You spotted it yourself, in your reference to the soviet programme

1) Stop innovating every goddam thing

2) Stick with something that works

3) Build a lot of it, and keep building it

So, for Instance, for the Mars programme:

Build 10 or 20 MER's per year

Each year change only one instrument

Each launch opportunity, launch 2 or 4 clusters of 10 for MIRV arrival at Mars

After 6-8 years, work out what to do next

For the Moon:

(Shouldn't have dropped the Apollo Ball in favour of the Shuttle cherry)

Now: Drop shuttle, Drop ISS (both are useless porking crap)

Go for a shuttle-tank derivative with more power but expendable crew /cargo modules

Send A Lunokhod/MER exploration programme as per Mars to find out more about the surface

Begin a manned programme and DON'T KEEP CHANGING your mind...