Burke nodded. He could see problems with the approach, but not as big as the system being hacked. And it wasn’t his field. Decker was the expert here.
“You will be able to communicate with it, though?” Burke asked.
“Oh, sure. And it can communicate with us. Like if it needs something it can’t find, and wants us to send it something. Some raw material or such.”
“All right. Well, there are other people we’re going to need to get in touch with on this, Bernd. To see if it’s even possible on the factory side. But it is possible on the computer side?”
“Oh, sure. It’ll take time, but time we have.”
Burke nodded again. Maybe there actually was a solution to his big worry. It was certainly worth some effort to find out.
“I’ll make some calls. Can you hang around for a while? A few days? Help me out with some conference calls? I can use help selling the computer side of this.”
“Sure. I wasn’t sure if my proposal would flop and I would be out of here in an hour, or if we would need more time. I took the liberty of bringing along a small bag, but I also hung on to the autodrone. Let me grab my bag and dismiss the autodrone.”
“All right. Then I can show you the guest room and we can get some lunch. I want to talk about this plan in more detail. And we have to decide how much of the real plan we tell people.”
“Even our partners?”
“Especially our partners.”
When he got home three days later, Decker’s wife Anna Glenn was interested in how it went.
“Did you sell him on the plan?” she asked.
“Oh, yes.”
“Now what happens?”
“Now I have to figure out how to do it.”
“The whole thing?”
“No. The computer part of it.”
“You don’t know?”
“No, but I’ll figure it out.”
Metafactories
“You want what now?” Greg Hampton asked.
“A factory that can make more of itself, deployed in space,” Russ Porter said.
The vice president of engineering at Colorado Manufacturing Corporation didn’t look satisfied.
“That’s what I thought you said,” Hampton replied. “But what’s the larger goal?”
“OK, think about it a second,” said Porter, the CEO of CMC. “You have this factory, right? It’s a dark factory, and it’s space deployable. You send it out to the Asteroid Belt, and the first thing it does is make another one of itself. Then it starts making products for shipment to Earth. No gravity well to fight in the Asteroid Belt, it just accelerates the products a bit along an intersecting orbit with Earth. They could take a couple years to get here, because it doesn’t matter.
“What does the second factory do? It moves on to a different asteroid and makes another one of itself. Then it starts making products for shipment to Earth. And so on. We make one factory, and the output keeps increasing over time with no further investment. In the Asteroid Belt, there’s no limit of raw materials, and we don’t have to build anything but the first factory. The rest is free.”
“That’s sounding a bit like a species,” said Kay Brady, CMC’s chief scientist.
“How so, Kay?” Hampton asked.
“The definition of a biological species is that members of the species can mate and produce offspring, and that those offspring are themselves fertile with other members of the species. This is different because the factories can reproduce themselves, but the offspring can also reproduce themselves. They’re fertile, in that sense. Not the same, but similar.”
“Ah. I see,” Hampton said.
Hampton turned to Porter.
“It’s a brilliant concept, if we can pull it off. Who are our partners in this?”
“Ted Burke and Bernd Decker.”
“That’s some big horsepower,” Hampton said.
“The biggest. And if Ted Burke has some long-range visionary idea, we want to be in on it. People thought his father was crazy for wanting to build on the Moon, but look where that went.”
Hampton nodded.
“The factory is going to have to have some sort of robotic or remotely controlled assembly units to build the new factory outside itself.”
“Then the new factory has to have the same thing, so it can build its own child factories,” Porter said.
“Unless we do something like human pregnancy,” Brady said. “Have the factory build the core of the new factory within itself, then eject the core and let the new factory finish itself. That would simplify the initial deployment as well. We would only have to get the core of the first factory to the Asteroid Belt and let it finish itself once it got there. Simplifies the transport problem and reduces the energy expenditure.”
“That’s brilliant,” Hampton said.
“It is if we can do it,” Brady said. “I’m not sure it’s doable.”
“I can see that you two are already on this, so I’ll leave it with you then,” Porter said.
“One last question, Russ,” Hampton said. “What’s the timeframe on this?”
“Completely open.”
Hampton raised his eyebrows at that, and Porter continued.
“This is a Burke project, Greg. Extremely long-term. With all the unit testing and building it in orbit and everything, it will probably take twenty years to pull it off. Let’s make sure we get it right.”
“And do it along with everything already on our schedule.”
“Yes, of course. But you can hire and move people around to staff this, because if Ted Burke is throwing a party, I want to be there.”
“So what do you recommend as a strategy?” Hampton asked Brady after Porter dropped out of the conference.
“Pull the very best ideas people out of every project, promote their seconds, and then replace them two-for-one with new hires. Maybe three-for-one.”
“Really.”
“Yes,” Brady said. “Figuring this out up front is the big issue. One big idea can change everything. Big-ideas people get bored in mature projects anyway.”
“Who do we make the technical lead?”
