Hypothetically, let's suppose risk of damage correlates with cross sectional area of DNA. It's conceivable that scaling up length with less critical segments reduces risk faster than the increasing cross section increases it.

It's an interesting idea from the OP. I'm basically imagining a box full of rope and someone is poking the box with a stick. The length of rope that can fit in the box increases faster than the box's cross section.

It is not square area, but cubic area (mass).