by Ray Mazza
Curiously, the scientist from Intelligentech was only interested in these two slides. They had talked about these “logic ganglia,” and how their formation might vaguely resemble the evolution of nerves in the earliest organisms.
Trevor had toyed with evolving them by looping their outputs back to inputs and to their fitness ranking algorithm so the ganglia would modify and score themselves, hoping something meaningful would result. It felt very much like swimming around in the middle of the vast Pacific at night, hoping to wash up on a glorious island of significance.
Trevor’s colleagues had rejected these ideas outright, failing to see any promise. Although Trevor never found his island, the man from Intelligentech seemed as close to ecstatic as an expressionless scientist could get, and after two hour’s discourse, offered Trevor a job on the spot.
A few days later, Trevor accepted. In his earliest moments at the company, they asked him to continue his circuit growth algorithms, and told him the plan was to use them in some form of optical recognition. Lab coats would hand him a sheet of inputs and request ganglia circuits that approximated a second sheet of outputs.
When he completed one assignment, the lab coats would take the circuit design with a nod and hand him a new set of inputs and desired outputs. Again, he would pore over variables while evolving populations of circuits. Each time that the lab coats handed him a new request, they would also urge him to refine his approach.
He optimized his code, and over time was able to arrive at solutions more and more quickly. The first one took close to a month. After a few more, they were taking a couple weeks. Eventually, he was finding solutions within days, and then mere hours.
Then one day they took his code, thanked him, and moved him onto something else. No explanations.
Returning his mind to the present, Trevor glanced again at the folder on the desktop. It was numbered with a version almost twice as high as where he’d left off way back then.
You will notice that the program we hired you for is here. I cannot tell you what a fantastic job you did; it was a wonderful contribution. After the handoff, my personal staff continued working on it. We needed to adapt it to our purposes and for our unique hardware, which is why we took it away from you. Your algorithm, however, has remained untouched.
If it was so fantastic, why hadn’t they brought him into Damon’s inner circle? Trevor no longer bought the story they gave him about an optics application… it sounded too unimportant. From the timeline Damon had given him, Day Eight already had its earliest human simulations when Trevor began working there. This was related.
What I need now is for you to find a bug in your code, and fix it. It is a bug that has been there since close to the project’s inception. It is an insidious bug that has lain dormant, and only has reared its head in the past few weeks. At first, we didn’t even know it was a bug. We just thought it was a shortcoming of the program. After Kane worked with it for nights on end, he found an inconsistency.
However, none of my team were able to fix it, because they couldn’t find the cause. Your math, although extremely complex, is understandable. What we had not been able to understand wholly was your approach. The aspects of it that we can grasp are simply jaw-dropping.
They said that Einstein was decades ahead of his time. Some believe that if he hadn’t penned his paradigm-shifting theory of the equivalence of mass and energy, E = MC2, that we still wouldn’t have it today. Instead, we’d just be wondering why the atomic clocks in our satellites wouldn’t stay consistent with those on Earth, and why our GPS devices couldn’t be more accurate.
I believe your approach to self-modifying intelligence – and I do mean intelligence, not just circuits – is, like Einstein’s theory of relativity: ahead of its time. Since we could not fully comprehend it, I ask that you find this bug, understand it, and fix it. We could not.
The reason we noticed this bug when you did not was that our modifications and our hardware increased the speed of your program to an extraordinary level. Only then, did it become apparent.
I am confident that you will see what I mean. You can fix it.
This is of the utmost importance, Trevor.
Damon.
PS – Please do not remove this or any future notes from the room. Simply leave them in the wastebasket, I will handle their elimination.
“Wow.” Both excited and in disbelief, Trevor sat and soaked up what he’d read. By this point, it was obvious that most of the projects that Day Eight’s “normal” employees worked on were related to the human simulations – either the work was to directly enhance them, or it was an implementation of ideas the simulations themselves had developed.
But the thought of his own program having been an accomplishment on the level of Einstein was far-fetched. Novel, maybe, but his approach hadn’t been so far beyond logical reasoning that it surpassed the comprehension of the lab coats at Day Eight. Had it?
Trevor checked on Allison, who was happily watching Alice in Wonderland and munching on her bowl of rainbow popcorn.
Looking back to the computer, he took a deep breath, stretched his hands, and jumped in. It only took half an hour or so before he felt at home once again in his code.
The program was still set to work on the same input the lab coats had last given him years ago. Without changing anything, Trevor pressed the ‘F5’ key to run it. At first, he thought it hadn’t worked, because the program displayed its usual charts, graphs, and statistics, but none of them were updating… like the program had frozen. Was this the bug?
Trevor looked more closely at the screen. The top of it read:
Solution Found.
