I wasn't aiming to disprove the hypothesis, merely looking for strategies better than my own.
In this case, I was also the person dealing with the person reporting it. Having had to close it with "cannot reproduce, won't fix" would have been a pretty disappointing outcome, and would make it really hard to discern between a 10x engineer and someone who's just bad at debugging...
They also might have looked at the code and said "oh hey, that won't work with multiple windows, I'll just change that".