Last year, a series of unfortunate events in Colorado Springs coincided with me being promoted onto the management fast track—and earlier this year a series of even more unfortunate events derailed me from said track, dumped me on a jet-propelled skateboard, and shoved me onto the career progression equivalent of a crazy golf course played with zombies for putting irons and live hand grenades for balls. Since then I’ve been subject to matrix management by bosses in different departments with diametrically opposed priorities who still think I work under them, while trying to establish just what is expected of me by much more senior people who think I work for them. It’s extremely fatiguing, not least because the furrow I’m ploughing is so lonely that nobody’s actually written a skills development manual for it: the Laundry is about procedures and teamwork and protocol, not super-spies and necromancers.
“I’m on the hook to the extent that I want to be,” I confessed. “Lockhart insisted, actually. Told me I’d never get anywhere unless I ‘set a course and stuck to it,’ to use his words. And Angleton just laughed, then told me to fuck off and play with myself.”
“Angleton said—” Andy’s eyebrow twitched again.
“No, that’s me; his actual phrasing was more . . .” Schoolmasterly was the word I was hunting for. A long time ago, Angleton spent a decade teaching in the English public school system (the posh, private school system, that is) and it had rubbed off on him—along with the extra special version of sarcasm generations of schoolmasters have distilled for keeping on top of their fractious charges. (Even his current nom de guerre, Angleton, was chosen with irony in mind: it irritated the hell out of our American opposite numbers, because of the one-time CIA legend of the same name. Really, he ought to be code-named SMILEY or something.) “But anyway, he gave me carte blanche, and my other boss expects me to—” I waved my hands, nearly knocked over my glass, and caught it just in time. “No, that’s not right. He just expects me to keep myself busy between External Assets jobs.”
“Paper clip audits.” Andy took a sip of Laphroaig. I didn’t bother to correct his misapprehension: External Assets, which Lockhart runs, is about paper clip audits the way the FBI is about arresting thieves, i.e. not at all but it’s extremely convenient for them that most people outside the organization don’t realize that. “Sounds to me like they want to see what you can do. Hmm.” Rueful amusement tugged at the sides of his lips. “So what are you going to do?”
“I’m building a spreadsheet. One with a lot of very interesting pivot tables.” Andy peered at me with an expression of mild disbelief. “Getting clearances for the data to feed it is a bitch, and it’s anybody’s guess whether it’s going to deliver anything useful, but if you don’t ask you don’t get . . .”
“Ask for what?” He hunched down in his chair. He was still a little shaky from the events of a couple of hours ago, despite all the whisky. “You’ve always hated admin work.”
“It’s not admin; it’s data mining.” I smiled blandly. “Big data, forward threat analysis. It’s a really neat idea from the suggestions box—my hat’s off to whoever came up with it. What I’m doing is proof of concept; there’s no way I could get a budget to do it properly. But if it works, then I can present it for discussion and maybe get something rolling.”
“Threat analysis and data mining?” Andy isn’t easily impressed: he has a habitual pose of arid detachment, an expression of distant amusement as if the slings and arrows of office politics (and the tentacles and curses of sudden-death engagement) are merely flying all around for his entertainment. But I’d got his undivided attention tonight: rescuing him from sudden death did that. “What kind of threats are you hunting, and where?”
“I’m looking for outbreaks. Not sure what, or where, so I’m trying to spread the widest net that comes to hand. Anything peculiar. A rash of spontaneous human combustion in Stevenage, or rabies in Ravensthorpe; could be anything. The point is to try and build a tripwire for anomalies.”
“But the police already—” He stopped. I shook my head.
