“The ATHENA research group is working on the spam-filtering problem by running a huge distributed metacognition app that’s intended to pick holes in the spammers’ fake social networks.”
MacDonald magicks up a big diagram in place of the graphs; it looks like a tattered spider-web. “Here’s a typical social network. Each node is a person. They’ve got a lot of local connections, and a handful of long-range ones.” Thin strands snake across the web, linking distant intersections. “Zoom in on one of the nodes, and we have a bunch of different networks: their email, chat, phone calls, online purchases…” A slew of different spider-webs, cerise and cyan and magenta, all appear centred on a single point. They’re all subtly different in shape. “Spambots usually get their networks wrong, too regular, not noisy enough. And we can deduce other information by looking at the networks, of course. You know the old one about checking the phone bills for signs that your partner’s having an affair, right? There are other, more subtle signs of—well, call it potential criminality. Odds are, before your partner snuck off for some illicit nookie, there was a warm-up period, lots of chatter with characteristic weighted phrases—we’re human: We talk in clichés the whole time, framing the narrative of our lives. Or take some of the commoner personality disorders: pre-ATHENA, we had diagnostic tools that could diagnose schizophrenia from a sample of email messages with eerie accuracy. Network analysis lets us learn a lot about people. Network injection lets us steer people—subject to ethics oversight, I hasten to add—frankly, the possibilities are endless, and a bit frightening.”
“Can you give me an example of what you mean by steering people?” Kemal nudges.
“Hmm.” MacDonald’s chair squeals as he leans back. “Okay, let’s talk hypotheticals: Suppose I’m wearing a black hat, and I want to fuck someone up, and I’ve paid for a command channel to Junkbot.D. First, I build a map of their social connections. Then I have Junkbot establish a bunch of sock puppets and do a friend-of-friend approach via their main social networks—build up connections until they see the sock puppets’ friend requests, see lots of friends in common, and accept the invite. Junkbot then engages them in several conversation scripts in parallel. A linear chat-up rarely works—people are too suspicious these days—but you can game them. Set up an artificial-reality game, if you like, built around your victim’s world, with a bunch of sock puppets who are there to sucker them in to the drama. Finally, you use the client-side toolkit to hire some proxies—neds in search of the price of a pint—who’ll hand your target a package and leg it, five minutes ahead of your colleagues, who have received an anonymous tip-off that the quiet guy living at number seventy-six is a nonce.”
Kemal is rapt, listening intently. He nods, perhaps once every ten seconds. MacDonald has got him on a string with this spiel. You look back at MacDonald. “You wouldn’t dream of doing that,” you say.
MacDonald grins and nods. “Indeed not. Morality aside, it’s stupid small-scale shit. What’d be the point?”
You peg him then. He’s not your typical aspie hacker, and he’s not a regular impulse-control case. MacDonald’s the other, rare kind: the sort of potential offender who does a cold-blooded risk-benefit calculation and refrains from action not because it’s wrong, but because the trade-off isn’t right. You won’t be seeing him in the daily arrest log anytime soon, because he kens well the opportunity cost of a decade in prison: It’d take a bottom line denominated in millions to lure him off the straight and narrow. But if he sees such a pay-off…
“What do you use ATHENA for?” you ask, bluntly.
“Right now we’re tracing spammers. ATHENA can scope out the fake networks: It can also tell us who’s running them.” There’s something about MacDonald’s body language that puts you on red alert. Something evasive. “ATHENA then probes the spammers to determine whether they’re human or sock-puppet. We’re working on active countermeasures, but that’s not green-lit yet; I gather there’s a working group talking to some staff at the Ministry of Justice about it, but—”
“What kind of active countermeasures?”
“Spoiler stuff, but more active than usuaclass="underline" using their own tools against them. You know it’s an international problem? Crossing lines of jurisdiction—a lot of them live in countries that aren’t signatory to or don’t enforce anti-netcrime treaties. So we’re examining a number of tactics that’d need to be approved by a court order before we could use them. So far it’s just theoretical, but: reverse-phishing the spammers to grab their control channels and shut down the botnets. Fucking with their phishing payloads to make them expose their real identity so you folks can arrest them. Stealing their banking credentials and applying civil-forfeiture protocols. Using their ID protocols to fuck with their personal lives—hate mail to the mother-in-law, that kind of thing. Having their computer report itself as stolen. In an extreme instance, ask the USAF to send a drone to zap them.”
“Uh-huh.” You glance down and try to look as if you’re making notes, so that he can’t see your face. One by one, the alarm bells are going off inside your head. “But you haven’t done any of this yet.”
“No.”
“But?”
“ATHENA is an international effort.” MacDonald leans forward on his elbows, fingers laced before him. “We are just academic researchers. We’re trying to find a way to, shall we say, enforce communal standards without turning the corner and ending up with a panopticon singularity, ubiquitous maximal law enforcement by software—nobody wants that, so we’re looking for something more humane. Crime prevention by automated social pressure rather than crime prosecution by AI. But… once you get into that territory? People don’t all agree on what constitutes crime, or moral behaviour. Some of our associate members live in jurisdictions where there are melted stove-pipes between academia and government, or intelligence. And I canna vouch for what those third parties might do with our work.”
ANWAR: Bluebeard
As soon as you open the front door, you know something’s not right.
“Honey? I’m home…”
It’s like that inevitable, deterministic scene in every horror video you’ve ever lost two hours of your life to: the dawning sense of wrongness, of a life unhinged. From the subtle absence of expected sounds to the different, unwelcome noise from upstairs in the bedroom, all is out of order.
“Hello?” you call up the stairwell.
There’s no reply, but you hear footsteps like a herd of baby elephants on the landing. Angry footsteps. Your stomach clenches. They are Bibi’s angry footsteps, and now you know what is wrong: All that remains is to find out why.