Silicon Valley vs NASA

Dennis Wingo writes in SpaceRef that in his judgment we can go to the moon with better and cheaper systems than before: calling the previous hardware "stone knives and bearskins." As examples he mentions the far faster and smaller computers available now, the much better alignment and positioning and inertial guidance systems.

All true. But he forgets four big problems.

1. Reliability. Just because a computer works fine here on earth doesn't mean it is radiation-hard enough to work in space. How do you know if it works in space? You have to test it. Testing takes time and lots of money. Dennis says "1969 the Lunar Excursion Module (LEM) sported the most advanced and compact computer built up to that time. " I'll bet it didn't--in fact I've heard the exact opposite. It was undoubtedly the best when it was built, but by the time the testing was done, the hardware would be behind the curve. Thing is, it was known to work, and the bleeding edge stuff wasn't.

   At some point the ever-smaller size of transistor gates starts to make them very sensitive to transitions or damage due to cosmic rays. (Our atmosphere blocks a lot of that.) If you want to use them in space you have to design for redundancy or for error checking to a degree not required for earthbound consumer electronics.

2. Systems engineering. OK, you've got the parts. They have to fit together. If you get a newer/faster/cheaper part, it doesn't save you anything if you can't fit it in the old position. If chip A doesn't fit anymore, you've got a redesign problem on hand: maybe a big expensive one.
3. Spare parts. Oops, maybe you don't have the parts. By the time you're done with systems engineering and testing, a few years have gone by. Where are you in the life cycle of the product? We've several boards we can't find components for anymore--and they're not very old... There are new products out there that do almost the same job, but their characteristics are slightly different, and they won't work.
4. Where's the beef? Ok, so you can save a few hundred pounds in designing a new Apollo capsule. Very good, you can use that capacity for lofting the tools you need for digging a moon shelter. But how does the decrease in the price of an Ethernet card change the price of a rocket motor, or the price of the liquid oxygen tank? Some things are still quite expensive: engineering, testing, and the beef needed to put the gadgets in orbit.

Sure, I'd like to go to the moon. But I think Dennis is overoptimistic about the cost.