| Next | Previous | Contents |
Tiger Leones here, in my lab on the second floor of the Whinx design and manufacturing complex in Salt Lake. This October I'm supposed to be getting accolades and stock options from my boss. I hope the package Alan left on my work table is better than last Thursday's. That dork, once I dug out of him the fact that he was overclocking my chip at 1.35 gigahertz... We told them in the data sheet that it has heat dissipation problems. It was kind of funny; I had a little cold that day and he asked if I could speak clearer and I wanted to give him a hard time so I told him my fangs kept my tongue from moving like his did, and he practically dropped his load when he realized he was talking to the chief designer and not some tech trainee. I shouldn't lie like that; I can talk as clear as any human and better than most bother to. Well, none of that is going to fix the W32 chip, if indeed there was anything wrong in the first place.
Hmm, this one is from Millstone again; they sent in a nothing-to-fix two weeks or so ago, but on the phone their EE swore that it flaked out for them. I slit the tape with the builtin package opener on my index finger. Three roaches here with their legs in the air, a disc, and a note, from the same woman, saying that ten seconds of their code fragment will do to a W32 what illegal parathion will do to a roach. Well, let's use the test jig to see if it's a nothing-to-fix situation again. No, they're dead, but not like a parathionized roach: they're doing something, all three the same, but like lobotomized. They can load and store, but anything that goes through the ALU comes out zero. Interesting. I brush a patch of midnight black fur into alignment on my muscular, allspice-scented shoulder.
So let's see this code fragment. Kind of short; they did some work to track it down and expended some expensive chips. I'll put a quick diagnostic after each repetition: two each add, multiply and divide. And how about a comparison just for completeness? This business about ten seconds puzzles me; it should either run forever or blow the first time, not that you'd expect that in a Tiger Leones design. I look at Kandinsky's cat in the print over my desk, but she offers me no clues. If I were male I suppose it would be a tomcat, but there would still be no clues for me.
Me: Hey, Alan, you want to look at something here? Bring a W32; I blew up my last one yesterday afternoon. This is all there is to their program. Can you see anything nasty there?
Alan: What are they complaining about? That blows up chips? This I have to see.
Me: They say it has to run for ten seconds, then is instantly fatal. Let's see what the reality is. I'm recording all the output, with time stamps.
The first error is at about 8.5 seconds, a bad addition. 200 milliseconds later it misses a compare. Things go downhill from there but oddly the multiplies and divides are perfect up to the very end, when the bits from the ALU start winking out one at a time in kind of a random order. I've never seen anything like it in 28 years of quantum logic work.
Alan: Crap, what's going on there?
Me: Jeez, I have no frigging idea! I feel like... Well, goals, issues and action. The goal is to maximize Whinx's profit by getting a great chip out the door, not one that has bits peel off one by one. I feel awful about this! Issues. When we find the problem we're going to send out new chips to everyone who's taken delivery, so we'd better freeze shipments in their tracks. We have to notify the public. Frank has to notify but I have to write the press release for him. There have been all those rumors about flaky behavior, and I'll bet... Either Millstone will want to be mentioned by name in the press release, or they won't. Now our real work is to find the bastard and kill it. Sorry about the claws but I take this personally. Would you do two things: put a note in your file to send Millstone four good chips when we have them, and then order a case of W32's and try to find a minimal set of that code fragment. And see if you can minimize chip wastage by stopping the test on the first or second error, letting it cool off; you know, see if we can detect partial failures. And I'm sure we're going to have to look for a thermal cause. You know what to check out from instrumentation, and ask for quartz lid chips in addition to the case you're ordering. While you're doing that I'm going to get raked over the coals by Frank.
Alan: OK, I'm on it.
I punctuate my conversation with lashes of my white-tipped tail. On the computer I start up a NetBoard session and invite Frank Encke, the chief of development for Whinx, number two person, and my boss. He answers promptly and I start typing.
Me: Frank, we have a problem. This customer sent in three dead chips and the piece of code that killed them. I just blew up another one. I really have a bad feeling that the rumors about flaky behavior are right, and that this code fragment is the key. Yes, I agree, you'll tell the shipping department? Yes, I already have an outline for the press release in my head. Moi? Sure, I'm responsible for the mess and I'll stand behind it, but I thought... Yes, I agree, negative publicity is better than no publicity at all. When do you think, 13:00? Can the reporters get here by then? OK, I'll have it up to you in about half an hour and we'll go over it. Bye, and I sure wish the announcement were different. Right, bye.
My mate Simba works at home. He's the editor of the Lion Disc. When we were built the Lion Foundation put equal effort into training us, and we sell discs of our lessons to the humans at cost. Simba's current project is to split up the physics lessons so we can put an introduction just after algebra, to put a little more meat into the basic science sequence at that level. I think I'd better tell him what's happening. I start another NetBoard session.
Simba: Hi there, Tiger. What's up?
Me: I have to keep this brief. The W32 flakeouts are real, and someone sent in a killer piece of code. We have to recall. Frank wants me to dou dzun at 13:00 today; you should watch my performance on the financial channel. I'm still going to be in a bad mood this evening and if I could prevail on you, I think some sushi for dinner would perk me up a little. I promise, if we're onto something by evening I'll come home to eat, then go back in for an all-nighter. And another thing: what do you think is going to happen to Whinx stock prices?
Simba: I can't legally sell Whinx short; it's inside information.
Me: Sell something else. This is a big recall; it may take down the whole tech sector. Then maybe tomorrow morning Whinx will hit bottom and we'll use the money to buy it.
Simba: Good plan. But let's skip the all-nighter, OK? Steady progress gets the job done, whatever we may have done in college. Surya says hi, and he has some questions for you about a lesson when you get home.
Me: Hi, Surya. OK, I'll do my best. I have to write the press release now. I love you.
Simba: I love you too. Be tough. Bye.
