I agree with the points made by @parker in the question.
The problem with "duplicates" is that a new question asked in totally equivalent terms to another older one, may not in fact be asking the same question, and will be dealing with different underlying assumptions by the questioner about what the question means and what a satisfactory answer looks like.
The problem with @EmilioPisanty's view that "the onus is on the poster of the new question to edit their post so that it delineates itself from the previous question", is that the new questioner may not be able to clearly articulate how their question differs from an older one, or why answers on the older question are not satisfactory.
In other words, they may have to ask questions about why their question is not clear.
It is usually only possible to formulate a question precisely once you broadly have the answer to it and have integrated the two, or have examined the question with the assistance of others, so expecting precise questions puts the cart before the horse on a site designed to answer questions.
Even trained academics do not formulate questions with perfect precision, or re-examine every assumption of their field of expertise in doing so - usually, their work carries so many assumptions that questions cannot be understood at all unless one has been through a process of gaining tacit knowledge of the field, which is why we have a system of education.
There is also the problem that, with very old questions and answers, none of the contributors may still be around to discuss them any further. There may also be a "clutter" of old discussion or shared assumptions, that new contributors do not want to be burdened with.
The structure of the site and the way its notifications work, also makes it difficult to be clear that an answer on an old question, is really intended to address a person who created a "duplicate" which was then linked to the old question. The person asking the duplicate also cannot mark the answer accepted, since that is a right which belongs to the person who asked the original question. This is probably why links to "old" duplicates, get little attention.
Also, the new questioner, whilst not rejecting answers to the existing question (or maybe simply holding that it is not understandable and cannot be evaluated), may also be seeking a fresh or alternative perspective on the same question without necessarily knowing how those alternatives will differ - and these can obviously only be solicited on this site by putting the question again.
Either way, if the questioner's position is not explicit, it may be necessary to pump the new questioner for further information on the question to establish whether they have merely overlooked an existing answer, or whether they think it is insufficient in some way, but such dialog obviously will create an administrative burden - quite possibly more burden than simply re-answering a question, or the dialog may simply start to resemble a re-answering.
There is also another beneficial aspect of true duplicate questions, which is that it allows those answering to rehearse satisfactory answers to the questions (and be assured that we have satisfactory answers), gain an insight into exactly how or why certain questions may be misconceived, and re-examine our own knowledge and ability to explain in the context of what may be a new set of contributors to the site.
Obviously, one does not want the same question asked separately every 5 days, but it might be reasonable to think that after as little as a few months (and certainly after a few years), the population of contributors may have changed significantly or totally - and that even if the questions are the same, new contributors may want to make a fresh attempt at answering them.
I would suggest that, unless a question is clearly about the formalism of physics and has a straightforward formal answer, we should hesitate to assume that questions ought to be treated as duplicates, merely because the form of words used in the question seems identical to that in another question.