QUESTION:
If the lander is not falling exactly vertically, could the solid rocket
motors carry it way off course, or even break the assembly free of the
parachute?

ANSWER from David Dubov:
Very good question. For the same reason a pendulum is rarely vertical, there
is a very low probability that the lander and the backshell will be exactly
vertical at the moment that the rockets fire. We spent a lot of effort to
figure just how bad that angle could get. It turns out that the aerodynamics
of the parachute and the backshell limit the angle to somewhat less than 20
degrees. This is the main reason why the parachute is shaped like it is. The
large (9 ft) "skirt" (the "band" in the Disk-Gap-Band parachute
configuration terminology) doesn't help slow the lander down so much as it
stablizes and helps minimize swing. Even with angles near 20 deg, we fully
expect that the rockets will do a pretty good job of imparting a large
horizontal velocity to the lander before it ever hits the ground (up to 50
mph). For this reason we had to design and test the airbags to handle
impacts at shallow fast grazing angles. So not only do the airbags have to
be good impact absorbers, but they also have to act like a giant aircraft
landing gear tire! Since we are only trying to aim the lander to an area 100
km by 200 km long, we don't mind if the lander rolls for a kilometer or
more.

As far as the paracute is concerned, during the rocket firing, the
parachute's role pretty much comes to an end. The tension on the parachute
suspension lines drops to near zero because of fast rate of deceleration. We
do not expect (and our many simulations confirm) that large angles of the
backshell during the 2 sec of rocket firing will change that significantly.