BUT earth will move with gM/m1
No. Multiplication is associative, you can switch the masses around as you please, nowhere in the formula does it say "the greater mass" or "the smaller mass" you could just as well re-arrange the formula and come up with "earth moves with gm1/M". Last but not least there's only one force acting on both objects... and gM/m1 is neither a speed nor a force. G * 100kg / 20kg is 5G. Measured in Nm²/kg² which is the same we started with because the two kg cancel each other out.
They both fall towards their shared centre of gravity. It's this "the earth revolves around the sun" thing again, no it doesn't, they both revolve around their shared centre of gravity (which, yes, is within the sun but still makes it wobble). That centre is very far away from the ball and very close to the earth and both are moving at the same speed towards it (because acceleration doesn't depend on mass), blip to the next frame of the simulation now the centre of gravity moved towards the ball, next frame still closer to the ball, that is the reason both reach it at the same time, not because one is faster than the other.
...or so it would be, if the shared centre of gravity of ball and earth wouldn't lie within the earth so they don't actually both reach it, the earth is in the way, the rest of the acceleration is turned into static friction: Because they both are still falling even when in contact. But really that complication only exists because they have volumes which is why I factored it out from the rest of the reasoning.
It's not nonsense when it makes people understand, buddy. And don't get all "oh be technical" on me when you say things like "earth will move with
<something with the same units as G>
". Something that's definitely something, but not m/s.I was talking about time-steps when I said frame. Hence "simulation", and "one frame, then another, then another", referencing successive moments in time.