The press conference is embarrassing; I much prefer to have people remind me what a wonderful lion person I am. But it's a good lesson for all the kids out there who use the Lion Disc to see someone they know stand up there and admit a mistake and take responsibility for it. I hope the general news service will pick up the video. I'm honest: while the problem is real we don't know yet what it is, but once we find it we probably won't have to do too much to fix it since the chip is basically sound. While we aren't happy about the cost of giving everyone new chips, we've been shipping for only a month and volume hadn't grown anywhere near that for the W30, so the company isn't in danger financially. I apologize for putting users through the bother and uproar, but I remind them that only one company developing a fairly unusual board had trouble, and it's my hope that nobody else will even notice the flaw, until the new chips arrive.
After that's all over I settle my nerves by making myself a cup of hot chocolate in the conference room kitchen, and one for Frank.
Frank: You seem pretty sanguine about this.
Me: Like I said, we find it, we fix it. I don't want to minimize the problem, but I don't want to maximize it either, like a bunch of fund managers probably are doing right now. I've left Simba instructions to sell other tech stocks now, and to buy Whinx when it hits bottom, probably tomorrow.
Frank: Putting your money where your mouth is: I like that. I have confidence in you; I did the same thing.
Me: Good luck to both of us. This morning I left Alan with a case of W32's and a lethal code fragment. I'd better get up there and see what it's mutated into.
Indeed, Alan has the code pared down to a sequence of twelve instructions, and he's able to demonstrate that if you run those once followed by a halt, then interrupt out of that into the tests, 23 microseconds later the ALU bits start disappearing. So it can't be thermal, but what is it? He's poring over the chip map as I come in.
Alan: See these flipflops here? The bit lines aren't in order due to the compression functionality, and it looks like something is marching out from this line, anywhere between the two rows of flipflops and the gates into the address adder, and cutting neighboring lines in sequence.
Me: Great work! That was fast. But it's also a disappointment: I was sure it was thermal, and I was hoping our overheating problem might be related. Let's check the hypothesis by targeting the bit lines on the other side; kind of make a mirror image of the data pattern.
It was too naive to just shift the bits, but shifting and then complementing gets the hatchet job done. From there we get a feel for what's happening: a particular relation between bits going through the ALU output register, which has two staggered rows of flipflops. If the data is duplicated in the two rows with a particular phase offset for a minimum number of nanoseconds, bad things start to happen. We're able to write our own instruction sequence that kills chips, proving that we know what's wrong, if not why. I send Alan home, write a report for Frank, and get myself back to our house for sushi.
At home Surya and Simba are waiting for me with broad grins.
Surya: Hi, Tiger! You were great on video! We saved the file. Now I know why Simba was selling stuff right and left. I sold out General Semi. When I buy Whinx tomorrow I'll be ahead almost a thousand dollars. We sold almost two million dollars of family stock, tech issues like you said, and they're down five percent already, and Whinx is down to 48. All us kittens will clean up when we buy Whinx tomorrow, plus Coyote and Mariposa, and the other lions too.
Me: I was great at making myself look stupid. Well, it's over. I'm going to do a few minutes of breathing exercises, and then we can eat. Or is your homework question restful?
Surya: Probably. It's physics, and Simba said I should hear your version of the explanation. The problem has a speaker at the end of a tube which makes a half wave resonance, right? So suppose I put two of these back to back so the speaker is in the middle. I'm supposed to get a full wave with the ends in phase, but when I do the calculation they're out of phase, like the half wave. Simba explained why it's wrong but I had trouble to understand. See here?
Me: I see it. Your speaker is like a balloon, pushing the air out from the middle of the tube. A real speaker would slosh it back and forth. Can you see that the phase would be fixed then? Try the problem again with the balloon idea for real: pump air into and out of the middle of the tube. You can see, can't you, that a real balloon is full of air and the waves would go right through it.
Surya: I see it, and I can see now that Simba was saying the same thing. But I'd better do the problem after dinner because Simba is serving the veggies. Let's set out the chopsticks and eat.
Me: Right. How's your college essay coming?
Surya: I made the changes we talked about, and Sandy Ramirez suggested I should emphasize being a lion person because they like people who are unique. So I rewrote it that way, but I thought it didn't sound good so I rewrote it showing ways I'm unique without actually saying I'm a lion. Like, I run a lot at night because I can see thermal infrared, and my claws help me climb trees.
Me: That's good; that improves it. Your job...
Surya: That's under work experience. It's not exactly academic, trimming trees.
Me: I know it's under work experience, but what are the admissions people looking for?
Surya: Someone who can handle the academics, who will do the work rather than goof off, and who has something to contribute that the other kids can learn from.
Me: And you study at home. If their profs give you assignments, what will you do with them? How do they know? What example can you give?
Surya: OK, I sort of see your point: they have to worry that I'm so used to supervising my own lessons, not that it's really like that, but they may think I won't do the teachers' assignments. But I have to do what the crew boss assigns me, and that's a good example to give them confidence in me. And I can work in a reference to claws, and it won't go over my length limit. That's a good suggestion; I see it now!
Simba: How's the sushi? Surya picked out the fish and steamed it. A skill you could mention: cooking. And Surya, can you squeeze into your essay a comment about today's stock spasm? That will give them an idea that you know something about money, a lot more than I did at your age, and are more likely to pay your bills. That's important to them.
We chatter about the day's events while finishing up the sushi and veggies. After we all wash the dishes Simba and I go over the three college essay versions and give Surya more detailed comments, and then we relax with some music: an impromptu arrangement of a Beethoven trio with me on violin, Simba on synth, and Surya's sax taking the cello part. I'll really miss Surya, our youngest, but the lion species has to grow, and he has to go to college and then form his own family with his fiancée Holly, Adam and Elsa's last kitten.
The next day Alan and I take the cover off a dead chip and put it in the scanning electron microscope. It looks like a chip. Hmmm... I zoom to the area between the two rows of main adder output flipflops that Alan identified yesterday. There's a very faint line there.
