The Decker and the 10M of Fiber Optic Cable

decker vs aiThe Market

Markets run in nanosecond time slices.

Computerization ruined the interpersonal dimension of stock markets. Once networked computers could make trades faster, smarter, and more efficiently than humans, the once sweaty and chaotic trading floors, hallmarks of stock markets worldwide, disappeared.  In New York, they disappeared into faceless impersonal datacenters across the river in New Jersey; enormous white hulking industrial buildings surrounded by barbed wire and patrolled 24 hours a day by man and dog alike.

Then, the fight for efficient trades was one of inches: who build the most efficient trading algorithm, who had the lowest routing latency, and who had the best physical position to shave off picoseconds from their nanoseconds when routing packets across a switch to get into the market, make a trade, and get out again.

And then other markets proliferated: black faceless servers blinking red and green lights into the air-conditioned darkness, siphoning off millions of trades from one market and to another.  Trading Markets are algorithms processing other networked algorithms and pushing completely abstracted bits which could, if stared at like a quantum function long enough, collapse into a real value in money, jobs, economies, and lives. Anyone with enough hackers can build a market, but it takes moxie to get it into the faceless New Jersey datacenters to start making trades.

Algorithms chase algorithms. A buy on a single anonymous stock at a penny on one black blinking box market morphs into a sell on another market at speeds faster than any other system can match. By time the markets synchronize and come to consensus on a price, the algorithm has snuck around the datacenter and made millions off teeny tiny race-ahead trades. But, if some other algorithm is faster – or smarter – it can snipe the buy and rob the first.

This leads to a natural military build-up of algorithms.  Boxes fight for position on racks to lower their routing latency to take most efficient use of their preferred trading platforms. Algorithms become faster, smarter, and more efficient.  Nastier.  More aggressive.

This is the natural first bloom for AI. Not AI in a traditional, Asimovian sense.  No one is building R. Daneel Olivaw’s positronic brain into a trading platform.  Human intelligence isn’t the right fit for this sort of job.  This is AI embodied in Black IC – designed, instead of keeping the Deckers out, keeping the trades in.  Going to war with other trading Black IC embedded in the other High Frequency Trading (HFT) platforms across a vast sea of cabling.  Processing vast amounts of data flowing through the nanosecond pipes into their machine learning algorithms and figuring out the spanning trees of the weaknesses of the various market systems and the other platforms.  And then exploiting those weaknesses.

They know everything about the world economy: how to build it, how to manipulate it, and how to destroy it.  They’re AIs in a perpetual state of entrenched and never-ending financial war.

Normal every day people, sitting at home, playing around with their day-trade accounts, playing with $100 here, $250 there, have no idea their money gets pumped through an intelligent, malignant systems who feeds off their petty trades to enhance their own algorithms.  This is the market of the future.

The Decker

The information Decker scores insider information from the Matrix about the future earnings report of a publicly traded partly owned LLC of the Mitsuhama Corporation. They’re going to take a bath on their quarterly report.  A serious bath.

She never tells anyone how she laid hands on this information. All anyone needs to know is that she has it, and it’s authentic stuff.  And she has a state-of-the-art deck.  Perhaps she got in via some social engineering instead of brute force hacking.  But she’s not saying.  All she’s saying is that she has a report, and she’s already placing her trades.

Mitsuhama will announce this in a week to their shareholders.

What the Decker also doesn’t tell anyone is that she’s been chatting on the Matrix with someone who develops those AIs that run on the trading platforms in the datacenter in New Jersey.  Turns out working in financial services IT is a thankless job without the big bonuses of the being-rapidly-replaced-by-AI very expensive money managers.  On one hand, no one working in financial services IT starves in the street. But, they’re all forced to work in sub-basements where the suited and moneyed class aren’t bothered by the sight of technical people.

The Decker and the financial services IT have a series of nice chats about the trading AI and ways to technologically mess with the system, and the best way to influence the system to make millions without anyone ever noticing.  They talk about nanoseconds and switch sensitivity.  The financial services IT maps the rack layout for the Decker.  He walks through the system hardening.

They talk more about the physical location of the datacenter.  It is pretty classified in bank parlance, but it’s not clear the financial services IT worker much cares any more.

