Unless you have to, with searoom and supplies, etc. -- I'm not sure why they didn't fall off. Close reaching into relatively steep or big waves really isn't something that a cat (especially one that a skipper isn't experienced in and "tested") seems strange to me.
They were on the west side of the Gulf Stream (as I recall) about 1/2 way to Bermuda.
From my discussions with big cat sailers (including one long discussion with a guy that had taken Playstation into three or four Oceans), cats really don't want to weather at sea. The guy from Playstation said his most difficult time was off of Nova Scotia when he got caught without a lot of sea room and had to stay close to the wind. He said that was one of the worst times he ever had on a boat (mono or multihull); and, he's got a lot more sea miles than Max Fletcher ;^)))).
By the way, one advantage of the Cat -- as one friend mused, he would have considered dumping the rudders and using the engine(s), sail, drogue to head for Bermuda (or even Norfolk). But as he said, he wasn't there.
I certainly don't have any offshore experience in multi-hull.
As for DDW's offer for a bet with Larry in a Dragonfly, I guess I agree with Jon (DDW) on the J/133 as long as: i) the wind is in the 25-30 knot range, and ii) the waves are not swell-like (i.e., not just rollers).
By the way, DDW: The analysis that you provided on the difference of section modulus and bending capacity was for a static load. The effect of shock loads in a variable flow increases the real effect on the system. It's not really something that can be calculated in the real world is it? You really have a variety of variables that you have to make assumptions of flow around the hulls, the variation of action in the trough, moving wave "train", etc.
You aren't just "dropping" the boat (accelleration) -- the whole medium of water is moving in different ways in combination. The angle of rudders in a particular situation is also and issue. One has the restraint provided by the autopilot and the stops on the rudder rotation, etc. working also.
Anyway, what is interesting is one rudder seemed to fail by detachment of the shaft from the rudder itself; and the other was bent and/or buckled so that it was essentially jammed.