Me: Alan, do you see that line? Let's put it in X-ray microprobe mode; would you turn on the detector? Thanks. What's it saying to us?
Alan: More oxygen, less aluminum, less silicon. It's an odd pattern but there's a lot of variation just from natural texture anyway.
Me: True, but. Let's look at that area when the chip is dying. Something has to cause the line. Can you feed the test jig signals through to the microscope? I know we have a cable for that somewhere.
Alan: Yes, I have it in my drawer. I'll set that up while you put in a good chip with no cover on it.
The SEM isn't like an optical microscope; it takes us about half an hour to change chips and get everything wired up. We run simple test patterns very slow, and we can see the information running through the chip, specifically the ALU main adder. Now for the real test. Alan can key in bit patterns, harmful ones, that will end up in the output flipflops. I zoom to the small space between the two rows. A four bit pattern gives no trouble. Six bits: nothing. Eight bits: there's something happening along the line but I can't see what. Ten bits: you can see the line clearly, but no clue why. Twelve bits...
Me: What in hell is that?
Alan: It looks like a hole. A crack, right along the line, and the central eight bits aren't getting out of the ALU.
Me: Put in all zero. The hole's gone. The twelve bit pattern again. I don't see it.
Alan: If the connections to the second row of output flipflops were cut by the crack, I can't load the pattern again, at least on this chip.
Me: Damn! I know; put in the sixteen bit pattern on the other side and I'll shift over there. Sixteen bits of death. Hey, what happened? Where's the chip?
Alan: Put it on low magnification... Holy shit, it took the whole side of the chip right off the header, a perfect cleavage! The bond wires are holding it above the header.
Me: That... is... incredible! Look, I have a feeling about this. I'm going to do a few minutes of breathing exercises, and would you pack up all the images in a compressed file and timestamp it? Make an index file for them and mail Frank the URL, then go get a cup of coffee and relax. At, let's say 11:30 we're going to make some heavy plans.
The plan has two parts. First, of course, is to cure the W32 chip, and I'm pretty sure that just turning around the second row of flipflops is all it will take, given how the bits seem to go together in the killer patterns. But Surya's physics problem ties in here: if something pushes apart the flipflops, even if not tearing the chip apart, it's going to generate sound waves in the silicon, which will eventually be dissipated as heat. There are pairs of related flipflops all over the chip, though not in such long rows, and that could be a cause of our overheating problem... and the cure, if the flipflops can be rotated to spoil the interaction. So the first part of the plan will be to make some test chips with just the ALU output flipflops, in various relative orientations. Assuming a favorable result in that step, we'll redo the W32 layout, test it, and get it into production. That will take embarrassing and expensive time, but not a lot of brains. And I'm sure that with the heat problem solved we can jack the speed up to 1.5 gigahertz; what a coup!
But spoiling the interaction is the opposite of what I really want to do. At 15:00 I'm meeting with Frank, who already approved phase one, and I'm outlining the real plan.
Me: We're calling it coherent quantum momentum transfer. There's got to be a lot of uses for CQMT. I can think of two right now. We now have the best, the fastest electroacoustic launcher in the business, and we should be bringing out a line of gigahertz mechanical filter chips. And second, if it can tear apart chips when we don't want it to, think of the precision if we put a border of these cells around each chip. Feed the right bit pattern onto the wafer and plink! The chips fall apart. No more scribe line alleys. No more fragments from the scriber. What do you think?
Frank: I'd be a lot happier, and so would be a bunch of non-technical people I could name, if we knew how it worked.
Me: So would I. I'm proposing to set up a shell corporation to deal with the new effect. Let's call it Coherent Quantum Momentum Transfer, Incorporated. Its assets will be the patent, which we'll file in a very few days, and my not inconsiderable self. Plus I want Alan; he wants to follow this through. And a few techs. It will sell acoustic technology and chip dicer borders to Whinx, and whatever else we come up with, which will give Whinx an excuse to feed money to us, and we'll figure out what we've got. And then the real commercialization begins.
Frank: The name will have to change. Can you have the W32 redesign done in a month?
Me: I'm assuming favorable results on the tests we're going to do, but with that assumption we can have a 1.5 gigahertz chip designed, tested and started in accelerated life tests in a month. Assuming I have priority on mask fabrication.
Frank: You do; I already told them. I think the board can be convinced about your corporation, but we have to put a time limit on it. I think I can get two years.
Me: That's fair. If we're not self-supporting in two years we'll evaluate when, if ever. I'm realistic; we could lose the game. But I have a feeling about this one same as I did about the W30 and the W32. I want a big option on the new corporation.
Frank: Like Simba did on Xylogen? Give the board a 1.5 gigahertz chip and I'm sure they'll give you something in return, particularly if it doesn't cost them much money. Go on, get out of here. You'll have your corporation, after the W32 is cleaned up.
October passes in a blur, and the first week of November. I keep the all-nighters to a minimum, but I have precious few weekend days free. Simba and Surya are perfectly able to keep themselves amused, but I'll be happy to get back to a normal schedule so I can join in the fun. Not that work isn't fun, but I'm really looking forward to finishing this redesign, and not least for having some time to think, to try to figure out what's causing the CQMT effect.
The job takes five weeks. That's over the promised month, no surprise, but not much, for which I'm proud. The masks worked fine for us but long experience has taught me that the production department will find lots of things to screw up, or as they would say it, lots of stuff the designers have screwed up, so we'll be putting out little fires for a few days more and driving the mask fabrication department crazy. By the new year we should have full production, and with a 1.6 gigahertz chip (better than I expected) we'll be cutting into the market shares of a whole new class of competitors.
