BlackWater wrote:

Sorry, but for this endeavour to be viable we need
an EXTREMELY low failure rate ... one in five
thousand perhaps, or better. ...

The Russian and American human rated launch systems
in use today have a mission failure rate of roughly
2% - meaning loss of crew. They also have a 2-3%
launch vehicle failure rate, which doesn't necessarily
coincide with crew loss. On two Soyuz missions, for
example, crews survived launch vehicle failures.
Three of the four fatal human space flight missions
(two Russian, two U.S.) involved failures during
descent. As for China, one of its unmanned Shenzhou
test flights reportedly ended with a parachute failure.

... It CAN be done ... but
it means not using bleeding-edge technology or
engineering and it means a major effort to simplify
and standardize too. ...

Space launch reliability in excess of "three-nines"
isn't going to happen with current technology. The
current state of the art in space launch vehicles are
the EELV launchers built by Lockheed Martin and Boeing.
These vehicles were designed to have hardware-related
vehicle failure rates of less than 1%, but non-hardware
failure modes like bad software loads or bad ground
processing steps (e.g. forgetting to remove a piece
of tape from a connector pair that is supposed to
separate - something that really happened) now account
for more failure modes than the hardware - resulting
in the total predicted failure rate for these newest
of rockets being roughly 2%.

At one time, NASA convinced itself that shuttle was
0.999 reliable, but 2 failures in 113 flights proved
otherwise. Given this reality, it is clear that human
launch systems must have crew escape systems to improve
survivability. But launch escape systems don't handle
reentry phase failures, or Apollo 13-type in-flight
failures. The risk to space flight crews will be very
real for the forseeable future.

- Ed Kyle