Would the same scaling for the scaled squares be achieved through the same scaling up/down or through the same additive change to the overall scale?

If we take that 60% square and add another -40% scale difference, we would get a 20% scale square, i.e. a 4×4 square. Would that appear 60% of that 60% square?

Thus, no, it would appear almost half the size of the expected square. In other words, linear scaling down is a lot faster than perceptually expected.(1a) 60%×60%=36%, — 60% of the 60% of the original size;

(1b) 4/20=20%, — the scale of the square that is the twice linearly reduced (by 40%) version of the original scale;

(1c) 36%≠20%

The same goes for the 200% square: adding +100% scale difference, we get a 300% scale square, i.e. a 60×60 square, while:

so the square we get is 3/4 of the expected one. Again, in other words, linear scaling up is significantly slower than perceptually expected.(2a) 200%×200%=400%, — 200% of the 200% of the original size;

(2b) 60/20=300%;

(2c) 400%≠300%

The case of scaling things down (i.e. the first one) is the zooming out case — you can feel (by holding Ctrl and scrolling down) that it's way too fast, especially when getting to 50% and below. And the case of scaling things up (i.e. the second one) is the zooming in case — it is painstakingly slow, try 200% and above. If we zoom out at a constant rate of 90% (i.e. 100, 90, 81, ..., 35, 31, 28, ...), we should get to 10% in ≈22 steps, not in 9, and if we zoom in at a constant rate of 110% (i.e. 100, 110, 121, ..., 259, 285, 314, ...), we should get to 5000% in ≈41 steps, not in 490. So please change the way scaling is calculated for zooming.