At last it's over, and the whole family are taking a break at the Lion Foundation territory up in the Utah hills: Crazy Felix Stenhorth's so-called vacation cottage. Living in Salt Lake we're the closest lions and we visit here frequently, but it's not the same since Dr. Franck and Dr. Newman died. They led the team that built us. I'm 44 lion years old now, counting from assembly, and Surya will be 17 next summer. Dr. Deutsch and Marjorie are still here but they don't travel much any more, and it's mostly new faces.
Surya is off exploring the other side of the property, or bothering the staff, or being teenaged. Simba and I are enjoying being together on our favorite flat rock along the stream, which has not yet frozen over. We pushed the thin blanket of snow into the water and enjoyed ourselves horizontally, and now are vertically considering our future.
Me: I've been letting this business soak into my mind for the last month, or more like, having it pressure impregnated. But I haven't a clue what's doing it. Look, let's go over the characteristics. Without quantum coherence you get nothing; not just averaging to zero, but zero zero zero. We can tell that from the acoustic data. It's a quantum effect; no question about that. It's proportional to the wave function overlap integral, sign included: if you swap one flipflop end for end you get compression rather than dilation. So we just went through the chip and turned a bunch of flipflops a quarter turn; problem solved. Well, we had to do the right ones or we'd just make new interactions, and we had to be creative routing the wires around the corners, but it worked out.
Simba: Do you get angular momentum transfer?
Me: Yes, but with the small feature size you can just barely detect it. Now here's something strange: you expect an exponential dropoff but at least within the limits we were able to test, the force is pretty much the same from a tenth of a micron to ten microns. You just need enough coherent units in the two groups, and it happens.
Simba: Hold it! Not 1/r dependence, which you'd expect for a row of active units?
Me: No. Let's see how could we test that over more range. We'll put a row of the things right up to the edge of two chips and put one of them on a movable stage. An acoustic detector; we cycle the bits for force, then no force. Now how do we maintain coherence on separate chips? Suppose we have a pseudorandom sequence? No, the starting values would have to be not just equal but coherent and we're back where we started.
Simba: Digits of pi? There's only one pi.
Me: Good try, but you have to start the calculation with coherent synchoneity; it's no better. Anyway, we can make a temporary connection between the chips, initialize coherently, then back them apart. I wish I could go to the lab and do it now!
Simba: The same from 1e-7 meters to 1e-5 meters, to... what? 1e17 meters? Somehow I don't think so.
Me: Let's keep this focused. 1e-4 meters would be spectacular enough. Let's not waste time speculating about something I haven't measured.
Simba: You're right. But we're talking momentum transfer here, right? By the principle of relativity that implies energy transfer. You should be watching for that.
Me: That's an interesting way to look at it. I see a chip being wedged apart; you see juice being transferred across the crack. Very interesting. Let's go in now; I'll whip up the design and send it in to fabrication, and the chips will be ready when we get home.
It's a good vacation; very restful. Simba and I hike up to an abandoned fire tower on the other side of the valley and illegally spend the night there, high up above the mountain. It's a long story why that tower is important to Simba. Surya tests himself by running a long route around the same mountain, on the day we hike back. It's tiring and he hasn't practiced enough, as I've pointed out to him, to be able to run so far continuously, but he leaves well before sunrise and returns just after dinner finishes: thirteen hours. By anyone else's standard it's superb performance and he's mature enough to understand that and to accept congratulations and hugs, and an enormous pile of freshly baked bagels, in the spirit that they're given. Next time up, after we've practiced properly, Simba and I should see if we can still do it in eight hours. Surya says we're old fossils and shouldn't risk it, but I feel like a teenager.
Returning to work on Monday, I brief Alan on the test I want to do. He checks out the equipment and sets it up while I catch up on paperwork.
Alan: OK, Tiger, do you want to see it?
Me: One second; let me save my file... OK, let's do it. How much signal do we get with the chips pushed together? Is the design doing anything at all?
Alan: Seventy millivolts. It's a modest signal, not enough to break anything.
Me: OK, move the stage step by step and record the signal at each step.
We're doing this slowly so it takes five minutes to get the chips ten microns apart, about the size of a largish germ. The signal varies erratically between 67 and 72 millivolts, thus reproducing our previous, more tedious measurements. Alan speeds up the steps, racing into uncharted territory. The signal continues in the same range. We wait for it to drop off. It doesn't. By 10:30 the moving stage has reached the end of its travel, two centimeters. Compared to the fifty Ångstrom feature size and the one micron size of the rows of flipflops, the mobile chip might as well be on the moon... or alpha Centauri.
Me: Alan, mortgage your kid and buy stock in this company. I'm serious; there's an employee stock ownership plan, which is mainly for me but any employee can use it. We've discovered something big!
Alan: I can see that, but what's going on? What can we do with it?
Me: I want to do three types of test. First, obviously, we have to measure the orientation dependence, so we'll disassemble the stage and just hold the thing in our hands. Assuming a coherent result, then I want to really find out about distance. I want something I can put in a package with a battery, and mail it to Singapore. Third, we've got to put a lot of these things on a chip and see how much force we can get out of it. How about we'll do the orientation part right now; then after lunch you make up the chip to go in the mail, and I'll start on the high force chip. I have to figure out how to get the units synchronized in bulk.
Alan's job is the easier; he just has to modify our test chips with proper wire bonding pads. He has the masks ordered by the time we go home from work, and tomorrow he'll do the chassis and microprocessor to control the three chips, one on each axis. My high force chip is harder due to the synchronization problem. Likely there will be some strange interactions between the units, and I'll have to be flexible in the bit patterns I load into them, to work around problems. It's like designing a memory chip. By quitting time I have the goals and issues worked out, supposedly, and I have a block diagram of my chip. Alan will help me with the rest of it as soon as the mailable unit is done, and with any luck a lot of the tedious parts can be copied over and over by the computer.
At home I slice green peppers, squash, eggplant and onions, while Simba nukes some sweet potatoes and Surya mixes up a spicy cheese sauce he likes. Over dinner I report my progress.
