“What are you looking for?” asked Sarkar.
“Anything,” said Sandra, entering the room. “Everything.” Then, to Jorgenson: “Any problems?”
“Not so far. He was already logged in, so we didn’t need to crack the password file.”
Sarkar was edging away from the group toward a console on the other side of the room — a console with a microphone stalk sticking up from it.
“Login,” said Sarkar in a low voice, then, without waiting for the prompt, “Login name Sarkar.”
“Hello, Sarkar,” said the computer. “Shall I terminate your other session?”
Sandra Philo had come up behind him, the rounded front of her stunner pressing into the small of his back. “Don’t do that,” she said simply. She reached over to the console and flicked off the switch marked “Voice Input.”
At that point, Kawalski, the liaison officer from York, appeared at the entrance to the room. “They’ve got a barber’s chair upstairs,” he said generally to the group, then, looking at Sarkar, “You give haircuts here?”
Sarkar shook his head. “It’s actually a dentist’s chair.”
Jorgenson spoke without looking up. “Scanning room, no doubt,” he said. Then, to Sarkar: “I enjoyed your paper in last month’s Journal of AI Studies. I’ll want to search that room next.” He went back to typing commands on his black-and-silver keyboard.
Sarkar sounded exasperated. “If you would just tell me what you’re looking for…”
“Damn,” said Jorgenson. “There are several encrypted banks here.”
Sandra looked at Sarkar. “What’s the decryption key?”
Sarkar, feeling perhaps that he had some measure of control at last, said, “I do not believe I’m obligated to tell you.”
Jorgenson got up from the stool. Without a word, a second analyst sat down on it and began typing commands.
“Doesn’t matter,” said Jorgenson with a shrug. “Valentina was with the KGB, back when there was such a thing. There’s not much she can’t crack.”
Valentina popped a new datacard into the card slot, and typed furiously with two fingers. After several minutes, she looked at Sarkar, her face full of disappointment. Sarkar brightened visibly — perhaps she wasn’t as good as Jorgenson had said. But then Sarkar’s heart fell. The disappointment on her face was simply that of someone who’d been hoping for a good challenge, and had failed to find it. “The Hunsacker algorithm?” she said in a heavily accented voice, shaking her head. “You could have done better than that.” Valentina pressed a few more keys and the screen, which to this point had been filled with gibberish, was replaced with English source-code listings.
She got up. and Jorgenson went back to work. He cleared the screen, then replaced Valentina’s datacard with another of his own. “Initiating search,” he said. The screen filled with a multicolumn list of two hundred or so terms in alphabetical order.
“There’s massive storage online here,” said Jorgenson, “under a variety of compression schemes. It’ll take a while to hunt through it all.” He got up. “I’m going up to look at that scanning room.”
Peter had an evening board meeting at North York General today, and rather than waste the morning fighting the telephones at the office he decided to do some work from home. But he was having trouble concentrating. Sarkar had said he’d have the virus finished today, but Peter still felt he should be doing something himself. Around ten-thirty, he logged into Mirror Image, hoping to see if he could fathom how the sims had gotten outside.
After dialing in, he issued the WHO command to see whether Sarkar was also online — Peter wanted to send him an E-mail hello. He was indeed. Peter then issued WHAT to see what sort of activity Sarkar was doing; if it was a background task, he probably wasn’t actually sitting at the terminal, and so the E-mail would be a waste of time.
WHAT reported the following:
Node | User | Logged in at | Task
002 | Sarkar | 08:14:22 | text search
Well, a text search could be either background or foreground. Peter had high-level supervisory privileges on Sarkar’s systems. He called for an echoing of the task at node 002 on his own monitor. The screen filled with a list of search terms, and a constantly updated tally of hits. Some, such as Toronto, had hundreds of hits so far, but others…
Christ, thought Peter. Look at that…
Sarkar was searching for “Hobson” and “Pete*” and “Cath*” and…
Peter tapped out an E-mail message: “Nosy, aren’t we?” He was about to send it when he noticed the full search parameters in the status line: “Search all systems; within each system, search all online and offline storage and all working memory.”
A search like that could take hours. Sarkar would never order something like that — he was too well organized not to have at least some idea how to narrow the search.
Peter glanced at the other search terms.
Oh, shit.
“Larsen,” “Hans,” “adultery,” “affair.”
Shit. Shit. Shit. No way Sarkar would be doing a search like that. Someone else was inside the system.
Node 002 was the AI lab at Mirror Image. Peter swung his chair to face his phone and hit the speed-dialer key for there.
The phone rang in the AI lab. “May I get that?” asked Sarkar.
Sandra nodded. She was watching the screen intently. Lots of hits on the common words — “affair” had over four hundred so far — but none on Hobson or Larsen.
Sarkar moved across the room to the videophone and hit the ANSWER key.
The Bell Canada logo backflipped away. Peter saw Sarkar’s face, looking worried.
“What’s — ” said Peter, but that’s all he said. In the background, over Sarkar’s shoulder, he saw a profile of Sandra Philo. Peter broke the connection at once.
Philo there, at Mirror Image.
A raid. A goddamned raid.
Peter looked at his screen, slaved to node 002. Still no hits on “Hobson.”
He thought for a second, then began tapping keys. Peter spun off a second session under Sarkar’s login name, using the password he’d heard Sarkar use before. He then accessed the diagnostic-tools subdirectory and called up a file listing. There were hundreds of programs, including one called TEXTREP. That sounded promising. He called up help on it.
Good. Exactly what he needed. Syntax: search-term, replacement-term, search parameters.
Peter typed “TEXTREP / Hobson / Roddenberry / AI7-AI10” — meaning change all occurrences of “Hobson” to “Roddenberry” on artificial-intelligence systems seven through ten.
The program set to work. It was a much smaller search — only one term — and a much narrower area to search — only four computers instead of the hundred or more that Philo was currently examining. With luck, it would make all the substitutions before it was too late…
The console beeped, signaling its task was complete. Jorgenson was back, having found nothing of interest in the scanning room. He looked at the screen, then at Sandra. Thirteen hits for Hobson. Sandra pointed at the tally. “Display them in context,” she said.
Two appearances of the word in an online dictionary entry for “Hobson’s choice.”
A user-ID file, equating “fobson” with Peter G. Hobson.
A computerized Rolodex with home and business addresses for Peter Hobson.
And nine more references, mostly within copyright notices, to Hobson Monitoring Ltd. as parts of various pieces of scanning software.
“Nada,” said Jorgenson.
“He’s got an account here,” said Sandra, turning to Sarkar.
“Who does?” he said.
“Peter Hobson.”
“Oh, yes. We use some programs made by his company.”