The problem is not so much the question of politeness I think as it's the question of how to value the content of a participatory website.
That a third pencil has no value and is not taken if offered does not necessarily mean that the first two taken have no value or had no value and should not have been taken and that if we have them in front of us we best get rid of them.
I suggest leaving questions and answers that are well received be and letting current preferences of site users who consume the site content filter additional content primarily.
That means once well received site content that would not have been well received much later should not be closed or removed.
It follows from a counterfactual analysis of the context. They are well received (otherwise they would be closed). Ultimately the site content is for the users. It is judged by their preferences because only they consume it. (Therefore it has value only so far as they value it and neither more nor less.)
So this content is not judged by an abstract standard of value above and beyond or other than their preferences at the moment as they are demonstrated. There is no entity with such a standard consuming site content.
This is all because of the participatory nature of the site.
Regarding precedents A.F. has a good point and I further suggest that the community do not consider any past site contents as precedents for deciding whether to take or leave additional future site contents.
The questions of taking or leaving are entirely questions of value. Consider the demonstrated value (votes total) to the community of site content as its value and let this solely determine whether to take or leave any given additional content and whether it keeps or removes the content taken in the past.
Actual site user preferences change. That an additional object is such that it satisfies past preferences if it were present in the past but was not so present and did not satisfy past preferences in fact does not give it any present value. This is determined solely by present preferences. So precedents cannot be valid rules in a participatory site regarding content.
So site content that was not well received and does not become well received simply is not well received.
Deciding by precedent the taking or leaving of additional site content is disregarding site user preferences and reducing its value.
However this means since we can ignore all precedents that there is no reason to close or remove (so that it cannot be taken as a precedent) the site content that was well received in the past but would not be additionally well received in the present if it was not already well received.
That something additional like it satisfies no other preferences, and would be left or removed does not mean that it the original satisfied and satisfies no preferences and should not have been taken or should be removed.
Concisely I suggest that altho much earlier desired content wouldn't be desirable now, this isn't a reason to do anything about it once it's on the site.
Preferences change not merely in eras but all the time. Current preferences don't imply much about actions caused by past preferences unless they are harmful not merely useless from the new perspective.
We get a better chef's knife to replace another which doesn't hone or hold its edge so well but usually we don't throw away the other one unless it's harmful (broken) or temporary (plastic).
Yet if precedents don't matter no harm exists. Why do anything if most users didn't feel the need to downvote reverse votes when their preferences changed, while some users may positively still like the choices they made?