for.
When you type stuff into the search box (or click a tag), what do you get? Questions. You get a list question titles and a blurb from the question body. If one of the questions doesn't quite seem like what you're looking for, even if it has the answer you need, it's going to get looked over.
I am not sure how the searching function works, but I suspect that questions marked as dupes get suppressed by some factor that favors what somebody thought was the "original" question. So, stuff gets missed on site search (and caught on a web search).
Similarly, when you create a question, you get a list of "related" questions that may or may not be "duplicates" based on your question title. Again, if one of those has the answer you want but the suggested question seems different, you're going to miss that and just keep typing your "new" question.
I just don't understand the urge that many people here have to categorize things as "the same" or duplicated when there are clearly nuanced differences. Instead of crudely marking something as "a dupe", why not have some way of marking it as "related" so that if somebody comes across a question, they can easily toggle to related questions that may be a better fit.
Words matter. "Dupe" suggests something unnecessary. Yet, they still exist on the site, people read them. People even take the time to answer them (before they're locked out of answering them). I understand there may be editorial reasons for some questions to get marked as duplicates, that is more likely to happen with simple questions which aren't very interesting. On workplace, however, we tend to get complex questions that describe complex situations and stimulate complex, faceted answers. It seems ham-fisted to me how such questions can get marked as dupe. It doesn't seem to solve a problem other than some peoples' need to categorize things with smug rigidity.