Monday, July 21, 2014

Humanizing Technology

(Picture from here.)

I'm a big fan of technology. There are technological solutions to many of the world's issues from hunger to clean water. The problems we face are usually how to get people out of the way of solving the problem rather than the problem itself.

That said, I'm definitely not a fan of humanizing technology.

Let me be very clear. I am all in favor of making technology easy. I want getting food and clean water to be so cheap and comfortable that you'd have to spend money and effort not to get it. I want to be able to ask my computational partner, think I'd like that? And have it come back to me with a cogent answer.

But I don't want it to try to act like a human being. I'm not interested in the illusion of an experiential entity.

There's a strong pursuit of this illusion going on now. The most recent one to come across my desk is Jibo, a robotic device that is intended to integrate into a family. To act in the role of a person. (See here.)

Jibo was developed by MIT's Cynthia Breazeal with the expressed intent of becoming "part of the family." (See here.) She feels emotion is the "next wave of this humanized high-touch engagement with technology."

I'm not a fan.

I like a humanized robot voice better than a voice that sounds artificial mainly because it's more intelligible. And I really like some of the capabilities advertised for Jibo. It has strong facial recognition coupled with a good camera. So it can take pictures like nobody's business. Since it can identify users, it's a no-hands portal to the net and can be a real internet helper with reminders and such. My objection isn't what it can do but how it presents itself.

For example, in the Jibo advertisement the robot does things like read a story to a child. According to the article Jibo will "put a smile on your face and make you feel better."

It worries me that someone thinks consumer machine is necessary for that.

I have a dog. It greets me when I come home and puts a smile on my face and makes me feel better. Sometimes it makes a mess. It has to be taken out and walked. If I step on its tail by accident, it yelps in pain so I musn't do that. The dog gives me a gift and requires an obligation. This is a transaction between us. This is a relationship between two experiential entities.

But Jibo is not an experiential entity. It is a machine that is intended to counterfeit the trappings of an experiential entity. It is designed to present the attributes of a relationship without the realization of a relationship.

This bothers me.

I think it's important that we understand who we're talking to. Who we're in actual relationships with. I think, for example, we should be concerned about the welfare of our children, parents and friends. The welfare of celebrities doesn't remotely interest me. I think we should concern ourselves with factual news rather then pleasant fiction. The relationships I have are with my wife, son and dog. Not my toaster. Not my microwave oven. Not my nice talking internet portal.

Fundamentally, it's a distraction. I talked about the Turing Test a while back. Turing had the idea that a device that could imitate a human being could be as intelligent as a human being. Back then most of the literature conflated intelligence with the ability to experience things. After all, how could you be smart and not experience the world?

Now we know that it is quite possible. Intelligence and experience are very separable. We can have quite intelligent systems that have zero experiential nature and really dumb systems that fool people sufficiently to be thought intelligent. In fact, some of the truly creepy ideas in SF are systems that are very intelligent in their execution of tasks without such a nature. Think of Terminator.

Or, even better, remember the old Colossus or War Games movies. Now think of them again and realize there was nothing in those machines but elaborate rules engines. Nothing inside but LEDs and a hum.

Now, if there were a person (entity, experiential organism, alien) in Jibo, I'd be all over it.

Sunday, July 6, 2014

Getting into Space

(Picture from here.)

The fundamental problem with getting into space is conservation of momentum.

Specifically, we need propellant. A space ship throw propellant out the back. Momentum must be conserved so when a mass is cast out in one direction, the ship moves in the opposite direction. Momentum is often defined in terms of mass (m) times velocity, such that:

  • m(1)v(1) = m(2)v(2)

Where m(1) and v(1) are, in this case, the mass and velocity of the propellant and m(2)v(2) is the mass and velocity of the ship.

It doesn't take heavy algebra to see that if you have a low mass, high velocity propellant you can get a higher mass to move a little bit. This is, in fact, what all propellant based systems do. It's a way of getting energy translated into velocity.

But with rockets and other space ships you have to carry your propellant with you. Consequently, that mv equation becomes much more dynamic and difficult. The mass of the space ship is really the mass of the ship plus the mass of the propellant plus the mass of the machinery to store, process and inject that propellant with energy. The ratio of propellant to payload is often as much as 10 to 1. So to impart a significant velocity to a payload of one ton requires ten tons of propellant. This is the source of step rockets like the Saturn V or almost any other launch vehicle. As soon as you've used up a portion of propellant, you discard the mechanism (and parasitic mass) that delivered that propellant.

Launch vehicles use chemical reactions to impart velocity to the propellant's mass. The Saturn V used liquid oxygen and kerosene for the first stage and liquid oxygen and hydrogen for later stages. Some in-space systems use things like a Hall thruster, where an ionized gas such as argon is accelerated out the back end with electrical fields without a chemical reaction. The experimental NERVA rocket used an atomic pile to heat a propellant and release it out the aft. In all cases, the propellant had to be brought along and expended. Once the propellant was used up the space craft was unable to control itself. There are several potentially operational satellites in orbit that aren't usable because they have no attitude control: they've used up their propellant. 

It's fairly clear that a real movement of humans and materials requires we reduce this problem. 

There are a lot of ways to do it.

