For the same reason, tracks generally end up doing more damage than wheels: you get somewhat less damage while going through the row, but they touch/damage far more ground when turning and that adds up.
The effect of large tires vs small tires vs tracks vs feet is insignificant and irrelevant, it's a rounding error compared to the effect of whether those tires/tracks will be pulling a plow/till or not.
Soil displacement is a function of machine weight AND surface area of tires/treads. Additionally, I'd expect the sharp edges of steel treads to disturb soil much worse than a similarly-weighted tractor with rubber tires.
Without doing extensive calculations, my anecdotal evidence is that I've seen bulldozer track "scars" from a single path that are decades old, while the damage from thousands of bicycles collectively veering from a paved path onto grass appears to heal itself in a year or two.
A single track every 30 meters where nothing grows is less impact on the overall field's growth than 30 tracks every meter where things still grow but not as well.
Figure 3 on page 279 explains it. https://www.researchgate.net/publication/224830979_Ecoeffici...
The reason I thought 1000 psi was low is explained on 278, the ground bearing capacity columns minimum value is 20 mpa or 2900 psi. The numbers in your link are dynamic pressures at depths of 14 / 28 inches. The area where the force is applied increases as the stress is distributed in the soil. The 2900 is a surface value.
[...More diverse crops, less pesticides, less damage to soil, less energy use, mixed instead of mono cultures.]
We already can do all of that, we don't need..
And now the potential to use less chemicals with robotic application is taking off even with large farm equipment. Cameras identify a plant off-row as the tractor traverses the field, and either plucks it or spritzes it with a microdose of herbicide.
It makes sense to me that farmers are best placed to optimize for current outputs (and probably poorly placed to optimize outputs).
Second, there's active intent to prevent people from knowing how to do things better. Almost all agricultural inputs are produced by a small number of large corporations. If farmers knew they could cut their fertilizer to 10% of what they buy now, stop buying pesticides, and spend less on fuel those corporations would lose a lot of business. So they pay "agronomists" and "soil scientists" to mislead farmers and keep them on the treadmill. There's lots of very deliberate FUD about organic agriculture, permaculture, etc, and when you've got millions of dollars invested in land and equipment, a little doubt goes a long way.