A client once watched a deletion discussion run its full week: nine comments arguing to keep the article, four arguing to delete it. He treated the tally as good news. The article was deleted anyway, and he wanted to know how nine people could lose to four.
They cannot lose, because nobody was counting in the sense he meant. Wikipedia says plainly that these discussions are not a vote. The administrator who closes one reads every comment and weighs its argument against policy, not its column. A keep that says the subject is well known carries no weight next to a delete that points out no independent source has ever covered them at length.
Decisions are not made by counting heads, but by weighing the merits of the arguments people give.
Certain arguments are set aside before any weighing starts. "I like it" and "it's popular" say nothing about sourcing. "Other similar articles exist" asks the closer to overlook this article's record because some other one, possibly just as flawed, survived. A sudden cluster of first-time accounts arriving to say keep, traceable to a forum post or a group chat, is read as one canvassed voice, not several independent ones.
When the arguments are genuinely close and well-founded on both sides, the discussion is not decided by a coin toss. It is relisted for another week, or closed as no consensus, which defaults to keeping the article until someone builds a clearer case either way. Whatever the outcome, the closing administrator writes a rationale, because a decision this consequential has to survive being read months later by someone who was not in the room.
We do not send clients into an open discussion looking for comments. A wave of enthusiastic new accounts is the fastest way to have an entire column discounted at once. What we do instead is surface, calmly and with our role disclosed, any independent coverage the discussion appears to have missed. That is the only kind of comment a closing administrator is obliged to weigh.