One mechanism is to remove the propulsion problem from the space craft entirely. This is the attraction of electromagnetic catapult or rail guns, solar sails, laser ignition-- all under the rubric of ground based or ground assisted propulsion. The idea is that you transfer energy and momentum from the ground to the space craft. You don't care how heavy it is on the ground. This frees the space craft to only carry what is necessary for its travel. It's hard to do this from earth: we have a deep gravity well and a thick atmosphere. A lot of easy mechanism that would work, say, on the moon, either can't overcome the gravity or can't push through the air. Which is, by the way, why I keep saying we should make a moon base our first priority. Forget Mars. Forget humans in deep space. Once you have the moon a lot of those things come for much, much cheaper. 


It's also the reason for the attractiveness of the space elevator. Ground based propulsion-- that is, the elevator's motors are powered from the ground-- and it gets you above the atmosphere. It's also true that the top of the elevator is high enough from the earth's surface that the gravity well is considerably reduced. 

But we're still stuck with the propellant problem. Our mythical deep space vehicle needs to dump something out the back end in order to move the front end.

Why can't we be like a submarine? It doesn't take water with it. Instead, it imparts momentum to the water in which it floats and thereby gets a return on momentum for itself. Why can't we do that?

This is the basis of Harold White's work at the advanced propulsion lab at NASA.

White's gotten a lot of press recently on his work on the Alcubierre drive concept-- and I'll talk about it, too, since that might someday get us out of the solar system. But his work on the "Quantum vacuum plasma thruster" (Q thruster) is much more exciting.

Remember I mentioned the Hall thruster? It's well documented and in use. The Dawn mission uses it. It's a low thrust, high efficiency system nicely designed for long term missions that require a little thrust over a long period of time. Not suitable for human spaceflight but good for many things.

If you look deep into vacuum the space is anything but empty. Virtual particles whip into and out of existence all the time. One would expect that these virtual particles-- being virtual, after all-- would have no actual effect on the real world except for the Casimir Effect. Turns out if you take two plates and put them extremely close together-- closer than a given wavelength of light. This precludes vacuum virtual particles from coming into existence of that wavelength or less. Outside the plates those particles can come into existence. This gives a density difference that can be measured as a pressure.

White's Q thruster operates on the same essential principle of plasma handling-- similar to how a Hall thruster operates-- and adds velocity to the virtual particles to give them momentum out the aft end, giving a net thrust. It is not a reactionless (or inertialess) drive. Momentum is conserved in that there is a wake of virtual particles. (Though where that momentum goes when the virtual particles disappear is a good question. One I haven't seen answered.)

He's gotten some interesting experimental results. (See here.) He's been working with the idea of getting between .4 to 4 Newtons/kW. As a thought experiment, he suggested a space mission with a 50 metric ton payload to Jupiter in 35 days-- about the same time it took Columbus to sail from the Canary Islands to the New World. The same mission to Saturn would take 70 days. 

Similarly, Earth's gravitation field is about 9.8 Newtons. Reduce the payload by two thirds but keep the same power source: you have a thrust of 12 Newtons and an effective lift off acceleration of 2 Newtons: 1 kg accelerating at 2 m/s**2. Chug your way into space. 

So this is very exciting. This has some proof in the lab-- White has a test harness that shows promise. But, if it pans out, it is an engineering exercise to put it into practice. 2 MW is nothing-- the solar cells on my house can do that in a few days. A tiny nuclear power plant can do that every second for years.

That's the one I'm hoping for out of White's lab. It is not (unfortunately, to me) what is getting all the publicity. That's the warp drive.

Relativity pretty much seals the deal on most ideas of faster than light travel. The speed of light in a vacuum is the iron clad speed limit of anything travelling in space. But, as the saying goes, you don't get down off an elephant. You get down off a duck. While the speed limit of travel in space is the speed of light, that says nothing of the speed of space itself. 

In fact, current theoretical models of the big bang incorporate inflation, where the early universe expanded many times the speed of light. While nothing in the expanding space, light or otherwise, could have exceeded the speed of light, the relative distance between points of space was increasing at multiples of c. This gives an effective travel speed greater than the speed of light.

Miguel Alcubierre was the first physicist see the possibility of FTL travel by altering space. In 1994, he posited that if one could warp space such that it was expanded at one end and contracted at the other, the net result would be a bubble that would transit normal space at multiples of c. This has since been dubbed the "Alcubierre Drive." However, it required exotic matter, negative mass and an energy requirement that was staggering: something like the mass of the planet Jupiter converted completely into energy.

In 2011, in an exercise for the 100 Year Star Ship Conference, White looked at Acubierre's equations. He realized that they were sensitive to the geometry of the system. (Paper here.) By altering the geometry of the warp bubble he was able to drastically reduce the energy requirements from a mass the size of Jupiter to something like the mass of a Buick. This is still not a small amount of energy. The numbers being tossed are more than the energy released in the largest nuclear weapon ever built: Tsar Bomba. And it would require actually containing and using such energy. 

But by scaling down the energy requirements to something imaginable, he also created a model that could, on a small scale, be testable in the lab. Currently, he, and other labs, have been attempting to determine if they can create a very small warp bubble. To determine this they are trying to change the nature of a laser interference pattern in a predictable manner. They've had some success but nothing sufficiently definitive. White has been willing to talk about his results in symposia but so far no paper has come out yet.

So we have two extremely interesting ideas that are right now being tested in the lab. The Q thrusters-- which, if they work, will pretty much open up the solar system to us-- and the warp drive that could give us the stars.

I won't live to see the warp drive in operation though I might live to see it disproved-- that would be a sad day indeed. But the Q thrusters. That could get us anywhere. That's the important research.

Further reading: