Pictures Only A Scientist Could Love and Prospecting

I've been spending a ridiculous amount of time at the Phoenix Mars Mission site. The engineering involved in this mission is just staggering and for it to have been so relatively error free up to this point is almost unthinkable, especially considering how much software is involved.

But all that planning and building and testing and travelling the hundreds of thousands of miles to give us pictures like this:


It's dead, Jim. Can't we at least get a faux-face print in the sand or suspiciously stacked rocks? I don't need little green men. But please, let's get a little bit of mystery.

I know. I know. They are going to dig for signs of water and potential building blocks of life to see if maybe once there might could have been primitive life on the planet. And yes that will be very cool to find out. For the record. Here's my prediction. The planet is dead. Always has been. Always will be. I predict they will find not even the faintest hint of a possibility of life there.

When I was at Balticon I sat in on an presentation on the current state of AI in computers and an overview of the different approaches to AI systems. Along the way the presenter talked about some decision that were made about how much AI to put into the phoenix mission and how much autonomy to give the lander. The trade offs are that if you give the lander more autonomy it can react to threats more immediately because it takes multiple minutes for it to phone home information and wait for instructions. On the other hand, no one quite trusts AI systems to make the right decisions. In the end, the scientists and engineers opted out of giving the lander. The idea being that buy putting it under direct control of the earthbound crew they would likely be able to get some data out of it no matter how badly things went wrong. The speaker at this panel said that they'd rather have a sure chance at getting 10% of the data they were hoping for than run the risk of getting nothing because something went horribly wrong.

In my amateur study of risk management for my job in IBM, I've had the chance to read up on prospect theory, which is the study of the idea that human beings do not respond to all risks the same way and in particular that people do not want to put at risk what they already have or think they can have, but they will have higher risk tolerances when they think they can minimize losses. It seems to me that the decision to not give the Phoenix lander autonomy is consistent with this principle, which in some way, is more interesting than any of the pictures I've seen so far.