"rk" wrote in message
...

I think I'll disagree, at least with respect to a lot of the avionics,
particularly computers/digital electronics.

Going back to Apollo technology, there you had engineers designing custom
CPUs
from gates for a custom instruction set with essentially no design tools
along
with the software to support it. The program memories were hand woven in
a
semi-automated process. Clearly we've come a long way since then.

For Shuttle computers, again we see CPUs and bus protocol devices (and
there a
lot of data buses on the Shuttle) designed with relatively small scale
integration. Today that job is much easier and 1553 protocol devices, as
an
example, are available either in a hardened chip or as a "soft core" that
can
be put in a corner of a relatively small IC. For the main engine
controllers,
there was a lot of work and risk involved in the plated wire memories.
Today,
that is a no-brainer, with hardened/hi-rel memories of very high density
available off the shelf, that are relatively trivial to use, even in
hostile
environments.

Along with the relative ease of design, the volume, power, and mass have
also
been reduced considerably while the reliability has gone up by a large
amount.

For the past 30 years, it's been common knowledge that the cost of any type
of system that uses general purpose or special purpose computers is driven
by the high cost of software development, not by hardware cost. This was
true in the Apollo era and it's true today. Case in point, Boeing's struggle
to develop the ISS software far exceeds the cost of the computer hardware on
the space station. Certainly the ISS computers are a thousand times faster
and have 10,000 times more fast memory and a million times more mass storage
capability that the Apollo era computers, including the shuttle flight
computers. And the cost of developing a debugged, verified and validated
line of computer code may have dropped somewhat during the past 3 decades,
what with all of the spiffy automated software development tools we have
now. But the size of the ISS software is measured in millions of lines of
code, while the Apollo and shuttle flight software has or had considerably
fewer lines of code.

Later
Ray Schmitt