The map on her screen described the posterior descending pathway, or PDP, of the birds’ vocalisation system. The contributors had all been adult males, each with a fixed song of their own that was somewhat different from the others’. Redland had chosen the PDP for the sake of those two characteristics: it controlled a single, precisely repeatable behaviour in each individual – the bird’s fixed song – but there was also a known variation between the contributors thrown into the mix: no two birds sang quite the same song. Unless the team’s mapping techniques could cope robustly with that degree of difference, making sense of anything as complex as the brains of rats who’d learnt to run different mazes would be a hopeless task.
Nasim slipped on her headphones and linked the latest draft of the zebra finch map to a software syrinx, a biomechanical model of the bird’s vocal tract. She had plenty of fancier, more quantitative ways to gauge her progress, but listening to the song these virtual neurons created seemed an apt way to judge success. The songs of the individual live birds had been recorded, and Nasim had heard them all; she knew exactly what the fast, rhythmic chirping of an adult zebra finch should sound like. As she tapped the PLAY button on the touchscreen, her shoulders tensed in anticipation.
The song was disorganised, weak and confused, more like an infant finch’s exploratory babbling than anything a confident adult would produce. She glanced at a histogram showing a set of simulated electrical measurements; the statistics confirmed that they were, still, nothing like the signals measured by micro-electrodes in the brains of real adult birds.
The different mapping techniques complemented each other, each one excelling at revealing certain aspects of the neural architecture, but for the data to be meaningfully combined she needed to find common signposts that could be used as points of alignment. It was easy to build, say, a composite human face by locating all the eyes and noses in a thousand photographs, then making sure that you merged eyes with eyes, rather than eyes with noses. But for a thousand birds with a thousand different songs encoded deep in their skulls, the signposts were subtle aspects of the neural network, and they had to be coaxed out of the partial, imperfect data that each individual map supplied. Right now, it sounded to Nasim as if she were merging pitch from one bird with tempo from another, to produce a musical concoction that was not so much generic as puréed.
She steeled herself and plunged back into the computer code for the map integration software. The task was proving more difficult than she’d expected, but she did not believe it was hopeless. She was sure that once she found the right perspective, the right mathematical point of view, the signposts would become clear.
Nasim usually brought a packed lunch with her, but all her routines were askew today. By two o’clock her concentration was failing, so she went downstairs to the Hungry Mind Café. She bought the vegetarian ragoût and took it to a table where three of her colleagues were seated.
‘How’s the revolution going?’ Judith asked her.
‘There was a big demonstration in Shiraz yesterday,’ Nasim replied. ‘Ten thousand people, according to some witnesses. Not quite a general strike, but it’s spread far beyond just students now.’
‘Have you still got relatives in Iran?’ asked Mike.
‘Yes, but I haven’t really stayed in touch with them,’ Nasim confessed. When her father had been executed, her aunts and uncles on both sides of the family had declined to speak out against his killers, and Nasim had been so angry with them that she’d cut herself off from everyone, even before she and her mother had fled. Fifteen years later she was less inclined to judge them so harshly, but she’d never tried to rebuild bridges, and the blameless cousins she’d once played with were strangers to her now.
Hunting for a chance to change the subject, she gestured at the empty plates on the table. ‘Looks like you’ve all been here for a while. So what gossip have I missed?’
‘Mike broke up with his girlfriend,’ Shen announced.
Nasim looked at Mike to see if it was true; he didn’t seem too devastated, but he didn’t deny it. ‘I’m sorry,’ she said.
‘It was going nowhere,’ Mike replied stoically. ‘We were philosophically incompatible: she belonged to True Love Waits… I belonged to True Love Wilts.’
‘So how can we take your mind off this tragedy?’ Nasim wondered.
Shen said, ‘Actually, we’ve been playing Thirty-Second Pitch. You want to choose one?’
‘Hmm.’ Nasim’s mind was blank, then she said, ‘Mike, you have thirty seconds to make yourself indispensable to… Amazon.’
‘Amazon?’ He grimaced with distaste. ‘I’d rather work for the IRS.’
‘Twenty-five seconds.’
‘Okay, okay.’ He closed his eyes and took a deep breath. ‘I offer to write a psycho-linguistic compression algorithm for text. MP3s for the written word.’
‘Compression?’ Judith interjected sceptically. ‘I don’t think Kindle is facing bandwidth problems.’
‘Not compression for the sake of bandwidth,’ Mike explained, ‘compression to save the reader’s time. Abridgement. Like Reader’s Digest Condensed Books, but fully automated, and based on a rigorous scientific analysis of what readers will actually retain. With music, we know that it’s safe to strip away certain sounds that are masked by others… so surely we can figure out what words can be omitted from a great slab of Melville or Proust without altering the impression that they leave behind. People are far too busy these days to indulge in rambling, discursive novels… but if they can feel just as Prousty in two hours as they would have in eight, every word lost is time found.’
‘Moby-Dick left no impression on me at all,’ Judith said. ‘I might as well have not read it. But other people can recite long passages from it verbatim. Doesn’t that undermine the whole idea of compression?’
Mike hesitated. ‘No, it just means it will have to be tailored to individuals, based on a personal brain map. So who better for Mr Bezos to hire than someone with brain-mapping experience?’ He turned to Nasim. ‘I rest my case.’
She smiled. ‘Well done. You’re hired.’
Shen said, ‘Can you improve their recommendations algorithm while you’re at it?’
‘Once they have your brain on file,’ Mike replied, ‘everything they do for you will be beyond reproach.’
Nasim spotted Dinesh approaching, beaming ecstatically. He was carrying an opened envelope and a letter.
‘I’ve got funding for HETE!’ he exclaimed, waving the letter. ‘Lab space, equipment and ten people! For three years!’
‘Congratulations!’ Nasim glanced back at the others and caught a flicker of irritation crossing Mike’s face.
Dinesh joined them at the table. ‘I can’t believe it,’ he said. That was usually an empty protestation, but he sounded genuinely dazed. ‘It’s really going to happen.’
Mike said, ‘So you’re just giving up on the HCP?’
Dinesh couldn’t stop smiling. ‘What difference does this make to the HCP? That will happen or it won’t, it’s not up to me.’
Judith said, ‘Where’s the money coming from?’
‘Bill and Melinda – bless his shoddy, monopolistic software.’
‘Sure it’s not the Turd Foundation?’ Mike quipped lamely.