other side without being seen. Lia did the same on the third floor, scrambling onto the roof.
She retrieved her gear, unloading and stowing the gun in its case and then pulling the strap over her rucksack. As she rose, there was a noise on the roof behind her. Lia swung around, not sure if it was Dean; a shadow appeared across the way near the fire escape ladder, then disappeared. Lia sprinted after it, pulling a small Glock handgun from her waistband as she crossed the roof. She reached the edge in a few bounds and bolted over the side, jumping down the ladder to the next landing, where the figure had just descended. The shadow tried to escape through the window, but Lia grabbed hold, pulling it out and sticking the pistol in its face.
The face of a nine-year-old-boy, shaking like a leaf, eyes like acorns.
She stared into them, saw his fear.
Lia pushed the boy away, then went down to the street. Dean was waiting on his bike.
“What were you doing?” he said.
“Making mistakes,” she told him brusquely, brushing past.
46
Rubens sat in one of the empty seats in the Art Room’s theater-style layout, staring at the screen saver on the computer monitor in front of him. A multicolored object floated across the screen, morphing into different forms as it came to each edge of the viewing area. The changes seemed random but were actually determined by a pair of complex mathematical formulas whose results only
Intelligence gathering worked like that. Seemingly random bits of information had to be stitched together from a wide variety of sources to reveal themselves as something generated by a purposeful formula. This was, in Rubens’ opinion, the reason that those trained in mathematics and cryptography were so good at intelligence work: they knew there were formulas behind what others saw as unrelated information flitting through the universe.
But there was also a danger that truly random events might be misinterpreted as being part of a nonexistent pattern. Or worse, unconnected strands might be erroneously connected, leading the analyst in the wrong direction.
Was that happening now? The man Dean had followed had used an antiquated analog coupling device to send a very short piece of information to a computer bulletin board in France. The technology was so old — it dated from the mid-1980s — that under other circumstances the NSA wouldn’t have bothered paying any attention.
Ninety seconds after the transmission, the weather site on the World Wide Web that Rubens had discussed with Johnny Bib had been rewritten, two minutes ahead of normal schedule.
According to Johnny Bib, the change was doubly significant not only was it outside the normal pattern of updates, but it also didn’t alter a temperature as the other ones had. Instead, it involved a forecast three days ahead.
The change was subtle: “cloudy” became “thunder-storms.” At the regular update time, the old prediction was restored.
Surely this meant something… or was it a simple programming mess-up?
The potential intersection of the two Desk Three missions — one to find missing atomic material, the other to find a chemist who might have made explosives for terrorists — was both tantalizing and frightening.
But was it a pattern or a non-pattern, meaningful or random noise?
“Boss, Lia and Charlie are clear. You want them back here?” asked Telach.
“No. Have them go to France. Tell them…”
He paused, not knowing exactly what to tell them.
“Have them go to Paris, spend the night, relax a little,” he said.
“Relax?”
“If nothing specific develops, they can back up Tommy Karr. Let’s let Lia catch her breath for a few hours. Contact Tommy — I want him to see if he can get more information from Monsieur LaFoote on the Web sites that were used to transmit instructions. Why wasn’t his friend suspicious? Or was he? What else does he know about the e-mails or domains? See if he can find out what computer his friend used to go on these sites, that sort of thing. And check if there were others.”
“He did already try. I’m not sure how technically competent the old guy is, let alone how much he really knows. Tommy worked him over pretty well while they were watching the police search the house. There are two more CD-ROMs, and possibly some information about an account that Vefoures had that we haven’t been able to track down on our own. Otherwise — well, you know Tommy. He’s everybody’s best friend. I don’t know that LaFoote is holding back too much else. Certainly not about computers.”
“Have him try anyway. LaFoote may not know he knows. Where is Tommy?”
“Tommy went back to Paris to dump the information on the disk for us, and LaFoote stayed back in Aux Boix, where he lives. We had a CIA agent follow him back to his house. He was still there last time I checked. I told Tommy to get some rest. He’s supposed to meet LaFoote at nine p.m. their time.”
“All right. Let Tommy get some rest,” said Rubens. “But when he checks in, have him try and move up his meeting with LaFoote.”
“You don’t want the CIA agent to get involved, right?”
Rubens gave her a withering look. “Do you feel he would succeed where Tommy wouldn’t?”
“I just wanted to check. It does seem—”
“Like a random pattern that doesn’t actually cohere?”
“I was going to say wild-goose chase.”
Rubens got up from the console. “What is Johnny Bib doing?”
“His team has the servers under surveillance,” said Telach. “They’re trying to trace anyone who accesses these pages.”
“Let me know if they come up with anything. I’ll be upstairs.”
At the very moment Rubens was asking about him, Johnny Bib was hovering over the shoulder of a cryptologist who’d been shanghaied into serving as a computer operator, trying to track the different queries onto the Web site they thought was being used to send codes through the terrorist network. The tracing work was done by a small program the techies called Spider Goblin 3. It was an extremely sophisticated version of the so-called spy programs used by commercial Web sites to track interested visitors to their Web sites. Spider Goblin 3 spat out a list of different nodes on the net showing where requests for the information had come from; this list was then compared to earlier captures of information to see if there were any matches.
There were literally thousands, as a typical request for a Web site might pass through twelve or more “nodes” on the World Wide Web network as it made its way to and back from the place where the information was kept. The difficulty wasn’t that all this information was impossible to obtain — it was there for the taking. The question was what significance it had. Working on the theory that a computer used for one bad purpose might be used for another — as if it were a car owned by a gang who robbed banks and gas stations — Johnny Bib decided to look at the contents of the computers used in accessing the site. The trick was choosing which ones to examine.
Fortunately, this could be placed in the hands of a complicated mathematical formula, which Johnny and his team had compiled the day before. It had involved a great deal of probability work, and thus much of it had been designed by the team’s statisticians — a fault, Johnny Bib believed, since statisticians were by nature imprecise and even messy. They were willing to live with errors in their work, which they classified as “inevitable.” For Johnny Bib, nothing was inevitable. Unknown, perhaps, but not inevitable.
But even a mathematician sometimes had to compromise by rounding off. Numbers existed in the real world, after all.
“Yo, Johnny Bib, Johnny Bib,” said Tristan Young. “Looky, looky, looky.”
Johnny Bib practically hopped across the room to the console where Tristan was working. Twenty-three years old, Tristan’s real calling was string theory and “real” cryptography, but he had been pressed into service in the
