Every day you go through things that have a chance of killing you. If you have first-hand knowledge of a friend of a friend dying while driving on the 401 to the cottage, then the odds are set at 1 in 10,000. Stunt odds are about 1 in 1000. Pouring gas on a fire with a plastic can is about 1 in 100.
Flying is about 1 in 10 million per trip. The old space shuttle was close to 1 in 100 by practice.
Engineering failure of a single component, like a clamp is about 1 in 10,000 for aircraft and nuclear-grade parts, and about 1 in 1000, for non-controlled parts. If you want to engineer something with lower chance, then you avoid single points of failure causing death. That's why window-washers have an extra rope. The odds of two independent failures multiplies the single failures. For ratty-tatty window washers, it's 1000x1000 or one in a million.
In the old company, I found a lot of 'handyman specials' which have a 1 in 1000 chance of failure. For nuclear plants we want to get up to 1 in 10 million. There are much fewer nuclears than airline flights. They want the odds at 1 in 10 million per flight because there are so many of them, and one airline crash suppresses travel plans.
This submersible can't be any better than the old space shuttle. You can see the backup -- there isn't any. A submersible should have a large tender, and several redundancies. A failure of a simple connection is death.
ps. a catastrophic implosion. Sounds like fatigue, but we'll never know.