Simba: So how did the test go? You look like the cat that ate the canary.
Me: Canary, an apt metaphor: flying like a bird. Surya, this is inside information and you mustn't breathe a word of it, not even a comment like ``my supervisor has this dynamite project...'' The force between the chips has unlimited range.
Simba: It can't be unlimited; that doesn't mean anything!
Me: Two centimeters. Alan is putting together something that we're going to mail around the world. I expect to see no variation in the force, except when the package rotates. I think we could send it to alpha Centauri and it would still work. I think we will send it to alpha Centauri.
Surya: Jeez!
Me: And mate, I intend to be on that ship. You coming?
Simba: Jeez! That means yes.
Surya: When do we leave?
Me: Surya, Surya! I'm not going to tell you to analyze your question; I'll just tell you. The first answer is, not for a long time. We haven't even done the Singapore test! We don't have the drive chip. Simba, you're absolutely right that there should be some way to transfer energy to the spacecraft through this effect, and I haven't even started on that part. But Surya, there's another answer. Your duty to the lion species involves eight kittens. If you and Holly follow the same schedule we did, which is likely, you'll be free to leave when you're as old as we are now, in 28 years. I doubt it will take that long for us to succeed, or to fail and to know we've failed. In either case you won't be going with us. I have a distinct feeling that we won't see any of our grandchildren go off to college, unless we survive the trip.
Surya: What do you mean, survive?
Me: The drive unit will have to operate for a minimum of eight years under heavy load. It could break. Or suppose we hit a dust grain? Or suppose we just do something stupid during the exploration?
Surya: But...
Simba: Developing the CQMT effect will take years. This whole idea, this plan: it's hard for me to have my whole life turned around yet again in an instant, like when we got married or when Tiger got injured and I got the Lion Disc job, and I'm sure it's equally hard for you, and for Tiger. Let's not try to understand it between one bite of sweet potato and the next. We'll all deal with it gradually. But I think Tiger's right: this may not be a suicide mission, but there's a quite high probability that we won't come back. I decided long ago that a trip to the stars, hypothetical of course, was the sort of thing I would be willing to expend my life for. And now it may come real, and I'm putting my life where my mouth was. I'm going with Tiger.
Me: Heavy philosophy is fine, but there's a prerequisite, and for that I'd like to do a little brainstorming: where is CQMT going?
Surya: CQMT; I wish you could make some kind of word out of that.
Me: Well, then, you think of something. I know that CQMT can produce enough force to crack a chip. I'm predicting that in the pusher test I can peel a chip right off a standard header; we already broke one chip that way. Hmmm, we'll put them in hockey puck rectifier cans; that should be heavy duty enough. Now what are we going to push?
Surya: Satellites into orbit!
Simba: Where we're going, we don't need... road!
Me: Bulk power lines, using the as-yet-undiscovered Coherent Quantum Energy Transfer. Your ideas are good, but I like the satellite launch idea better than hover conversions because you have fewer bigger customers, less administration, and less work to distract me from the main job, which is turning that cash flow into a working starship. Someone else, like Surya if he turns out to have that kind of talent, can do the hover conversions. You may think this is kind of a stupid planning session, but I need it to get my ideas focused: what are we going to do with CQMT? That determines what aspect I work on. I wish I had some idea how it works! We need someone with advanced training in physics; I'm an amateur.
Simba: The W32 designer calls herself an amateur in quantum mechanics.
Me: Be realistic. I know how to kick the bits, but as far as I can see, quantum mechanics works and we don't know why. It works on something else now, and we still don't know why.
Surya: You could go to grad school.
Me: Charlie did, with a good result: he's a deputy commissioner now with the SEC. Diana did a CPA course. Leo did what you might call postgraduate work at Julliard. Adam and Elsa both went to law school. We'd have to decide which of us has the talent, if either do, and whether studying for a doctorate would hold back me or Simba from essential work. You're welcome to try, if you have the talent.
Surya: I thought I was doing hover conversions.
Me: After college. Bill Gates might have been justified quitting school, but not you for CQMT. Actually, I'm guessing about time scales and a thesis on CQMT would come at a pretty convenient time. I don't want to push you into a field you're not suited for, but if you have something to offer there, kitten, it would be most welcome.
Surya: OK, I'll be prudent.
It takes to the end of the week to get a mail package assembled, tested by driving around town, then packed off to the airport for shipment to freight forwarders in Seattle, Singapore, Cape Town, Frankfurt, New York and then back here. Friday afternoon my pusher chips come in. I can load an eight-bit pattern into the first few units of the first very long row, or into one place anywhere else on the chip, and it works fine. After that, either the force doesn't increase or it goes to zero. Well, you can't expect miracles on the first day.
Unfortunately the chip continues to not cooperate for another day, and another. But the Singapore package is holding up well. Our test chip turns the force on and off at about 100 kilohertz and we detect it by the acoustic waves in the silicon of the partner chips in our lab. Alan put a little microphone in the package and he frequency modulates the pulses. His idea was that the four battery cells might look on X-ray like dynamite or a detonator, and if nervous security agents wreck our device we'll at least know it was through no fault of the chips, and when and where it happened. I'm glad we did that, because yesterday at Cape Town some people paid very close attention to the package for two hours. They spoke Afrikaans, which is kind of like Dutch, which is kind of like English, and from a few cognate words and the sounds of handling we figure they went as far as to unwrap the package and see the transparent plastic chassis with the batteries. Then they laughed sheepishly, wrapped it back up and sent it off to the freight forwarder. I hope they didn't just throw away the bubble wrap.
Me: Hey, Alan, it just came to me: remember the security check yesterday? We have a commercial product there: point to point communication, unlimited range, and way faster than 100 kilohertz. Let's spend the rest of the afternoon putting together a writeup to get the patent people started. I'm not making any progress on the pusher's bit pattern, and I don't think you are either. And I'm going to sell all my telecom stocks.
