You see the chances of you having your real question. Answered are directly related to the number of people reading it. (With the added complexity of it actually having to be the right person reading it but why get overly complicated?)
If you decrease the signal (real question) to noise ("fun" question) ratio, you increase the amount of garbage and reduce the number of serious questions that can be answered.
Imagine the "endgame" of fun questions over time, what if they start allowing "rage comics" about programming (f7u11overflow.com) ...
There is absolutely a place for "fun" on the internet, there is absolutely a place for "fun programming questions"... But it isn't stack overflow.
I'm not sure I fully agree with the tools research questions being OT, but I understand that they cannot have objective answers. Compare for example the answers to the question which IDE should I use? The programmers at JETBRAINS logon and say "jetbrains" then some MS programmers log in and say "VisualStudio" then the team that wrote eclipse log in and say "actually yeah, visual studio... Eclipse sucks"
Now someone has to put a green check mark beside one of those answers or leave it unmarked as unanswerable. Both are unacceptable really.
Best to avoid it.
I'd even argue that in the early days the "fun tone" made it more of a hangout and 3rd place rather than the boring and dry site it is today.
Fun questions could be tagged (and even hidden).