Payback

120

Rubens bent over the console at one of the stations at the rear of the Art Room, watching the screen of an analyst who was monitoring and assisting police activity in Texas. He had a feed from a Customs Service aircraft, supplying an overhead view as an emergency response team of FBI and border patrol agents surrounded a tractor trailer south of Houston.

A series of links with Homeland Security gave the Art Room access to police networks as well as the FBI. The Texas Highway Patrol had found a truck with Mexican plates apparently abandoned at the edge of an auto wrecking yard. The plates were not on the watch list the Art Room had developed, but otherwise the truck looked very much like what they were looking for.

“No audio?” said Rubens.

“They’re using standard radios,” said the analyst. “We can get to the commander through the Homeland Security line and vice versa, but we don’t have real-time communications with the people on the scene.”

Rubens watched as the emergency response team — essentially a SWAT squad — cautiously approached the truck. Rather than risking a booby-trapped door, they climbed to the top, where they began cutting a hole to get in.

What if he was wrong? What if the warhead didn’t exist or was back in Peru? Babin certainly could be there, hiding somewhere in the Amazon with Túcume.

That mistake Rubens could live with. Far worse was a mistake that led to the destruction of an American city.

“Won’t be long now,” said the analyst.

“You have a list of stolen vehicles from this area?”

“As soon as the alert came in, I got it. I went all the way back to the border, tracing the route. There are only a dozen.”

“Only a dozen?”

“Not many people steal trucks, I guess.”

“Expand the search to include any sort of vehicle, anything large enough to hold a crated weapon. Sixty-six inches,” added Rubens. “Approximately.”

“Are you praying?” asked Johnny Bib.

“Uh, just stretching my back,” said Robert Gallo, twisting up from the floor. Though perhaps prayer would not have been inappropriate — he was having a very hard time nailing down any sort of indication that the container truck had made it into the U.S., let alone where it was going.

“Nothing in the state databases that you can use to find what sort of truck it is?” asked Johnny Bib, coming over to look.

“The states don’t keep very close tabs of trucks,” said Gallo. “I’ve, like, checked through the lists of, you know, road stops and stuff, those weigh-in things. Looked for mismatches and stuff. But nothing jumps out. This kind of isn’t my thing, you know? Searches? And like, what would I check? Vehicles most unlikely to be stopped?”

“I hope that’s not going to become your slogan,” said Johnny Bib. “Defeat.”

“It’s not defeat.”

Johnny began to shake his head back and forth without saying anything. He looked like a kid’s wind-up toy gone berserk.

“What if I looked for the target rather than the truck?” said Gallo. “What I was thinking was, check everything we have related to Babin, right? And then see if there are any links. Jeez, Johnny, could you stop? You’re giving me vertigo.”

“Exactly,” said his supervisor, without explanation. And he turned and walked from the room.

Puzzled, Gallo returned to his computer. He began another set of searches, this time a keyword search of NSA South American intercepts over the past week using “Stephan” rather than “Babin” as a keyword. Not surprisingly, he got about ten thousand hits.

He was about to run a Dredge search on the hits when he noticed that one of the entries on the last page had a name very similar to Babin — Baben. He looked and confirmed that it was simply a misspelling by the computerized transcription programs, which often relied on phonetic spellings and best choices if the intercept data was unclear.

The search system was programmed to find near misses in spelling. But did that tool apply when you were searching in foreign character sets?

Or rather, had the tool been in place three years before when the Babin intercepts were being compiled in Russian and the Cyrillic alphabet was being used?

As a matter of fact, it had. But to keep the tool from matching every possible word, Gallo realized as he played with it, it assumed that the first two letters were phonetically correct. So if “Ba” Б (the sound that began “baby”) was entered incorrectly as “Ba” (the sound, in Russian, that began “vacation”), the tool rejected the string as a match. That made sense in most cases, but in this particular instance, a human operator who was a native English speaker could easily choose “B” by mistake rather than “E” and the error would not be picked up.

“This isn’t my thing,” mumbled Gallo as he retrieved a list of old databases to apply the new search term to.