Sockpuppet: Book One in the Martingale Cycle (16 page)

‘Better a brogrammer than a hogrammer,’ said Colin blithely.

Danielle picked up the keyboard and slapped it hard on the desktop.

‘Oh, cock off, Fatnav, you enormous
bollock
!’

‘Ooh. Eloquent.’

Dani let out a hoarse shout, stood and grabbed a canister of DVDs from the desktop. Ripping off its lid she strode to Colin’s chair and dumped the contents over his head. A hundred shining discs cascaded down the slopes of his body and spilled to the floor with a metallic splash.

Everybody froze, intent on their work. J-R concentrated on Graham’s screen but couldn’t help seeing, in the periphery of his vision, Colin sitting placid and upright. Vindicated.

Danielle shied the empty canister into a bin with great precision and walked past J-R, who made an effort not to flinch. She stopped very close at hand, barely two inches from Graham, and reached for a device on the shelf above – a small beige rectangle with a dangling cable. Her neat right breast brushed the back of Graham’s head. He continued to watch his screen cooly. J-R might have existed on a wavelength invisible to them both.

She glanced at Graham’s screen and spoke quietly.

‘Way to set these fuckwits an example. You’re meant to be analysing Parley data, not proffering.’

‘There’s nothing there,’ said Graham. ‘I told you.’

‘Oh, oh, so you’ve read every line of this three-thirty-two gig file? Pony!’

‘I showed you the keyword results. I don’t care what analytics voodoo you pull. You’re not going to find a thing. Nothing sic_girl read even mentions a politician, let alone Bethany Lehrer. She can’t have said what she said.’

‘Shit me, if only I’d thought of that. She never said it! My troubles are over! Fuck-a-lujah!’

‘I’m just saying, the evidence. It doesn’t make sense to me either. I can only think—’

‘What?’ Danielle’s fury was interrupted, her interest piqued.

‘Well, it has to be sic, doesn’t it?’ said Graham.

‘Of course it’s her.’

Danielle gestured at the screen where sic_girl’s continuity stuttered on.

‘No, it has to be
coming
from her. She must be
thinking
this stuff. Becoming able to.’

Danielle sized Graham up, apparently scenting a tease.

‘No, OK, OK, I know,’ said Graham. ‘Like I say, it doesn’t make sense. But to have my whole team trawling source files for patterns – in data that doesn’t include a single mention of the subject sic_girl spoke about – it’s pointless.’

‘
Pointless
? I’m
pointless
?
Jesus!’

J-R raised an apologetic hand. The two combatants turned to face him.

‘Might I ask?’ The pair were attentive. ‘As I understand it, you’re looking for information sic_girl’s programme may have – ah – read about Bethany?’ Two nods. ‘And you want to find this because sic_girl must have – ah – copied her messages from somewhere?’ Graham nodded. Danielle stared, brow bunched up. ‘So the Parley characters do just
copy
information? I thought they had some kind of – personalities – of their own? How can that be true if all they’re doing is, so to speak, dragging and dropping from elsewhere?’

Danielle and Graham looked at each other and exhaled in unison. J-R’s idiocy had allied them to a common cause. Danielle volunteered herself as spokeswoman.

‘OK. Explainer.’ She spoke as if to a pre-school student. ‘So everything sic_girl says comes from somewhere else – and that includes her personality. That’s the whole point of Parley. It’s a kludge, all right?’ Her hands circled each other. ‘A, a, a, mash-up. When I coded it, I wasn’t trying to create a thing for actual users. It was down-and-dirty, for a bet.’

J-R looked from one to the other, searching for signs he was being mocked.

‘You created Parley – Britain’s most popular digital environment –’

Graham shook his head in mock amazement.

‘Is
that
what it is?’ he said.

‘– for a
bet
?’

‘A bet with this fucking feller here.’ Danielle punched Graham’s shoulder, apparently quite hard. ‘Right?’

Graham did the cub scout thing with his fingers.

‘I cannot deny it.’

‘This hairy tool here bet me I couldn’t beat the Turing Test.’

‘Yes, of course,’ said J-R. ‘The Turing – ah, remind me?’

‘Machine intelligence,’ said Graham.