It's a big coup, a definite moneymaker for TransForce (the new corporation), and Frank tells me he knew I could do it, but I'm just not able to be as happy as I ought. The main work isn't progressing.
It's December first, and for the last several weeks I've been struggling with the pusher chip, trying to get my mind around it, but I just don't get it. Well, there's no point dragging my tail over it; we have plenty of time and plenty of credit with the board of directors, but I'm used to having my way with silicon chips. Speaking of which, CQMT is a complete dud on gallium arsenide, but it works just fine on chemical vapor deposited diamond, once you get over the bizarre voltage levels you have to use and the complaints of the production department about the horrendous processing problems. And it works on germanium, not that that's particularly useful. It likes pure elements and doesn't like compounds. Strange. My thought this morning is interrupted by an eep from upstairs, instantly cut off.
Surya: Tiger, Simba! Could you please boot your machines and join a NetBoard session? Holly's machine.
We do so. The network delay from Salt Lake to Illinois is noticeable. Adam and Elsa are already online.
Simba (on NetBoard): Hi, people; what's up?
Holly: We both got into MIT!
Me (and Simba simultaneously): That's great! Wonderful news! Congratulations, Surya and Holly; we're proud of both of you.
Surya: What we wanted you supervisors in for is this: You people got married at about sixteen years, but you were able to live at the Lion Foundation and your supervisors were right there. Then you had to split up to go to different colleges. We've talked to all of you about this: we felt after mate selection that we weren't mature enough to go off on our own, and it was going to be best to have a close friendship over the net, and to visit as much as we could, but not to actually form a family unit until we graduated. That seems to have worked out best for the older kittens. We asked MIT for early action and hoped we'd both get in, and hinted to them that we're a mated pair, and they came through! So we're changing our plans; we'll be married at MIT. Do you have comments, so far?
Elsa: Marriage was a big adjustment for me. College was, also. Many of our lion policies are designed to allow you to do one adjustment at a time.
Holly: We agree. Surya and I would like to start living together, like, soon.
Adam: That's a rather sudden additional change of plans.
Me: Here's an issue: where would you live? I see three possibilities: here, there or neutral territory.
Surya: Could I say this, Holly? As the outsider... Well, like I said, neutral territory is a distant last choice. I've always been the little kitten in my family and I'm used to the idea of not being ready for what the older ones can do, like living separately, and then growing into it. But who switches supervisors? Simba and particularly Tiger have a very direct style; if they don't or do like something, they say it right out, and I'm used to that and while I don't enjoy getting chewed out, I think it's easier for me Tiger's way, than if I have to, you know, interpret. And I have to really work to fit with your style. I've been nice as I knew how to be when I visited you, but quick enough you'd find me getting disrespectful, where Tiger and Simba appreciate my candor. Can you see where this is leading?
Elsa: Beyond that you'd prefer to stay where you are?
Surya: Right, beyond myself.
Elsa: What does Holly think about this?
Surya: You know my supervisors pretty well. You've been dealing with Holly for a long time too. Thinking objectively, which group do you think she fits in best?
Adam: Holly is being awfully quiet, while you answer our questions.
Surya: Yes, sir. What she's feeling is important right now, not as said by her but as understood by you. Put yourselves in her place.
There's a rather long pause. I originate my own NetBoard session and get Simba and Surya on it.
Me (privately): Do you think they're talking it over with Holly?
Surya: I have the impression that she's alone in her room. They ought to be thinking, and maybe talking with each other. I hope they go along and tell her to come here, because she's too loyal to tell them she wants to leave. We had an argument about that, a few days ago. Look, different topic: I really enjoy being with you people, like when we play music, and I'd miss what I learn from you both. But I will go to Holly if she won't come here. I'm sorry.
Simba: Does she want to stay there? Is she avoiding us?
Surya: No! She loves to visit here, and Adam and Elsa are so damned legal, as she puts it. But she doesn't want to just walk out on them; that's not right! Adam and Elsa can set this up so someone loses on any choice we make, and you've trained me too well to be ruthless when I have to be.
Me: Look, I'm not proposing this as a good solution, but suppose you stay here, she stays there, then you live apart for half a year at MIT, get your first semester jitters under control, and then join up?
Adam (on the main session): We've been talking over here. It would be with our blessing, if Holly moves to Surya's territory.
Simba: We welcome Holly to our territory.
Me: We've enjoyed Holly's visits and we would enjoy having her here.
Holly: I didn't want to run out on my supervisors. But I'm so happy to be able to join with Surya now rather than in four and a half years! I'm coming. Thank you, all of you, particularly Elsa and Adam.
Me: Call this a wedding gift: I'll set up the airplane ticket through the Whinx travel office. I'll try for Friday morning, if that's OK.
Elsa: What shall we do?
Simba: Does she have a mountain pack frame? We have some camping planned around the Lion Foundation. And a long run.
Elsa: She will have.
We work out a few details, exchange a little news, and log off. Things do change suddenly, don't they? Surya comes down and hugs both of us.
Surya: I'm happy, but I'm scared too. You two are never scared.
Simba: Of course we are. What do you do when you're scared?
Surya: Same as you, like you taught me: be extra careful, and do what you need to with minimum fuss and risk. My mind is jumping too much! I've got to run it off before I do any lessons. Do you want to come along?
Me: I'm late for work anyway. Wait just a second while I send mail to Alan; when you called I was just thinking of a bit pattern I want to try, and he can get started on it.
So we run around the park two blocks from our house, silently. My mind won't let go of that bit pattern: if it's coherent then why won't it link up with the pattern on the partner chip? For that matter, there must be some interaction with the second bit group both locally and on the partner chip, because the second set of partners are coherent too. Except they're not. The bit patterns are the result of separate computations, so the partners are coherent pairwise, but the two pairs of partners are independent. So the interaction comes out to zero; why? It won't be zero initially. But after the phases affect each other... randomly!
