De Romanicorum Physicalium 10

Pensées sur l'SF.

• For those who are sanguine about the Pre-Tribulation Rapture Singularity, here's a little reminder of things Kurzweil thought would happen by 2010 (from a talk presumably somewhat before then):

  1. Images written directly to our retinas.
  2. Ubiquitous high-bandwidth connections to the Internet at all times.
  3. Electronics so tiny it's embedded in the environment, our clothing, our eyeglasses.
  4. Full-immersion visual-auditory virtual reality.
  5. Augmented real reality.
  6. Interaction with virtual personalities as a primary interface.
  7. Effective language technologies.

  #1 ain't happened. #2 ain't happened. #3 ain't happened. #4 ain't happened. #5 ain't happened. #6 is vague but if it means AI snerk, and if it means "virtual avatars" snerk, guffaw—it ain't happened, the only question is how ridiculous it was to think it would. And #7 ain't happened.
  
  That's a track-record that shoots right past Harold Camping and into William Miller/Samuel S. Snow country.
• I think that zled skulls, while they have the flare at the back of the mandible, like a hippo, to accommodate a swell of jaw muscle, still have the wide-looped zygomatic arch, like on a felid's skull, because even though their jaw-muscles no longer attach on the top of their head, they still go most of the way up the sides. This translates to big "cheekbones".
  
  Also, instead of having that gap on the upper, outside edge of their eye-sockets, seen in most of the Carnivora and, indeed, in many other Laurasiatheria—bats, peccaries, pigs, rhinos, tapirs—their eye-sockets look like those of a tamarin attached to the zygomatic arch of a jaguar. Except, zled eyes (unlike mammal eyes but like bird ones) are immobile...and they have the sclerotic ring that supports that (they also have very shallow eye-sockets compared to mammals, which, one, gives more room to jaw muscles but two, and more importantly, lets them have more of the image in focus without having to move their eyes—a bird's eyeball is shaped like an M&M, a flattened ellipsoid rather than a spheroid).
  
  Reading up on tamarins suggests zledo are, in many ways, basically a carnivorous version of them; tamarins, after all, are mostly monogamous (I think a large minority of them are polyandrous), live in nuclear-family groups, and have anatomy that can be described as halfway between cat and (other) monkey—they have claws as well as thumbs, for example. They even have jaw-shapes designed to accommodate big muscles.
• People are awfully optimistic about using graphene (2-d carbon hex-grids, like flat buckminsterfullerene—they sometimes make it by literally cutting nanotubes down the side and unrolling them) to store hydrogen. The trouble with storing hydrogen is its tendency to blithely outgas right through the very walls of the tanks; it's smaller than the gaps between the atoms most tanks are made of. One proposal is to bond the hydrogen directly to the graphene to make graphane; you can get it out again by heating it. Another is to have the graphene fold itself into "origami boxes" around the hydrogen; you get it out with an electric charge. But the origami-boxes are apparently only 9.7% hydrogen by weight. The rates I find for graphane with the hydrogen directly bonded vary by the alkali metal used as (I think) a catalyst for the bonding, but the rates listed are 12.20%, 10.33%, and 8.56%, for lithium, sodium, and potassium respectively. Meanwhile, boring old methanol gives a by-weight efficiency of 12.6%, and it's almost certainly cheaper even if you use a ruthenium-catalyzed process to get the hydrogen out. Water is 11.19%. Ammonia is 17.75%, and apparently you can separate it out with sodium amide, which costs very little; that, I think, is the real wave-of-the-future for hydrogen-powered vehicles.
  
  (Gasoline has exactly twice the energy density of ammonia, by volume, and six times by mass according to the only source for that second number I can find—ammonia isn't very dense—but fuel-cell vehicles are apparently 75% efficient while internal combustion engines are only 20% efficient, frittering away most of their energy as waste heat. Apparently you actually need 69% more ammonia to fuel a fuel-cell car than you do gasoline to fuel an internal-combustion one...but ammonia is literally 20 times cheaper. You'd probably close the gap, price-wise, with refrigeration: pure ammonia actually boils before room temperature, so you'd need something to keep it cold. A full-size Ford sedan only got 14.4 miles to the gallon in 1952, whereas one now gets 23, meaning you needed 60% more gas—and 1952 was the middle of the Golden Age of Route 66, so I don't think the fuel-economy will pose much of an issue.)
  
