Blind Lake

Eyeball Alley was a replica of the installation at Crossbank, which Chris had also visited. Structurally identical, at least. The differences were all in the details: names on doors, the color of the walls. Some halfhearted seasonal decor had lately been installed, a festoon of green and red crepe over the cafeteria entrance, a paper wreath and menorah in the staff library.

Charlie Grogan wore a pair of glasses that showed him things Chris couldn’t see, little local datafeeds telling him who was in which office, and as they passed a door marked ENDOSTATICS Charlie had a brief conversation (by throat microphone) with the person inside. “Hey there, Ellie… keeping busy… nah, Boomer’s fine, thanks for asking…”

“Boomer?” Chris asked.

“My hound,” Charlie said. “Boomer’s getting on in years.”

They took an elevator several stories down, deep into the controlled environment of the Alley’s core. “We’ll get you suited up and into the stacks,” Charlie said, but when they approached a wide door marked STERILE GEAR there was a flashing red light above it. “Unscheduled maintenance,” Charlie explained. “No tourists. Are you prepared to wait an hour or so?”

“If we can talk.”

Chris followed the chief engineer back to the cafeteria. Charlie had not had lunch; nor, for that matter, had Chris. The food on the steam tables was the same food they served back at the community center, the same prefabricated rice pilaf and chicken curry and wrapped sandwiches delivered by the same weekly black truck. The engineer grabbed a wedge of ham-on-rye. Chris, still a little chilled by his walk to the Alley, went for the hot food. The air in the cafeteria was pleasantly steamy and the smell from the kitchen rich and reassuring.

“I go a fairly long way back in this business,” Charlie said. “Not that there are any novices at the Lake, apart from the grad students we cycle through. Did Ari tell you I was at Berkeley Lab with Dr. Gupta?”

Tommy Gupta had done pioneering work on self-evolving neural-net architectures and quantum interfaces. “You must have been an undergraduate yourself.”

“Yup. And thank you for noticing. This was back when we were using Butov chips for logic elements. Interesting times, though nobody knew exactly how interesting it was going to get.”

“The astronomical application,” Chris said, “you were in on that, too?”

“A little bit. But it was all unexpected, obviously.”

In truth, Chris didn’t need this playback. The story was familiar and every general astronomy and pop-science journalist of the last several years had recounted some version of it. Really, he thought, it was only the latest chapter of mankind’s long ambition to see the unseeable, embellished with twenty-first-century technology. It had begun when NASA’s first generation of spaceborne planet-spotting observatories, the so-called Terrestrial Planet Finders, identified three arguably earthlike planets orbiting nearby sunlike stars. The TPFs begat the High Definition Interferometers, which begat the greatest of all the optical interferometer projects, the Galileo Array, six small but complex automated spacecraft all operating beyond the orbit of Jupiter, linked to create one virtual telescope of immense resolving power. The Galileo Array, it was said at the time, could map the shapes of continents on worlds hundreds of light years away.

And it had worked. For a while. Then the telemetry from the Array began to deteriorate.

The signal faded slowly but relentlessly over a period of months. After an intensive review NASA pinpointed the source of the failure as a few lines of bad code so deeply embedded in the onboard Galileo architecture that they couldn’t be overwritten. This was a risk NASA had assumed from the begi

“NASA didn’t have an O/BEC processor back then,” Charlie said, “but Gencorp offered them time on their unit.”

“You worked at Gencorp?”

“I baby-sat their hardware, yeah. Gencorp was getting good results doing proteinomics. You could do the same stuff with a standard quantum array, of course. Engineers used to think of the O/BECs as u

“Spooky?”

“Unexpected. Counterintuitive. Anybody who works with adaptive self-programming will tell you it’s not like ru

Charlie trailed off, as if he realized he’d overstepped the bounds of engineering propriety. He doesn’t want me quoting this, Chris thought. “So you went to NASA with the O/BEC processor?”

“NASA ended up buying a few platens from Gencorp. I was part of the package. But that’s another story. See, basically, the problem was this: as the Galileo Array’s output got fainter, it was increasingly harder to separate the signal from the noise. Our job was to extract that signal, hunt it down, subtract it from all the rest of the random radio garbage the universe belches out. People ask me, ‘So how’d you do it?’ And I have to tell them, we didn’t do it, nobody did it, we just posed the problem to the O/BECs and let them generate tentative answers and bred them for success… hundreds of thousands of generations per second, like this big invisible Darwinian evolutionary race, survival of the fittest, where the definition of ‘fittest’ is success at extracting a signal from a noisy input. Code writing code writing code, and code withering and dying. More generations than all the people who ever lived on the earth, almost more generations than life on the earth. Numbers complexifying themselves like DNA. The beauty is the unpredictability of it, you understand?”

“I think so,” Chris said. He liked Charlie’s eloquence. He always liked it when an interviewee showed signs of passion.

“I mean, we made something that was beautiful and mysterious. Very beautiful. Very mysterious.”

“And it worked,” Chris said. “Signals out of noise.”

“Whole world knows it worked. Of course, we weren’t sure of that ourselves, not while it was happening. We had a few what we called threshold events. We’d almost lose everything. We’d have a good clean image, then we’d start to lose it, almost pixel by pixel. That was the noise wi

“Until one day?”

“Until one day a man in a suit came into the lab and said, ‘Boys, we got confirmation from upstairs, the Array just stopped broadcasting altogether, shut down entirely, you can get ready to close up shop and go home.’ And my boss at the time — that was Kelly Fletcher, she’s at Crossbank now — she turned away from her monitor and said, ‘Well, that may be, but the fact is, we’re still making data.’”

Charlie finished his sandwich, wiped his mouth, pushed his chair away from the table. “We can probably get into the stacks now.”