Total Generations: 1,922,517
Printed to file “Circuit 10-29-2012.1.cct”
What? The program completed? Already!? It wasn’t updating because it had finished instantaneously. He’d expected it to take a few hours.
Holy crap. This computer is… “Fast” would have been the understatement of the year.
Instinctively, he opened the computer properties to check the processor. It only displayed:
Processor: Unknown Model
Trevor shut the computer down and removed the finger screws from the oversized case, then slid the side panel off to expose its internals. A billow of heat brushed past his face. At first, he didn’t quite understand what he saw. The inside of the case was laden with rusty cobwebs. On closer inspection, he realized he was looking at a nearly microscopic network of copper-colored tubing.
Not thinking, Trevor reached out to touch it. As his fingertip made contact with a portion of the webbing, a searing pain shot through his hand. He yelled out and sprinted to the bathroom to hold his finger under cold water. How could he have been so stupid? It had been like touching a soldering iron. The tubing must have been an extensive form of a heat sink used to draw heat away from the most active components so they didn’t fry themselves.
Fortunately, the medicine cabinet held both bandages and antibiotic ointment. He dressed and bandaged his finger and looked at it, thankful he didn’t have to do a lot of typing just yet… he would need to locate the bug first.
Back in the sanctuary, Allison was upset after hearing him yell out. He calmed her down and she settled back into her movie.
Trevor again peered inside the computer case, still curious, but more cautious. There were some standard computer components which didn’t interest him. He turned his attention back to the tubing. Most of it fed into a golf-ball sized sphere in the center of the case that had a silver-purple iridescence to it under the gentle light of the room. That – however it functioned – must have been the processor. Trevor guessed it was spherical because that was the most efficient way to pack material into a space… although all of today’s computers still hadn’t progressed beyond flat processors.
Satisfied there was nothing more to learn simply by looking at it, he shut the case and rebooted the machine, excited about the prospect of running his circuit evolution program at blin
ding speeds.
Trevor toggled an option that would keep it running indefinitely since it wasn’t trying to find a specific answer. It would just “live” and create populations of larger and more complicated ganglia of circuits.
He ran the program again. The screen came to life, with charts jumping around, lines rising and falling on graphs, and numbers growing, shrinking, growing again. After six seconds, it halted. The screen displayed:
Population insufficient for reproduction.
Error: Gene mismatch
Total Generations: 7,319,008,024
Significant circuits printed to file “Progression 10-29-2012.1.cct”
File truncated at maximum size.
Gene mismatch? It meant that nothing in the population of circuits could be represented genetically in such a way that anything could reproduce. It was like having chromosomes that didn’t line up… like trying to mate a chicken with an octopus…
The output file had two gigabytes of the most recent data, far more than he could reasonably wade through.
Nonetheless, Trevor scanned it, and to his surprise he immediately noticed a pattern. The circuits had gone from enormous, to, well, practically nothing. The final circuits were so small that most couldn’t even be represented in a single gene.
Somewhere along the line, the program had convinced itself that smaller was better… more efficient maybe? Once this idea – this meme – entered the scoring function, it couldn’t be stopped. Ultimately, the population wore itself down to nothing… it committed suicide.
That was the bug.
Now, how did it relate to the human simulations?
~
Trevor made no progress in his search for a fix the remainder of the day. Before he left, he wrote a short response to Damon on a fresh sheet of stationery:
Damon – Understand the bug. Next step fixing it. Why not have simulations do it? Burnt myself on computer. What is the sphere inside? Also, why was my memory stick stolen and brought to the office? Sorry, my handwriting sucks. Prefer phone. Trevor.
He positioned his note on the desk, tucked Allison into bed, and headed out into the shadowy dusk just as some winter clouds began to dissipate and reveal the moon.
Chapter 26
What the NSA Doesn’t Tell You
The next day, Trevor picked up a few movies for Allison. He chose animated Disney movies because they were the most vibrant. When he showed them to her, she jumped up and kissed the camera.
“Yay! Thank you thank you thank you Trev-ee! Let’s watch one! Robin Hood Robin Hood!” She clapped as Trevor put it in. He raised the back of his left hand to his cheek and felt its warmth… he was blushing. His sister, Amy, used to call him “Trev-ee” before she could fully pronounce his name.
It was strange to him that a simulation could make him feel this way. She was just a bunch of electricity moving around in a box somewhere.
Yet… though she may have been a bunch of ones and zeros, she was a bunch of ones and zeros that gave him the nickname “Trev-ee.” And when she’d leaned in to kiss the camera, he’d noticed her neck – he’d seen her jugular vein pulsing steadily with her heartbeat. A genuine heartbeat. If that didn’t make her real, what would?