  I'm not sure what the solution is for space-travel, where every gram of wasted mass counts (every kilo of hydrogen, even if you store it as ammonia, comes with 5.63 kilos of "waste" nitrogen). Maybe just budget around outgassing a portion of your propellant. The current state of the art in storing (presumably gaseous) hydrogen, "quantum passivation", where you electropolish a stainless-steel pressure vessel to a mirror finish and then fire it in a furnace between 673 and 803 Kelvin, has an outgassing rate of 2×10^-15 "torr-liters per second" (i.e. 2×10^-15 liters per second at a pressure of 1 torr, which is 1/760 of a standard atmosphere). At that pressure, 1 liter of hydrogen masses 0.08988 grams, which means an outgassing rate of 1.7976×10^-16 grams per second. Stainless steel hydrogen tanks routinely have pressures of 20 megapascals (150,012.337 times as big), which would give an outgassing rate of 2.696621810×10^-11 grams per second, or .85 milligrams per Julian year, so I'm pretty sure you can take it. I think when you store it in liquid or "slush" form its outgassing-rate is reduced, though—my guess being you divide the gaseous-outgassing rate by what percentage of the mass of your liquid or slush hydrogen is actually bubbles of gaseous hydrogen. (Presumably for aerospace applications you're not going to use stainless-steel vessels, but then, for a spacefaring civilization you can probably get the same performance with other materials—if you google "quantum passivation" most of your results pertain to semiconductor research.)
• Slush hydrogen, incidentally, is 16-20% denser than liquid hydrogen, meaning that every megagram of it takes up 11.29 to 11.86 cubic meters, instead of 14.11 cubic meters for liquid hydrogen (liquid hydrogen's density is 70.85 kg/m³, so slush's density is 84.35-88.56). Spacecraft propellant tanks are spheres (because they're the ideal shape for pressure-vessels); the slush-hydrogen ones require tanks 2.78 to 2.83 meters in diameter for a megagram of propellant, while those of liquid hydrogen need tanks 3.00 meters in diameter.
  
  Suppose you go with a rocket with dimensions comparable to, say, the SASSTO—but sporting a proton-chain rocket that gives an exhaust velocity of 10% c. (I find single-stage-to-orbit designs a good model for high-end "starship" type ships, I think I've mentioned that before.) If you go with the SASSTO's 10.34 mass ratio—remember, it's not the weight with propellant (97,976 kg) over the dry weight (6,668 kg), it's over the dry weight plus the payload (2,812 kg, which brings the total to 9,480 kg)—and a proton-chain rocket, you wind up with a speed you have to circle the block a few times to come down from. 7.5% c seems to be about the cutoff point to brake in a timely manner, for which you want a mass ratio of 4.5; that would bring the "dry" weight (if the payload is unchanged) up to 18,960 and 4/9 kg. It's up to you how you want to interpret the change; me, I interpret it as the ship getting bigger. Take the cube-root of the increase in mass, and you wind up with a SASSTO 26.63 meters long and 9.35 meters in diameter. (Incidentally, if you look at the SASSTO, I think the closest approximation to its volume—if you don't want to screw around with figuring out how to take the volume of each conic section of the tapering parts—is an ellipsoid.)
  
  Anyway. A 4.5 mass ratio means your interplanetary SASSTO carries 76,203 and 5/9 kg of propellant. Slush-hydrogen, max density, would require 860.45 cubic meters of tank to hold it all; supposing you cart that around in seven rings of seven balls each (the most efficient way of storing spheres is to stick them in hexagons, with one at each point and one in the middle), you get 49 spheres each 3.22 meters in diameter. A quick "eyeballing it" diagram in Inkscape suggests they'd come to a total of 19.1 meters long, if you nested them front-to-back as well as side to side (offset each ring by 90° and you can nest each sphere into the space between the spheres behind it); if not, of course, you're just getting (3.22×7=)22.54 meters. You probably have to make your ship longer to accommodate that; to keep the mass, remember to divide the diameter by the square root of whatever factor you increase the length by. Make it twice as long, for example, and you have to make it narrowed by a factor of √2.
• Looks like DARPA has Jossed me. Their Warrior Web suit is worn not as chaps, but under the clothes like long underwear (sure, "wetsuit", that's what it's like). The Wikipedia article on DARPA (the Warrior Web project doesn't have its own entry) refers to it as an "exosuit". So I guess that's what I should call the equivalents in my books. Mark your calendars: on 5 May 2014, DARPA said they planned to actually equip a squad with them to compete against another squad in various carrying and mobility tasks—"30 months from today". Thirty months from then is 5 November, 2016.
  
  Also, things need 100 watts. An average laptop battery provides 72 kWh (or 259.2 MJ, if you prefer SI units like I do). That means it can power a Warrior Web suit for 720 hours, which as you may have noticed is thirty freaking days! I'm sure an "average laptop battery" is not what you want to be depending on in the middle of nowhere while people are shooting at you, but the point is, powering that for an extended period of time is not exactly Arc Reactor business for our current capabilities.
• A four-minute mile—something else the suit's supposed to enable—doesn't sound all that impressive; it's the standard for all male middle-distance runners, although it was a standard that was breaking a record in 1954. But...middle-distance runners don't compete with a backpack and a rifle, wearing body-armor. A guy wearing an exosuit would take 23 seconds to close from the (arguable) effective range of the M4 carbine (it's actually effective to about 400 yards, but at 200 yards its performance starts dropping off rapidly, because of its shortened barrel), to its "battle zero" range (which seems to be 25 yards, in most sources I can find). He also needs 32 seconds to go from the effective range of the M1 Garand to its battle zero (440 yards to 200 yards—yes, the Garand's "zero" was the range at which the M4 starts to suck). That second number is important to me, remember, because my Peacekeepers' round is based on the .30-06.
  