“Not the police. Sure they’ll be on the line as soon as they confirm a fire-breathing lizard has come ashore in Liverpool, but what about the other stuff? We live in dangerous times. What got me thinking was, how many of the sort of problems we get called in to piss on start out small and get treated by the wrong emergency services? Body snatchers in Bath, zombies in Z—Zurich.” My metaphor engine had just broken: I took another sip of whisky. “A lot of possession cases show up as anomalous behavior, and while the ambulance service often bring the police in, it’s frequently mis-categorized as a mental health issue. So I’m trying to work out how to mine the National Health Service data warehouses for early signs of demonic possession. That’s what the task was about. ‘Everybody knows vampires don’t exist,’ it said: ‘develop a data mining utility to provide three sigma confirmation of the null hypothesis based on evidence from the NHS Spine.’ I don’t know who put it on the stack but it’s inspired! I mean I couldn’t have come up with a better ten-percent project if I’d designed it myself.”
Andy stared at me slack-jawed for a couple of seconds, then raised his left hand and theatrically closed his mouth. “Refill time.” He shoved his glass across the desk, towards the bottle. “Then you’re going to tell me why you’re telling me this.”
“You haven’t guessed already?” It was my turn to raise an eyebrow.
“You need a minion to run interference for you with the nice Data Protection Commissioner with the taser, right?”
“Right.” I topped up his tumbler.
He hesitated momentarily. “Deal. Because it’s crazy but it just might work, and it sounds a fuck of a lot less dangerous than what I was working on turned out to be. What could possibly go wrong?”
• • •
WHICH IS WHY, IN THE END, ANDY DIDN’T GET TO DEMONSTRATE his coding chops by summoning up an Eater.
And why I eventually sneaked my way into the clearances I needed to log onto the SUS Core Data Warehouse.
And, ultimately, why all the deaths happened.
2.
MEET THE SCRUM
HOW IT STARTED: ONE MONTH AGO.
“Hey, Alex, did you hear the one about the dyslexic sailor?”
“No—”
“He spent the night in a warehouse!”
Alex threw a bean bag at the joker—John—who caught it out of the air. Their supervisor was unamused: “Pigs!” said Mhari.
“It’s okay, hen, we’re committed,” said John.
Then she laughed. “You will be.”
It was lunchtime in an open-plan office, eight floors above the lobby level of a tower in Canary Wharf. North of Barclays, west of Santander, deep in the beating heart of global commerce. The office was a small clot of strangeness congealed in the pulsing circulation of an investment bank. They were in the bank, but not quite of it, this scrum of half a dozen Pigs and Chickens. They wore the suits and sometimes talked the talk, but held themselves apart; and when they left at night, they passed through a glassed-in corridor lined with metal detectors before they retrieved their personal phones and wallets and watches from metal lockers beneath the eyes of security guards. Some of them had worked in proprietary trading before joining this group; others had come straight out of academia, trailing the long shadows of student loans behind them (taken on by the bank as part of their golden handshake). But now they were in the bank but not of it, for the Scrum were not permitted any customer-facing contact at all. Indeed, they were employed by a shell company, the better to enable the parent’s corporate management to deny their very existence.
There were other signs of distinction about them. Their hours were not governed by the ring of the trading floor bell in Paternoster Square, or any other market for that matter. (Electrons never sleep.) Nor did they directly support any of the parent institution’s trading teams. They had a designated Product Owner, it was tru
e, to whom the Scrum Master (or in their case, Scrum Mistress) was answerable, both for handover of deliverables and negotiation of new goals, and the Owner actually worked with a committee of traders and analysts. But the office was carefully structured to keep the members of the Scrum as tightly insulated from their parent organization as possible.
That way, deniability could be maintained for as long as possible.
• • •
I WASN’T HERE, I DIDN’T SEE THIS, AND I CAN ONLY OFFER THIS fictionalized reconstruction, dredged from the turbid depths of my imagination and seasoned with facts.
