>Just writing the
>software for the lander would likely cost upwards of $100 million (good geeks don't come
>cheap, either) when you count all of the testing necessary for such critical code.

This is a good example of all that is wrong with the aerospace industry.

Yes, I can easily see Lokheed or Boeing employing a huge team of software developers using the latest CASE tools, and burning through $100 million.

But it would be ABSURD.

Going to the moon has not gotten any harder in the last 30 years. The software that was used for that task run on a computer with less computing power than a modern dishwasher. A sensible developer would take advantage of the capabilities of modern hardware by working in a higher level language, avoiding the error prone resource optimizations that were necessary then, and give it a decent user interface. At that point it would have the functionality required for the task. Everything else is frosting. The original software was only written by a handful of people, and if you were given the same task with the resources of today and the benefit of hindsight, it is a job for a single developer.

John Carmack
www.armadilloaerospace.com
www.idsoftware.com