‘Google it,’ said Dani, simultaneously.

Graham shrugged.

‘It wasn’t so much a bet,’ he said. ‘I just pointed out one night to Dani there was a coding challenge she couldn’t beat. She still hasn’t, as it happens, since she decided to cheat.’

‘I so did not cheat. It’s a valid solution.’

‘Look, I don’t want to be this guy about it, but you should go back and read
Computing Machinery and Intelligence
.’

‘It is, too, valid. Your problem is you haven’t read beyond the first three pages. Intelligence as search, numb-nuts. Eat it.’

J-R had not the slightest idea what this brewing argument was about but he needed to get them back on track. His hand went up again.

‘So, Dani, you wrote Parley to solve this – test?’

Thankfully, Dani was less inclined to argue than to explain her own skill.

‘So OK,’ she said. ‘The thing with the Turing test is, how can you get a machine to answer any question someone asks it? And how do you fool that person into thinking they’re talking to another human being?’

She gave J-R a challenging look. That had not been a rhetorical question.

‘Ah. I suppose –’ he began to extemporise. ‘You’d have to interpret the grammar of the question, look up the words in some sort of dictionary – perhaps the underlying concepts as well, to understand the context.’ There! The man from Westminster does know a thing or two after all! ‘Perhaps a kind of – ah – neural network? A network of related concepts the system could refer to. I imagine it would have to be incredibly sophisticated and—’

Danielle cut in on him with a dismissive
Pfft!

‘D’you have any idea how much storage we’d need to accommodate the knowledge of even a five-year-old child?’ she said. ‘Let alone give it semantics?’

‘No, I clearly do not.’

Graham cut in again:

‘Point is, she didn’t need to when she had –’ he gestured at his screen.

‘– the Internet!’ continued Dani.

The way they finished each other’s sentences: there was surely history here.

‘What Dani did, it’s sort of genius – and I’d only say this when she isn’t listening –’

Dani made a dumbshow of blocking her ears. She was surprisingly sweet when she smiled. Her cheeks became chubby and took on dimples.

‘– what she did was, she set up a system where when you asked it a question, it could search online for related topics being talked about somewhere – and Frankenstein meaningful answers from other people’s conversations, other people’s grammar, vocabulary. For a while it sounded completely unnatural.’

‘Ha!’ said Dani. ‘Like remember one time we asked it,
Who are Queen?
’

‘Yeah.’ Graham grinned at J-R. ‘It said
Queen Elizabeth the Second and Queen Latifah are Queen.
’

Dani gazed at Graham with a kindness J-R hadn’t seen before. Their relationship ran a spectrum from spiky bitterness to time-worn affection, with no middle ground. J-R found it hard to calibrate. She continued.

‘The turning point to be honest was, we got bought by Mondan. Which gave us access to their spiders. It got several orders of magnitude easier overnight. They scrape essentially the entire thing every hour. Much more often for some sites. They keep one of the copies – fully indexed – in the data warehouse at 404 City.’

She gestured towards a metal panel in the wall. A battered film poster –
Iron Man 2
– was taped to it.

‘A full indexed copy of –?’

‘Of the Internet. I shit you not. They scrape basically the whole fucking Internet every hour –’

‘– just so many data, so fresh,’ Graham picked up. ‘Me and the data geeks down here, we had to filter it, whittle it, before Dani could feed the algorithm.’

‘But the sheer scale,’ said Dani. ‘Something changed. There’s still no intelligence in this thing; but when you get that volume of real human data, you get thousands, hundreds of thousands of new rules about what word goes with what word.’

‘What patterns of speech are natural –’

‘– I never taught the algorithm English but it knows how to speak it from the data. It could have been Russian, Japanese –’

‘– and at some point, as the data volumes ramped up exponentially, at some point, it started to surprise us.’

‘And so the thing was, I found out you could give the system a personality – create a Persona – by pointing it at sites and threads with a shared character. One of the unbuilt personalities of the Internet.’

‘Here were characters Dani never designed. Obsessive geek, political activist, doting mother –’

‘– all aggregate personalities we found online. The dramatic personas.’

J-R, struggling to keep up with this contrapuntal routine, chose not correct Dani’s Latin.