I got to sit in on the autopsy later, as we uncovered the gruesome history of the Scrum. Not that it was particularly awful, up to this point. So, an investment bank has a bunch of in-house proprietary trading teams who specialize in taking positions on the basis of quantitative analysis of which way the markets are trending? There’s nothing illegal, immoral, or fattening about that. Investment Bank has a group who work on minimizing the latency in their trades, to reduce the market spread? Ditto. Algorithmic trading teams shoveling out and revoking put offers every few milliseconds to see what falls out of the high-frequency trading tree? We’re getting into dubious territory here, but everyone else has been doing it, ever since the Yanks deep-sixed the important bits of the Glass-Steagall Act and their commercial banks all sprouted high-stakes gambling arms—but that’s ancient history. The point is, the Bank wasn’t doing anything they shouldn’t have been doing, at least within the context of the crisis of early twenty-first-century capitalism.
Until we get to the Scrum.
A scrum—generic—is, depending on which dictionary you look it up in, either an elaborate way of restarting play in a rugby match (rugby: sort of like American football only with less body armor and more biting and gouging—I was forced to play it at school: did wonders for my character), or one of the quirkier cognitive disorders to which software project management is prone. The rugby version involves getting head-down and having a shoving match with the other side, sometimes involving ear-chewing, scrotum-grabbing, and neck-breaking (although the latter is frowned upon); the software variant is not dissimilar. It has its origins in Agile methodology, although it’s Agile hopped up on crystal meth and spoiling for a fight: exactly the sort of thing you’d expect a bunch of city high-flyers to find appealing (at least in principle, and as long as it didn’t look likely to detonate a large landmine under their bonuses).
The Scrum, singular, was the brainchild of Oscar Menendez, the vast unsympathetic brain at the heart of The Bank’s Data Analytics Support Division, and sometime Algorithmics star. Back in the prehistory of the early noughties they headhunted him from Google to show them how to apply map/reduce to the very large data sets they were processing in real time—
(You’re not listening, are you? Damn it, I suffered through the briefing. I don’t see why you shouldn’t have to suffer, too!)
Moving swiftly forward: the Bank set up the Scrum to try and bring some of the culture of an agile, highly responsive software start-up to bear on the job of developing new and improved tools for high-frequency trading. The Scrum was elite; the Scrum had esprit up to here; the Scrum’s long-term planning threshold was about twenty-four hours, marching to the beat of the daily stand-up meeting. They had a monstrous array of high-power data mining tools, live feeds from every exchange on the planet, their own individual compute-server farms. They had a PhD to headcount ratio close to 1:1 among the pigs—the hard core of math quants and algorithm developers. And every day they went in search of new and better techniques for identifying patterns in the data, trends that might be good for an extra 0.1 percent margin on every transaction for the handful of seconds it took before their competitors cottoned on.
They were brilliant, widely read, incisive, and effortlessly effective analysts and programmers. Which is another reason why, ultimately, so many people died.
• • •
PICTURE ALEX, ALONE IN THE OFFICE ONE EVENING.
Picture a twenty-something with spectacles and the remnants of a late heavy bombardment of acne cratering the ’80s designer stubble under his jaw. The spectacles are, of course, aggressively black-rimmed and thick-lensed. His suit is expensively tailored, his shirt of finest quality linen, and the collar is just slightly askew, because Alex has an image to defend: that he is a quant, that he is Oscar’s intellectual successor, that his mind soars as high above such mundane preoccupations as the City institution’s dress code as an SR-71 roaring on Mach 3 afterburner across the abstract vistas of category theory and algebraic topology.
It’s actually a total bluff. Like many truly brilliant minds, Alex suffers from an inordinate case of impostor syndrome, with mild hypochondria on top. He routinely arrives in the office each morning as a sweaty, seething blob of fear, certain that at any moment one of his colleagues will expose him as a fraud, unable to prove something as basic as An = Bn + Cn for any arbitrary integer n. He’s convinced that the reason he doesn’t have a girlfriend is that he has halitosis. (He doesn’t; the real reason is that he works over fourteen hours a day, six days a week, and spends the leftover hours sleeping, eating, and trying not to fall apart.) And he used to think that his acne was a symptom of malignant melanoma; every winter cold is multidrug-resistant tuberculosis.
