QUESTION:
Would the project have to be started over if the airbags broke before the
probe could be safely landed or would it be able to continue from that
point? Could one of the airbags fail to inflate?

ANSWER from Cheick Diarra on November 20, 1996:
The air bags will be inflated only seconds before landing. It will take
about 1/3 second to inflate them. If at that point one develops a leak, the
pathfinder will touch down before it lose its gas.

ANSWER from Tom Rivellini on December 10, 1996:
If the airbags do not inflate, then the lander would hit the ground MUCH
MUCH harder than it was designed to.  If the lander was going slow enough
(because the rockets and parachute happened to work better than expected),
the lander MIGHT survive and still open.  More likely, though, the impact
with the ground at about 30-40 mph would be rough enough to damage some very
important parts of the lander and prevent it from performing its mission.

ANSWER from David Dubov:
Yes, anything could happen! We all know Murphy's laws. But we strongly doubt
it. Why? Because we tried to design the bags so that it is very unlikely to
happen (that's our only defense against those laws of his). We asked
ourselves, "What might prevent one or more of the bags from inflating?".

Of course, it is possible that the EDL (entry descent and landing) software
could somehow forget to schedule the airbag inflation (it turns out the
software could forget to hit the "button" and it still works due to a timer
backup system we designed). Testing and verifying software takes a lot work
(we are still at it!).

Each of the four bags had to be "strapped down" to their respective petal.
To untie them the software must command a (dual) relay closure which applies
voltage to a pair of (dual redundant) explosive squibs (also called
pyrotechnic initiators) which activate cable cutters. There is a cutter on
each of the four bags that all fire at the same time. If a cutter doesn't
fire then that bag won't inflate. Fortunately, we know from experience that
these cutters will cut provided one of the two initiators fires; so we made
sure that they fire by using dual redundant circuitry and power to the
initiators.

Once the bags have been unstrapped (this occurs 1/4 sec before the gas
generators ignite), the three gas generators (GGs) are ignited in the same
manner as the cutters, each with dual redundant circuitry and initiators.
Although the bags work best when all three GGs work, if only two of them
work, then all four bags will still inflate because the bags are
interconnected. They are interconnect so that they can share gas and
equalize the pressure among them. In this case the bags will still do an
adequate job of protecting the lander from impact provided that the impact
conditions are also not the worst case.

To make this complex system reliable, we had to balance risk throughout the
design process, juggling the cost of additional complexity with relative
reliability.