This is a nice presentation of the changes made to Discovery.

I read Feynman's Surely You're Joking, Mr. Feynman (I hope I have the right book reference) where he talks about the Challenger disaster and how his analysis showed the risk of catastrophic failure due to the design of the shuttle was about 1 in a 100, and not 1 in 100,000 as Nasa had claimed.

It turns out he was wrong - based on actual history, the chances of catastrophic failure are about 1 in 50 - there have been two disasters out of about 100 launches.

His concerns had to do with the design and not so much the implementation. The design was just too complicated and worked too hard to defeat physics. For instance, a good nuclear reactor will shut down automatically if the water leaks out. This is done by making physical linkages that put the rods into the reactor automatically - it's not done by a sensor, or a computer, but by the nature of the physical world.

The shuttle design flies in the face of that and has all kinds of little hacks (the more recent ones are documented in the link above) to 'work around' the nature of the physical task it is made to perform. (My favorite is the 'bolt catchers'.)

So, the odds are good that the shuttle will fail catastrophically again. And it will be in some system or subsystem where the engineers already knew there was a problem, as happened the last two times.

And it will be very sad.