“Peter Moore. He’s the best ideas man we have. And he’s open to good ideas from anywhere, including being receptive to other people’s ideas contrary to his own.”
“Moore?” Hampton asked. “But he’s got issues driving a project to completion. Who do we make the project manager?”
“Valerie Dempsey.”
“Dempsey? Have you ever heard Dempsey and Moore go at it? They argue like an old married couple.”
“Yes, like an old married couple that’s still together,” Brady said. “That’s the point.”
“So you think that could work?”
“Yes. If they don’t kill each other. Which would actually solve the problem.”
“All right,” Hampton said with a sigh. “We’ll give it a try.”
“Holy, crap, Bob. Did you hear that?” Tim Fender asked.
“Yeah. Everybody on the floor heard it,” Robert Abrams said.
“How are we going to do this project with Val and Peter after each other like that?”
“Oh, it’ll work out.”
“You think so?” Fender asked.
“Yeah. You know where they are now?”
“No.”
“They went to lunch together,” Abrams said.
“After that?”
“Yeah. It’ll be fine. It’s just how they are.”
“It’s too bad Kay Brady’s pregnancy idea won’t work,” Valerie Dempsey told Peter Moore over lunch. “That would really simplify the construction and transport of the first factory.”
“Yeah, but the other things the factories need to make won’t be able to finish themselves,” Moore said. “We need the factories to have an exogenous manufacturing capability to make things that big.”
“Yes, they’re talking about spaceships for the Earth-Moon run and all manner of other things. Which means all the assembly ro
bots and welders and all that. Which also means the capability to manufacture all the robots and welders and all that. Which makes the whole job much tougher.”
Moore nodded. It was Moore’s punching holes in the pregnancy idea that had led to the shouting match this morning. He understood Dempsey’s position, but it just wouldn’t work.
Dempsey sighed.
“It would have been so beautiful. We ship out a small factory that finishes itself once in place, then starts making new factories. With a fraction of the lift to orbit and much easier transport to the Asteroid Belt for the reduced mass.”
“Wait. Wait just a minute.”
Moore held up a hand and Dempsey waited. She had seen him do this before. He was on to something, but it was clear he wasn’t sure yet just what it was.
Finally Moore slapped his hand on the table.
“We can still do it,” he said.
“Do what?”
“Ship out a small factory that finishes itself once in place, then starts making new factories.”
“But I thought that’s what wouldn’t work,” Dempsey said.
“Not the way we were thinking about it. We artificially constrained the problem.”
“In what way?”
“That there can only be one type of factory,” Moore said.
“Doesn’t that make the problem simpler?”
“No. Or at least not by much.”
“So...?” Dempsey encouraged.
“What we do is we make a mother factory. An infant one. We transport it out there, and it finishes itself. It has no exogenous construction capability. But what it can do is hatch new factories that do. But there’s no reason that first factory needs any capability beyond making new factories.”
“Which makes it much easier to build and transport.”
“Exactly,” Moore said.
“Doesn’t that make it twice the problem, though? Designing two different types of factory?”
“Not precisely. I think what we do is design the more capable factory, then back into the mother factory, leaving out the things we don’t need. So it’s more of a design problem, but it’s not close to twice.”
“So we start on the big factory design,” Dempsey said. “That doesn’t change what we decided this morning. But once we have that, we start seeing how small a piece we can make for the initial factory to minimize lift and transport costs.”
“And if there need to be some slight mods to the big factory plan to make the initial mother factory as small as we can, we back that into the big factory design at the end.”
“I like it. I like it a lot. I think we get the best of everything out of that setup. Nice.”
“Thanks.”
Tim Fender and Robert Abrams were working up a list of basic capabilities required of the factory platforms and marking the ones that weren’t in CMC’s current portfolio when Moore and Dempsey came back from lunch. They were all smiles, talking and laughing as they walked by.
“See,” Abrams said. “Told ya.”
“I’ll be damned,” Fender said.
“OK, so back to the power unit. We have nuclear, a closed system so there is no ongoing mass requirement, and we have to radiate away the extra heat somehow. What else?”
“We need to get North American Power involved in that. Didn’t you say they were on-board?”
“Yes, but we can get some of the parameters down, I think,” Abrams said.
“I think we need to worry about the case of a factory working an asteroid without fissionable materials. We don’t want to have to limit our source material to the ones that do.”
“Agreed.”
“So I think we need some kind of material transfer system between the factories,” Fender said.
“Long ways between asteroids in the Belt. Average is, I think, over half a million miles.”
“Yes, but in the thickest part of the Belt it’s a fraction of that. Even so, time-in-transit isn’t an issue because there’s no humans aboard. No environmental support issues.”
“We also don’t want that to be some sort of system that requires reaction mass,” Abrams said.
“Can we use an electromagnetic impeller?”
“Shoot containers back and forth. Probably. How do we brake them?”
“Maybe we just catch them,” Fender said.
“Catch them?”
