On Sat, Jan 4, 2025 at 12:22 AM Bruno Haible <br...@clisp.org> wrote: > Hi Jim,
Hi again :-) > > I'm surprised that you would use "more sources use Y" as a rationale > > for using Y. > > I do so for language-related topics. Because language gets defined by > how people use it in their majority. > > Like you, I'm not generally in favour of accepting majority votes > - for determining what are facts or > - for topics in the domain of programming. That is a fine approach for determining which spelling of a word to use, but not for which algorithm to use -- the majority is often satisfied with a suboptimal solution. I think the same risk applies to prose. For determining what may be a prose-related best practice, wouldn't going with the majority ultimately lead to reducing quality to some merely "good-enough" practices? Gnulib sets a very high bar for code quality, but that's relatively easy to measure objectively. I would argue that it's fine (and expected) to choose prose guidelines that select less-common constructs, because estimating the quality of documentation is fuzzy and subjective, and fewer people care enough to spend the time to get "mere documentation" right.