  Some looking around on things that go that fast (other than middle-distance runners) revealed to me that I was wrong, zledo don't go as fast as bikes, when they run. They go faster. Well, not faster than racing bikes; those go 40 kilometers per hour, which I think I'd set as the specific speed of a running zled (a four-minute mile is 24 kph). But the average bike only goes 15.5 kilometers per hour. Zledo weigh the same as an ostrich, and while their legs aren't built like ostriches' (they're built more like a carnosaur's), the carnosaurs could put on impressive bursts of speed, with Allosaurus doing something in the 30-55 kph range. Zledo aren't distance-runners by any means (which is why they domesticated the zdhyedhõ'o, cursorial predators like giant dogs), but for trips of a couple of blocks they can do on foot what we'd need bikes to do. They can also, in a fight, jump 8 meters forward (closer to 10 with a running start) or 3 meters straight up. They come from a planet that's somewhere between the Mesozoic and the Pleistocene on the "murder world" scale, and it also has 8% higher gravity.
• Atlas, the bipedal robot from the folks who brought you the BigDog (Boston Dynamics, a robotics company owned by Google but with, y' know, concrete accomplishments instead of eschatological bafflegab), has recently been equipped with a battery, instead of an off-board power tether. This battery is 3.7 kilowatt-hours, and powers the robot for an hour—so I guess we can conclude its typical operations require 3.7 kilowatts. (Convenient ratio, that.) A 3.7 kWh Li-ion battery would, apparently, weigh between 37 and 13.962 kilos (because they run from 100 Wh/kg to 265 Wh/kg). Atlas' mass is 150 kilos, so its power plant takes up 24.667% to 9.308% of its mass (that second number is very nice).
  
  Now, Atlas is not really that much like a really humanlike robot (I'll get to those in a second). It is, however (not least because it was made for DARPA) an excellent model for a walking mecha. Scale a 1.8 meter, 150 kilo Atlas up to 10 meters tall, and you get a mecha that weighs 25,720 kilos; assuming that power usage scales with mass, it uses 634.43 kilowatts to operate. We scaled its power-plant along with the rest of it, so its 2,394-6,344 kilo battery can still provide it with an hour of operation. You probably want it to operate for longer than an hour, though, so I'd go with lithium-air batteries for the mecha. Those (11.14 kWh/kg) give you, on the small end, 42 hours of operation, and on the large end, 111. Or go with a smaller battery, shorter operating times (hey, a full day of operation would only require 1,366.8 kilos) and use the extra weight for armor and weapons.
  
  Maybe I'll go with lithium-air batteries after all, for my mecha; methanol is less than half as good a power-source, only 5.472 kWh/kg compared to 11.14.
• But for humanoid robots that you might actually mistake for human in good light and clothing other than a Mother Hubbard dress, the model has to be Kenshiro. (So, I guess Atlas, as a model for robots, is already dead.) Kenshiro is a Japanese robot that is actually scaled like a human, albeit a not-full-grown one—at 158 centimeters and 50 kilos, it's the size of a Japanese twelve-year-old boy (and about 4 kilos lighter, but roughly the same height, as an American thirteen-year-old girl). If it were as tall as Atlas, it'd only weigh 74 kilos.
  
  I can't find its power-usage stats; since its structure is monumentally more complex than Atlas it probably, at least for now, uses much more power. Eventually a day will come when they get it down to the energy use of a human being, which is 115 watts for a man the size of the scaled-up Kenshiro (and 105.5 watts for the original-sized version). Yeah, you can figure out how much power it takes to be you by converting your calorie intake to kilowatt-hours (hint, one kWh=3600 Joules) and then dividing by 24. (I just do it with Google's calculator.)
  
  Of course, a robot with performance superior to a human is probably going to mass more, and will certainly require a bigger power-pack to justify its superior power. You burn energy moving things, remember. Significantly superior strength would also require reinforced joints, which adds more weight and power-pack. This is the thing: while a robot could probably be given superior performance to a human, you can't realistically give it drastic upgrades and still let it ride around in our cars or pass for human for one second after it accidentally pulverizes someone's foot on a crowded sidewalk—and, again, power requirements. Tôka kôkan is the first law of a lot more fields than just alchemy.