Me: I've got something on the pusher chip problem. You two keep running; I'm going straight back and start working on it. And you'd better call me to remind me to come home.
It's not easy to modify the loading program so multiple bit patterns stay coherent. There's also the issue that we don't want the chip to tear itself apart, so every bit pattern has to be different. We can get away with eight bits at a time without breaking the chip, so there are only 32 different patterns that will produce a force: not enough to even cover our test chip, much less the anticipated production version. We're not able to do the tests today and I can see a tough path ahead of me, so when Simba calls late in the afternoon I send Alan home and go back myself.
Me: Well, what's all this food I smell? Smoked fish curry; this seems to be Surya's style.
Simba: We're celebrating getting into MIT, and his marriage.
Me: Right. Holly likes ratatouille. Let's make it for her on Friday; I got the airplane reservation and she'll be here at 13:30. Surya, would you see to getting the vegetables? And we should have some kind of wedding celebration on Saturday. Let's plan that after dinner.
Simba: So how did the chip do?
Me: Tough, but we made progress. Tests Friday, I hope. You two can handle picking up Holly from the airport, right? Drive me over to work and keep the car.
It's a fun dinner, and the break into family business gives my brain a chance to digest something too. Later in the evening we're playing Leo's third string quartet (with Simba handling both second violin and viola on the synth) when the 5/4 rhythm at the start of the second movement gets me thinking pentagonally: Penrose tiling! I restrain myself from working on it until the morning, but then I outline the idea to Alan, warn him that the arrangement of bit patterns may have to change, and get to work to see if Penrose tiling is at all relevant. The idea is that the pseudorandom arrangement of the tiles should lead to a nonzero but statistically small within-chip force, unlike the square tiling we'd been planning on, where all the forces would add with the same sign and destroy the chip.
Yes, indeed, the Penrose tiling will be helpful. The chip will work a whole lot better if the groups of registers are laid out on a pentagonal pseudorandom grid: oh, joy for drawing out the masks, for we normally copy a square or at worst hexagonal grid when doing memory chips or similar repetitive designs. We'll have to write a special program, with graphics, to create hectares of layout with every eight-bit register individually positioned in its pseudorandom location. And then I have to figure out how to address the things, coherently, to load their bit patterns. Well, we can't expect to get something valuable just by snapping our fingers, can we?
Anyway there's an upside: by loading a tenth of the registers on the test chips we can approximate the Penrose grid close enough that the chips will hold together. I work on the code that picks the spots and Alan continues writing the loader itself. Friday just before noon we finish the program. We want to see the result. We load just forty registers, and the force is enough that I can feel a very faint ticking when I put my finger on the chip cover, the modulation being turned down to a few hertz. With 640 registers the ticking is actually audible. With 5120 registers the two circuit boards move apart in little flea hops. Each of us holds one circuit board and I enter 20480 registers as the number to be loaded, and press return. Our circuit boards with the test chips on them push very definitely into our hands. That's about as much force as I want to risk putting on the thin metal legs of the chip carrier. Clearly if I loaded all the usable slots on the chips I could put one board through the wall into the parking lot, and the other through the wall into the neighboring office and, possibly, its occupant.
Frank is pleased with the progress. I negotiate with him to borrow a Whinx programmer, because the chip layout program will take a lot of work and Alan and I are most productively used in designing chips, not in big programming jobs. Alan and I work on fleshing out the program specifications, and the programmer joins us just before quitting time to pick them up. We impress on him that we'll need to get into the program and fiddle, for example changing the size or orientation of the registers, or feeding additional data to them for initialization, so he'll have to make everything modular and flexible. We'll have lots of conferences before the job is done, but next week Alan and I can get to work on another important aspect: how to control the chips. Some applications like satellite launching can use a pusher that's all on or all off, but the more interesting ones like hover conversions require delicate adjustment of the force.
On my return home (by bus), Holly is there. I give her a big hug, then Surya and Simba.
Holly: Thanks so much, Tiger, for letting me live with you.
Me: Thanks for coming, even if only until next fall. I'll miss you two a lot.
Holly: I wish we could be in two places at once. You know, there's something a little embarrassing I need to ask you. Financial. We didn't exactly decide what my status is.
Me: Legally you're Adam and Elsa's kitten, here on an extended visit. Your right to our fortune is through Surya, not directly as our offspring.
Holly: I didn't mean that! I meant cash flow; the food budget. Not the education budget.
Me: Oh. Be frugal; don't waste stuff. You know how we operate around here. But don't worry about the budget. If we send you to the store for groceries, make sure we give you money. Surya has a job and his own account; do what he does, as for what to pay for yourself. Don't worry about the budget; remember the word ``fortune'' I recently used?
Holly: Thank you, but I don't want to sponge off you just because you're rich.
Me: Not legally, but in fact you're part of this family. Do it; be a member, including with your mouth. That's not sponging. Now here's a lesson in family life: Surya is supposed to have gotten some vegetables, and let's get them ready to eat.
Surya: They're in the refrigerator.
Holly: Oh, look, we could make ratatouille out of these.
Me: Our thought exactly, and Surya has changed his recipe since you were here last. You'll like it. You do the eggplant; remember, knives are in this drawer. I'll take the onions and squash; Surya, do the tomatoes, and Simba can start the rice and then cut the peppers.
Holly: Yes, I remember. Thanks, Tiger, for telling me how things are going to be financially in about two sentences. Surya was right that I'd get along better with you. My supervisors would have gone around in circles and I probably would have gotten mad at them. I don't want to badmouth them, but it's the truth.
Simba: Elsa is a cop; I'd expect a very direct style from her. She's pretty direct with me; so is Adam.
Holly: Probably it's because I'm their little kitten. I hate being the youngest; they're always trying to make things easy for me. And I fight.
Simba: I had to work hard to avoid that with Surya: to supervise him rather than making him a live toy.