“Yeah. Big net on springs and shock absorbers. Have a catcher’s mitt stick out from the factory and catch them.”
“That might work, depending on velocity,” Abrams said. “I’ll have to play with the numbers.”
“If we can make it work, it solves two problems. The two things we’re probably most short of. Fissionables and water. One factory runs into a bunch, it sends the extra to others.”
“Sounds good, if the numbers work. Oops. There goes the message alert.”
Abrams fiddled with his display.
“Huh. Get this,” he said. “From Dempsey. ‘First factory on pregnancy model. Subsequent factories with exogenous manufacturing.’”
“Wonder what Moore will say to that?”
“He’s copied on it.”
“No shit,” Fender said. “OK, so now we know what happened during lunch.”
Fender thought through the implications. Abrams saw the distant look in his eyes and let Fender think while he acknowledged receipt of Dempsey’s message. Fender finally shook himself and his eyes refocused.
“Thinking about it, that solves a lot of problems on both ends,” he said. “Easier lift, easier transport, but no limit on manufactured size on the other end.”
“Yes, and it gives us the big manufacturing bay of the initial factory as an enclosed volume we can use to send supplies along. So what do we want to send along?”
“Fissionables and water, I think. Maybe copper. Let me work on it.”
Artificial Intelligence, Sort Of
Bernd Decker was dealing with his own set of problems in planning the Belt Factory Project. Mostly because it was impossible.
“Ok, Bernd,” John Butler said. “You want a machine that is next-level artificial intelligence, right? Much more capable than JANICE. Next-generation more capable.”
“Right.”
The Joint Artificial Neural Intelligence Computation Engine – JANICE – was the most advanced machine ever made. It was not the fastest at basic computation, but it was the most independent in crafting solutions to problems.
“And we have a decade or so to do it,” Decker added.
“It doesn’t matter. We couldn’t even use JANICE for this.”
“Why not?”
“Because we can’t replicate the fab facility in space,” Butler said. “If every factory is going to have one of these machines on it, and the ability to make more such machines, we simply can’t do it. You want a space-based fab facility, for next generation parts, and you need to be able to build that fab facility in a space-based factory under computer control. Ain’t happening. Not in ten years. Not in twenty. Not even to current Earthbound capabilities.”
“Ah.”
“You need to figure out a way to do it from here.”
Decker’s mind raced. Could they build the machines here and ship them out to the Asteroid Belt for inclusion in new factories? Or build the fab facilities here, and ship them out? Shipping the machines would be easier than shipping the fab facilities. Either would make the ongoing effort dependent on support from Earth, though, and the project was likely to take more than a lifetime. Which made it dependent on other people carrying on the project.
Wait a minute. How many machines did they need, anyway? Did every factory need one? Was he actually creating a problem there?
The more advanced machines weren’t strictly deterministic. The closer you got to something like human intelligence, the less deterministic they were. Like human geniuses in some field, they would get you a good answer for a problem, but two of them would not necessarily come up with the same good answer to the problem.
Was he effectively sending a committee into space?
Probably one machine would be better, as long as it kept working. What were the odds of that? A corollary of Murphy’s Law said the more important and unique a component was, the more likely it was to fail. Include redundant units and neither would ever fail; include just one of some critical component and it would always fail, and at the worst time.
So what if they made three of the high-end machines here, and sent them out with the first factory? One could be installed in the first factory, with the other two as cargo. When the first factory built the second and third, it could install those machines. The later factories would get less capable machines.
Wait.
“Can we space-build a fab facility that could make lesser machines?” Decker asked.
“Run-of-the-mill stuff? Like current factory computers, say? Good enough to run a factory? Sure.”
“Then we could send out three of the advanced machines. Build them here. Put one in the first factory, and have it take the other two along as cargo. It puts them in the first two child-factories. Probably not even start them up unless the first one fails somehow.”
“That we can do,” Butler said.
“We need to worry about the comm links between the advanced machine and the slave machines, then. Something that can’t be hacked.”
“Simplest way to do that is to make them directional. Directional RF antennas, or maybe modulated lasers. The only links that count are the ones that don’t intersect Earth’s orbit, which is a pretty small part of the sky as seen from out there.”
“All right. Let’s plan on that, then,” Decker said.
That decided, Decker bent himself to considering architecture. The thing humans did that computers didn’t do was make intuitive leaps. These came in two forms. One was coming up with an answer from partial data, like seeing the punch line coming halfway through a joke. The other was an answer to a problem just popping into one’s head, and knowing it was right. How did you do that with a computer?
Of course, answers didn’t just pop into one’s head. They popped into one’s consciousness. Decker thought of the human brain as a hundred computers, only one of which the user had a terminal into. That was one’s conscious thought process. The others continuously worked problems in the background. That’s why, when you were beating your head against some problem, you could sometimes set it aside and come back to it later, and the answer would be obvious. Those other computers had worked it out while you were doing something else.
QUANT (COLONY Book 1) Page 2
