Here's why things get interesting.
If you turn back the clock 12-18 months, you would find the board chock-full of posts "begging" for more variation in the play-by-play text. I remember those posts because I took part in many of them.
Many wanted more variation, and others offered that no AI code was going to simulate a real human being - no matter how hard you tried.
Although the Play-by-play could be improved - and it was - it would always be short of a real announcer.
So, here we are today, basically saying the straightforward "play results" would be better than the attempt to simulate a real announcer... the pendulum continues to swing
Not taking either side, but it's interesting to note there is no answer on most things that will please everyone.