Holly: Oh, they didn't do that; they did their job, but they had to work at it too, and it just took so many words that I had a hard time to cope with. Maybe I'm going on about that too much. May I ask... Surya said there's a lot of stuff going on here and you warned him to keep his mouth shut until I'd had a chance to talk to you two. Could you tell me?
Me: It's inside information, and there are regulations. Charlie, not to mention your supervisors, would be honor bound to nail your and my hide to the wall if we violate them. What I suggest is that you decide for yourself how much Whinx stock you want, and buy or sell as appropriate, putting the order in tonight for execution when the market opens tomorrow. I'll write up a letter which we'll both sign, saying that the order was made before any inside information was given to you. After that, and until you go off to MIT, you can trade the stock only when we do. We know when it's legal to deal in it: basically, when the inside information has been announced publicly.
Holly: I have a hundred shares, but I don't have the cash to buy any more.
Me: Well, that simplifies the letter. We can do that after dinner, and if there's to be a dinner we'd better get these vegetables and Surya's garlic sauce into a pan. OK, now, if you're to be a member of the family you need to keep this absolutely secret. You don't say to your siblings, ``Tiger has a secret project''. You say nothing at all, not even rolling eyeballs. If you're asked what Tiger is doing, you tell them there's a legal restriction on TransForce internal information and you aren't allowed to say a word. If you can't do that, we're going to have to work out some arrangement; you get all the information or you get none.
Holly: I can keep a secret.
Me: How do you know? I'm serious; is there any objective evidence?
Holly: In games, like poker, I win them by not letting out secrets. One of my human friends asked my advice last year, and I could have told other people but I didn't. I'm not telling you what it was either. I've had the lesson on secrets and I was able to do what I was taught.
Me: Good kitten. I believe you. By the way, I've talked with Surya about this issue and he judges that you'll keep the secret. OK, here's a quick summary, none of which is public knowledge. You know about the W32 recall, but not what was breaking the chips: the coherent quantum momentum transfer effect. Whinx set up a separate corporation called TransForce to commercialize it, and of course I have a considerable chunk of its stock. We already have four products, not in production but in various stages of development. Two are nice but not world-shakers. The third is long range communications. A little package this big; we eavesdropped on airport security people in Cape Town. From here. Computer networks, phones, they won't be localized any more and every one of them will have a Tiger Leones CQMT chip in it.
Holly: Wow! Is that big, or what?
Me: That's just the appetizer. In stories they have tractor beams. CQMT isn't a tractor beam, but you can exert a force between two chips, and there's a good chance you can equally well feed energy between them. The applications we're planning to target initially are hover vehicles, and space launches and propulsion. Plus high level power transmission, if I come through with the energy component.
Holly: Jeez!
Me: Now the real inside information. Neither Whinx nor TransForce know about this one. Think about space propulsion. Where would you send a ship, if you had a CQMT pusher?
Holly: The moon? I've got it: asteroid mining!
Me: Good thought, but we've decided that's too complicated for me personally to handle; TransForce will just provide the chips for it. Pick one destination.
Holly: Jupiter? No? Tell me.
Me: Understand, we could run into a stone wall at any time, but the tentative goal is to build a starship and to go somewhere in it. Suppose alpha Centauri. Sublight, but near c.
Holly: Jeez!
Me: Does that give you an appetite for dinner? It does me. Smell that garlic! Would you put this on the table, please, and Simba, serve the rice.
Holly: Tough as a lion; I'm not some flighty teenage human girl, but that puts butterflies in my stomach. The stars! What a family!
Me: Well, we have to make it work first. We could lose. But it's worth trying, in my opinion, and Simba's. So how do you like Surya's style of ratatouille? We shouldn't get totally unbalanced; worldly things have a place in our life too.
We have a nice mundane dinner, but the discussion before it has gotten something moving in my brain. After washing the dishes we try a little Mozart: Holly has brought her clarinet. Holly's circadian rhythm is on Peoria time, so she and Surya go upstairs early.
Me: Hey, lovebirds, you don't have to giggle so much. This nosy mother-in-law has a suggestion. Holly's tired and jumpy. You two get a good, peaceful, quiet rest tonight, so you can have a fun time tomorrow. Get the picture?
Surya: Yes, mother-in-law.
Holly: Good night, Tiger and Simba. And thank you.
I'm not going to take my own advice. The feeling stirring in the back of my brain won't wait until Monday; I'll feel miserable until I work on it. I copy down the eight dimensional Dirac equation, and a simplified solution of it that I believe is pretty close to the wave function of one flipflop in one of my registers. Now I put in another one for the partner chip. Wave functions spread throughout space, which is why, I think, the product integral is independent of where the two flipflops are relative to one another. At least that's true for the simplified solution. Now if there's a force between them then the phase of the wave function would vary periodically in the direction of the force. The odd thing is, the bit patterns that create a force have their phases at all different angles. I write out the wave functions for the other bits in the two registers. Well, actually if you look at it, the phases don't add up to zero; there's a nonzero sum same as if... Let's assume that each bit has an adjustable phase. If I start with identical wave functions, then impose the periodic variation consistent with there being a force... All I have to do, to control the force, is put a phase shifter on each bit and somehow get them set at the right angles! And Simba was exactly right: if the phase is rotated in time, as it would be if the chip were made to move against the force, or if I programmed a continuous rotation into the phase shifter, energy would be transferred into or out of the power supply! It's the last piece of our puzzle. I get right to work designing the more complex register cells for a new test chip, considerably more complex due to the programmable phase shifters and the hand-drawn Penrose tiling. Simba doesn't like all-nighters, but this one is going to be worth it.
And that cockamamie acoustic coupling scheme for the communicator chip: the right way to do it is through CQMT energy transfer. Come Monday I'll have to tell the development team for that chip that they're going to see a big change.
| Next | Previous | Contents |