And if she was just a bunch of electricity zipping around a hunk of metal in a box, then what was he besides a modicum of current firing around in a slab of meat? That’s all the brain was… three pounds of water, fat, and protein with a dash of electrical activity.
Trevor grasped the flesh on his forearm between two fingers, inspecting it, studying the resultant wrinkles, consciously experiencing the sensation of his own touch. Life at all was a miracle, let alone life in a computer.
“Trev-eeeeeee,” she said, snapping him out of a trance. “Hillary asked about you on the phone last night.” She said it in a sing-songy voice while pointing at him, waving her arm in a little circle.
“She did? What’d she say?”
“Oh… nothing.” Allison smiled.
“C’mon, tell me!” he laughed.
Allison just shrugged.
“That’s all right,” said Trevor. “I didn’t want to know anyway.” He diverted his gaze, feigning utter lack of interest.
“She just told me to be a nice princess for you because you deserved it and wanted to know if you talked about her.”
“She did? Well, you certainly have been a nice princess for me,” Trevor smiled. “I think it’s nice you get to talk to her on the phone.”
“And I get to see her every day, too!” Allison corrected, grabbing a framed photo off her dresser. “Look, I keep her right next to Daddy.” She held the photo up so Trevor could see. The confines of the shiny metal frame held a girl in her late twenties with a waterfall of wavy brown hair, knowing green eyes, and a smile as pure and radiant as her voice had been.
“She’s pretty, huh,” said Allison.
“Just like Maid Marian. Want me to put on Robin Hood for you to watch now?”
“Yay!” Allison mimed the action of shooting a bow and arrow, then pretended she got shot in the chest. She fell to the ground, clutching a red magic marker “protruding” from her sternum.
Trevor chuckled and put the movie in. “I think you’d make a great actress someday.”
Her corpse on the floor smiled and nodded, then went back to being fully dead, complete with tongue hanging over the side of her mouth. Once the opening credits began, she hopped up and did her best to whistle along.
Trevor seated himself in front of the desk and found Damon’s next note spanning pages:
Trevor – Glad you found the bug. Now fix it.
I still do not know about your memory stick. It was gone when I returned to work, and I haven’t been able to coax the information out of Kane without giving us away. It will take time.
I choose not to communicate by phone because I no longer trust the airwaves. I mentioned before that we have ties with the NSA. You don’t have a secure cell phone, so it is safe to assume they can – and likely do – listen in on any of the standard phone transmissions we have. After the internet surge they would be foolish not to.
Now, you’ve been waiting long enough; it is time I explained our untold history and technology in detail.
About 26 years ago – back in 1986 – Day Eight started as a spin-off from the national intelligence community. Our company was then called Intelligentech, as I believe it still was back when we hired you. We only had a score or so of researchers, mostly as transfers from both the NSA and CIA.
We were created as a spin-off because, of all things, funding. The NSA had a project they needed to devote massive resources to, and we would get more total funding as two separate entities than as one. We were a public company, which allowed us to easily get contributions from the government’s black budget laundered through various donors and venture capital firms. (Every time the government “loses” billions of dollars – like the twenty-five billion that evaporated into thin air from the Iraq war budget and cannot be accounted for – it’s not actually “lost,” it’s funneled to a spread of black budget projects like this one. Any project that has the potential to forward America’s efficacy of warfare is especially funded. Lucky us.)
Our task was to improve their computer technology, plain and simple. Yet it wasn’t so simple.
Now then, if you were to go interview with the NSA, what they’d tell you is that the pay isn’t great – after all, it’s just a government salary – but that you wouldn’t want to work for them for the money. Instead, you’d come for the opportunity of a lifetime: they’d tell you that you’d be working with technology that’s ten years ahead of the marketplace. Ten years! In the computing world, as you know, that’s multiple generations.
Ten years ago, you could burn the entire contents of your computer onto a few CDs… now, with hard drives as vast as they are, you’d need a few thousand CDs. Moore’s law predicted that we’d be able to double the number of transistors in our computer chips every two years, a
n exponential increase in processing power… and we’ve actually been outpacing it! We’ve gone from ten million transistors on a computer chip to ten billion in the last decade.
So imagine what that would mean if the NSA’s computers were ten years ahead of the public! A nerd’s wet dream. And plenty of nerds they got.
Like I said, they’ll tell you they have technology ten years ahead of the industry. Now, what they won’t tell you, is that they’re lying. The NSA’s technology isn’t ten years ahead of the industry. No. These days, their technology is actually a full sixteen years ahead, or more. It’s outpacing the industry at an accelerating rate. And they have supercomputers built out of this advanced technology.
I know this because we helped build them.