107344732772682756

SPACE: As you should know, the Mars lander Spirit is now on Mars and sending back pictures. Nova had a great show a few days ago about building the landers that I really liked. It was a great look at all the work and science that goes into a huge endeavor like this.

One thing the show pointed out to me, unintentionally, was the problem with the NASA/government science way of building projects like this. The engineers from the JPL were very proud of the fact that they built every single part of the landers themselves. To me, this is a colossal waste of money and engineering. If you build everything yourself, you have to test everything yourself and then test everything in conjunction with everything else. Buying parts from people who already know how to build motors or parachutes means that they can use their expertise in that area to get you a product that works. And they'll do it for a lot less money probably. Of course that deprives you of the right to say you did everything yourself, which for a lot of scientists is not something they are willing to give up.

As big of a fan of this project as I am, I wish they were doing things quite a bit differently. Instead of building 2 very expensive, very overbuilt robots I would love to see NASA go for the "Fast, Cheap, and Out of Control" plan espoused by MIT robotics wizard (and Roomba vacuum inventor) Rod Brooks. Brooks's idea was instead of building 2 irreplaceable $5 million robots (for example), you build 100 $100,000 robots. Then you can use a less over-engineered landing strategy and spread those guys over the surface of the planet. Even if you lose 25 or even 50% of the robots, you've still got 50+ robots roaming around. Even if each one is only 1/10th as powerful in terms of science ability, you have many times more overall scientific capacity, and its spread out over a greater area. Since we're just at the beginnings of scientific exploration on Mars, it makes more sense to do a shotgun approach to exploration than the laser-like focus of the current system. When we have initial data and we've used cheap robots to find better landing strategies, we can send the super-powerful robots and really get into the hardcore science.

Even off-the-shelf robotic components are very powerful and although they might not be engineered to JPL specs, I would venture to say you could build a far more than adequate robot with pre-made industrial components. Switching from the usual monolithic top-down system of the current NASA to a more distributed approach would be a pretty big paradigm shift so I'm not holding my breath. It's something to think about though. If you happen to know any NASA bigwigs, pass along the idea. :)