Discussions and votes that are directly related to RFD closures:
Discussions and votes that are indirectly related, by pertaining to formal votes in WT:VOTES:
Some example past closures
Dan Polansky (talk) 10:20, 5 September 2022 (UTC)
Evidence shows votes are actually being tallied even if other considerations may play a role: 1) people post keep and delete in boldface; 2) people post "Delete per nom" without adding any arguments to the discussion, which only makes sense if tallying takes place; 3) some discussions contain tallies, practice used by multiple editors. The question of what overridable threshold defines consensus therefore remains, and there's no denying it.
To me, evidence suggests the best supported threshold is 2/3 (66.6%) of support / support + oppose. The supporting evidence is the following:
This is not to say that there is actual consensus for 2/3 threshold. As for 1/2, it was mentioned by one editor in a discussion but no one has tried to put it as an option into a poll. As for 1/3 minority sufficient for deletion, one user supported that in a discussion but no one else added their support. Since no one has bothered to engage in polling or other consensus-detection method to determine other thresholds than 2/3 and 3/5, there is no way for us to think that 1/2 or even 1/3 is supported by consensus. Dan Polansky (talk) 13:12, 5 September 2022 (UTC)
RFDs are neither pure votes nor pure discussions: they are a combination of votes with discussion.
RFDs are not pure votes since they contain discussions: pure votes would only include bare votes "keep", "delete", "abstain" and the like without a further comment, and people would not respond to each other. In RFD, the discussion sometimes leads people to change their stance and it influences subsequent participants.
RFDs are not pure discussions either. Evidence shows RFDs are being predominantly treated and closed as being not only discussions but also votes:
The above does not preclude that some votes are ignored as poorly reasoned, but examination of past RFD discussions shows bare deletes are not being routinely discounted as featuring no reasons. Dan Polansky (talk) 07:48, 7 September 2022 (UTC)
The text from the header was originally in Wiktionary:Requests for deletion, making it hard to track changes to it; it was put into the header by me in 2010.
The requirement no consensus => entry is kept was introduced in diff from 13 September 2009:
{{look}} to the bottom of the discussion. If there is no consensus for a period of several months, the article maybe be kept as a 'no consensus'."The period of several months was later changed to one month in diff from 9 October 2009. I changed "may" to "should" in diff from 5 February 2010, and then again in diff from 27 October 2010. Then the text survived in the header with almost no changes until diff from 24 October 2021. I changed the text back to the spirit of the discussed requirement in diff from 3 September 2022.
I don't understand the purpose of "may" instead of "should"; is it to suggest that this is just a hint to be freely overridable? Formally speaking, "should" is weak enough, although in Wiktionary policies such as WT:CFI, "should" is interpreted as "shall" and "must". WT:CFI uses a mixture of "should", "shall" and "must" with the same meaning, as far as I can tell; I don't believe the "should" parts are interpreted as mere recommendations. Dan Polansky (talk) 12:48, 7 September 2022 (UTC)