In the midst of an article in which he recommends that the administration encourage the Chinese to race us to the moon, Dwayne Day writes: "There is nothing that a human can do in low Earth orbit, other than the study of other humans, that a robot cannot do better."

I hope that he didn't give very much thought to that statement, because it's demonstrably untrue. Could a robot have done this better? How about this? Or especially this, which happened over three decades ago?

I doubt that we have the robotic capability today to do those things, let alone at the time. Dwayne can argue if he likes that they weren't worth doing (I would disagree in all cases, especially in the case of Skylab), but to say that there's nothing that robots can't do better than humans in LEO is...mistaken.

Interestingly, of course, this is being discussed as an alternate means to save Hubble, but it will clearly be a technical challenge, and it's not being done because it's a better way, but because NASA is unwilling to send a crew.

Rand, you were correct to point out that Day's statement about "nothing huans can do in LEO that robot's can't do" was incorrect, but not for the reasons you stated.

In order to definitively state that there is nothing that a human can do in low orbit that a robot cannot, empirical evidence provided by the operation of a teleoperated robot is required, to provide a comparasin to human performance. So far the only data available is for the Canadarm on the shuttles and the Canadarm II on the ISS - in both cases, teleoperation is from a distance measured in a few meters rather than hundreds or thousands of kilometers, and astronauts are close by to jump in and assist if needed. There is no data for any other robots operating in low earth orbit, so no meaningful comparasins can be made.

"nothing humans* can do... that robots cannot"

remember, robots don't make spelling mistakes, people do

Um .. dont misunderestimate what modern robotics is capable of these days.
Robotics has and is still undergoing a kind of quiet revolution, that began in late 90ies due to several factors coming together. Its not covered often by western media, or when it does it gets severely skewed opinions.
Pete Markiewicz wrote a short summary here: http://www.roboticstrends.com/displayarticle364.html , he's been talking about those trends at length over @ http://www.plyojump.com/weblog

That’s one heck of an assertion. Certainly I don’t agree. Robots are good for specialized tasks, but have little flexibility. I would be curious of his definition of “robot.” I don’t consider a teleoperated device or a manipulator arm to be a “robot” because they are not independent. Still, I wouldn’t mind seeing the development of teleoperated devices for LEO that could be operated from the ground. We certainly don’t have them NOW.

And, kert, if you can show me a robot with two arms that can take ambiguous voice commands and follow them as well as a human, then I’ll agree with you. Robotics finally is starting to go somewhere because of the exponential improvement in computers, but it is EXTREMELY difficult to “teach” robots many things we take for granted, because we don’t know how we do them. Look at the recent DARPA race. Knowing some of the difficulties, I was impressed they got as far as they did, but it was called a failure in the news. And compared to what a general purpose robot would need to do, that was a very simple task.

You could make the same sort of arguements against tourism or spectator sports. There's really no need for people to go, in person, to exotic places, or gather in huge groups for no logical reason. You will see more and learn more from your television set than you will in person.

The reason to go to space is for the experience, and to not live one's life vicarously through the efforts of professionals. Witness also the popularity of various "extreme" sports-- people want thrills, and if their life is so otherwise comfortable that it doesn't provide them, they'll make the up.

Any space enthusiast who can't understand that needs to find another line of work.

Well, one thing a human can do in LEO that a robot cannot do in LEO is think. Granted, this is a skill that is undervalued in today's marketplace, but thinking is still important, and can be very helpful when trying to make decisions about the unexpected.

I always find it laughable when people make the argument that humans are inferior to machines. These are the arguments of couch potatoes, who have no appreciation for the difficulty or complexity of everyday physical tasks.

"And, kert, if you can show me a robot with two arms that can take ambiguous voice commands and follow them as well as a human, then I’ll agree with you. "
Visit Robodex this year. But note that its difficult to "agree with me", because all i was saying is that people often severely underestimate capabilities of modern robotics. Partly because of a certain western media bias.
BTW, tech that you can see on MER rovers or this years DARPA race is far from state of the art in robotics.
And one more thing, AI currently has very little to do with practical robotics.

"I always find it laughable when people make the argument that humans are inferior to machines."
That would be incredibly a stupid argument both ways. How can you answer a question like this: which is inferior, Britney Spears or Fiat Panda ?

Funny, my impression was that people, in the general public at least, tend to wildly overestimate what robots can do. They expect C3PO and are shocked to find that, for a robot, getting through a room without hitting something is pretty tricky stuff.

“BTW, tech that you can see on MER rovers or this years DARPA race is far from state of the art in robotics. And one more thing, AI currently has very little to do with practical robotics.”

Those are mutually contradictory statements. Or perhaps you have a very different definition of AI? Making sense of visual information and reacting to it properly are usually considered to be major areas in AI research. Probably the single biggest problem in development has been the fact that we don’t understand how we do things consciously, so we don’t know how to program for it. Fast hardware isn’t nearly enough, you also need to know how to use it.

I’ve been closely watching robotics develop since the ‘80s, and finally, with the increase in computer capability, they are starting to get somewhere. But I certainly haven’t yet seen anything that even begins to approach the “Apple II” (a general purpose, useful, not too expensive, reasonably programmable) robot. If you have seen one, please provide examples.

"Or perhaps you have a very different definition of AI?"
Probably. I take the 'I' part of AI as in real intelligence, conscious thinking. I think its completely unnecessary for practical useful robotics.
Take, bees for example. Would you say that bees are intelligent ? IMO, no, but they are still able to function very well and produce high value practically out of nothing ( well, natural resources and energy are ultimately consumed, of course )
The "AI must come first"-style of thinking wrt robotics is prevalent in western parts of world, so that commonly found laboratory robotic platforms in western universities consist of very pathethic sensory and mechanical appendages, while often employing frightening amounts of computational power.
Eastern robot developers have apparently grasped the fact that for the robot to be useful, it has to first have necessary means to interact with the environment, and computation is just a tool to govern these interactions.
Like i mentioned before, Pete Markiewicz has covered these subjects at length over his weblog.