Right now he’s a bundle of pure anxiety because of his salary. He’s theoretically due for a bonus—determined by the overall financial performance of the entire Scrum at the end of the financial year—but his basic salary of a little over £60,000 seems as dwindlingly inadequate as a Jobseeker’s Allowance when stacked up against the Croesian excesses on display in the basement car park. And he gets to see the basement car park every morning when he chains up his bicycle after cycling in from his room in a house in Poplar. He’s only been in the City for ten months (although it sometimes feels like a life sentence superimposed on a millisecond), and the habits of impoverished frugality he learned as a student weigh him down even as he tries to put on a good face for the people he works with.
You wouldn’t want to be Alex. Being Alex, aged twenty-four and alone in the City, is awful. But being Alex in three hours’ time is going to be much, much worse—and he’ll still be alive to feel it.
• • •
ONE OF THE GREAT BESETTING PROBLEMS OF THE MODERN AGE is what to do with too much information. This is especially true of high-frequency share trading, where every second a Sahara-sized sand dune of data must be gulped down and sifted for the fragrant cat-turds of relevant market movements. The ebb and flow of share prices is familiar from a thousand ticker-tape parades and ever-shifting number grids, but that barely scratches the surface of the problem. Your stake in a corporation that makes robot milking machines for the dairy industry can be affected by a press release from an upstream supplier announcing a better image recognition algorithm for udders. Or by a newspaper article in which a farmer explains that, because the release cycle for new cows (nine months and one week) is shorter than the release cycle for new udder-recognition software (eighteen months), they’ve been breeding cows for teats that are compatible with robot vision systems. Or by a supermarket capping the price of dairy produce, leading to a liquidity crisis in Ambridge.
The Bank had been addressing the problem of drinking from the data firehose for many years, of course. Indeed, one of the Scrum’s major tasks was to develop the banking equivalent of a self-cleaning litter tray: tools to help traders visually explore dizzying multidimensional arrays of ever-changing data without oversimplifying it into uselessness or causing them to throw up last night’s Premier Cru.
Alex was working late on a surge effort, trying to hook a new data set up to a funky fractal visualizer Dick and Evan had knocked up two months earlier: a Spanner-based widget that turned sixty-four dimensional data sets into rolling three-dimensional landscapes, the gradient and color and friction and transparency of each crusty outcropping encoding some aspect o
f the object of fascination. The goal: pour in the popularity of babies’ names over the past decade, sales of movie tickets in matinee showings, the Top 40 tracks pirated on BitTorrent, and the phase of the moon: get out an ordered list of toy manufacturers to buy or sell on the basis of their spin-off movie merchandising prospects. The reality so far: get out a scary-looking ski slope with black flags on the off-piste runs, not entirely suitable for traversal by a trading desk strapped to a snowboard.
Not that Alex knew much about skiing—his early lessons at school had been terminated by an unpleasant fall that had convinced him he’d broken his fibula for three days—but after eight nearly uninterrupted hours of staring at the screen he’d begun to go scooshy-eyed, his bladder was filing for divorce, and the interlocking mass of Möbius gears squirming behind his eyelids still resolutely refused to come into sharp focus.
I should go home, he realized. I’m not getting anywhere. Maybe if I sleep on it, it’ll come to me in the night . . . But what if it didn’t? There’d be a reckoning at tomorrow morning’s stand-up meeting. In the theater of his mind’s eye he could see Mhari smiling at him pleasantly and voicing all the apologies he’d have loved to keep to himself. He could hear Evan exercising his sense of humor, warped and slightly patronizing, playing off the weaknesses of his pair-programming partner. Maybe Oscar gracelessly and grumpily demanding to know what his salary was good for, if not this? I can’t go home, he thought dismally. Not until I’ve eaten my dog food.
Rhesus Chart (9780698140288) Page 4