‘Something real just, just –’

‘– they emerged,’ said Graham, gesturing again at his screen, ‘So, sic_girl here. Really strong character, one of our most popular, right?’ Danielle gave a short burst of nodding in reply. ‘But she’s just a mash-up of all the drug-dependent borderline bipolar obsessives on the Internet.’

‘Which is by the way a lot,’ added Danielle, still nodding. ‘Excellent source data.’

J-R looked again at the Parley continuity on the screen. This wistful, vulnerable voice, a voice he’d got to know well in the last twenty-four hours, was no more than a patchwork of other voices, from somewhere invisible on the wider network.

‘To someone like myself,’ he said, ‘without the technical background, this seems impossible.’

Danielle nodded and screwed up her face.

‘Honestly?
All
of this seemed like hoodoo when it first happened. We just – got used to it.’

Graham began to quote.

‘
Every sufficiently advanced technology 
. . .’

J-R completed the statement.

‘. . . 
is indistinguishable from magic.
’

Graham toasted him with his coffee cup.

‘Arthur Clarke.’

‘Yes, the minister is fond of quoting that.’

But then, most technology seemed like magic to Bethan.

J-R continued to study sic_girl’s words as they scrolled up the screen. It was like an optical illusion: even knowing the personality behind these words was fake, he couldn’t make himself see it as anything other than real. Are we all so similar at root that someone can stitch our voices together and create a thing so much like life? Words trickled down sic_girl’s continuity. Daft commentary on a reality television programme. Obsessing about her appearance. Desperation for her next dose of medication. And then –

‘Ah – my God!’

‘What?’

‘Shit-monkeys!’

It was happening again before their eyes.

¶sic_girl

Sigh.

The only thing worse than being talked about is . . . well, actually loads is worse. Pain is worse. Ouchie. Don’t talk to me.

But since we’s talking about being talked about, let’s talk about Bethany Lehrer . . .

Seven

‘You want trust? Trust is based on surrender. Always, at some level.’

The auditorium had fallen in a trance. The only sound the electrical presence of the PA.

‘Once upon a time, if you wanted something, you gave up nothing more than money in exchange. Money. Disposable, fungible, easy to obtain for those with the wit. A commodity.’

Huge speakers pumped the voice, kicking it back around the packed hall.

‘But today when you want something, you first need to give up a piece of yourself, of your identity. This is the way it works now. We have to grow up and accept this or we will not move forward.’

This was premium Sean Perce. Before an audience he lost his coiled fury, presented an accidental quality. From the darkness of her panel-member’s chair, Bethany watched him lean on the podium like a club comic and survey the crowd. This was how she’d first seen him, four years ago at an open-computing event at the ICA. His boyish face, gravity-defying pomade and whiff of the ’70s rogue had disarmed her at once. He had the gift of making everyone in an audience – especially women of Bethany’s demographic – feel intimate with the man on the distant stage.

‘At Mondan,’ he continued, ‘we believe in the overriding importance of trust. We care – passionately – about the security and privacy of the millions of people whose data we handle every day. But if people want to take something out of the system, access a product or service, they need to accept that everything they say and do is recorded. And it will be used.’

Bethany scanned the pool of faces in the spill of stage light. She picked out journos, longing for the lecture to end so they could land their questions on her. This was the first shot they’d had since this pumped-up nonsense of a scandal began – and how convenient that this industry event offered presentations from the minister responsible for the Digital Citizen, and the CEO of the company providing the technology. According to Krish, acceptances for this session had doubled since yesterday and they’d had to move it to a bigger hall. Bethany could handle the hacks. More importantly, this was her chance to talk to Sean.

He was close to wrapping up. She was next, then the Q&A: then the two of them could talk.

‘But I’m going to leave you with a challenge.’

Sean snatched her attention back. The screen switched to a plain Mondan logo. No more slides. Was this off the cuff?

‘All this talk of trust is fine and dandy. But we still aren’t getting it. We don’t understand the cancerous power of
dis
trust. These systems for identification and security are all founded on trust. Suppose you don’t trust me. Well, all right, no problem: find someone you
do
trust. This trusted third party can tell you whether or not to trust me.’ He pivoted to look directly at Bethany. ‘Minister?’