The Decker probes the New Jersey datacenter but discovers the same thing that every Decker before her has learned: knowing about its existence isn’t enough.  While it is not completely cut off from the Matrix – after all, trades must flow and money must flow out – many of the lines in and out are hard-wired dedicated lines into banks which only flow bank traffic.  The actual ingress and egress points in and out lead almost nowhere.  The nastiest black IC she’s ever seen protects one useful ingress point.  She probes it, and fights it anyway, and gets bruised.

It takes weeks, but she finally sneaks in around the nasty black IC and makes a brief, and strange, contact with one weird mathematical AI who runs a trading platform.  They can barely communicate. The trading AI works completely counter to human thought.  It wants the Decker to feed it information to allow it to destroy the other trading platforms.  The Decker has only the quarterly report.

The Decker figures out a basic language. She persuades the strange, murderous AI she wants to destroy its foes with her data, and she can destroy them.  She arranges the right time to place a series of Mitsuhama shorts with this strange horrible thing.  It triggers several internal high-powered machine-learning algorithms and figures out how to place the trade while screwing every other AI in the datacenter.

But to make her trades successful, it tells her, she needs extra nanoseconds.

The Run

Should a Shadowrun team do the bidding of a high-powered financial AI trapped in a state of perpetual war and gaining intelligence through an endless stream of data through its machine learning algorithms?  If it pays off in millions, why not?

The Decker explains to her Shadowrun team this: if they slow down the rest of the trading platforms as the AI places her trade, they will have exclusive access to massively short Mitsuhama.  The stock will go into momentary free-fall and they’ll make millions of Nuyen.  Which, of course, they will split.

All they need to do, during the announcement, is to plug a 10M cable into a specific port in a router in a rack at a specific time.  This forces all the packets traveling over the cable to take a sliver of a fraction longer to process.  Some of the trading AIs working their magic high frequency trades will slow down enough that the Decker’s contact AI will place all their trades first, and complete them, and drive down the Mitsuhama price. This will  generate a panic.  They will short off the panic and make vast sums of cash. And then, when they remove the cable and let all the trading go forward as normal, the trading AI will buy up stock at bottom rate prices with the cash they just made.  They will all be paper and actual millionaires.

The Decker will finish the transaction by funneling the money through a series of shell companies into off-shore accounts.  Then the money will be poof. Gone.

Sure, it will crash Mitsuhama’s stock price and likely trigger waves of future layoffs but they’re an evil crypto-corporate overlord so… it’s all good.

Technically, this isn’t a heist, the Decker explains.  It’s a momentary IT upgrade to a router using a single loop of cable to introduce a bit of lag.  The Decker isn’t suggesting they mess with code, or programs, or even break into any computers.  They’re not going to fight black IC.  All the alarms are physical.  And the Decker can place her trade with the insane trading AI via her deck while she waits outside.

Simple, right?

Here’s the list of problems to this crazy plan:

  • The physical coordinates of the datacenter is a highly guarded secret.   The Decker has them… roughly… from her financial services IT contact but she will have to use the Matrix to get some overhead geographic scans to actually find it. And that, in itself, is likely a dangerous run through the Matrix.
  • Automated systems monitor and guard the approach to the datacenter.  Without some sort of faked-up ID with matching uniforms and marked vehicle, Shadowrunners will never get a truck full of their gear near the facility.
  • Direct front-door approach to the facility isn’t going to happen. Drones, security servicemen in power suits, and various dogs protect the facility. Even flashing IDs and gaining entrance will mean monitored passage while they fix HVAC or whatever they’re claiming.  Getting partway in and then completing the trip by air at night will be more successful than going straight up to the front door.
  • If the Shadowrunners get past the front door physical security of barbed wire, drones, security, dogs, and high walls, they’re going to have to get in and wait because, to reduce detection, they’ll have to go in at night and trading doesn’t open until 9am.  Even the trading AIs are asleep when the banks close.  It’s a long wait to avoid detection, not make a sound, and not need to visit the bathroom.
  • Security cameras blanket the place.  Either the Decker needs to turn those off through the backdoor she’s already found, or the Shadowrunners need to take out the security guards watching the cameras.
  • Once inside the facility, the Shadowrunners will discover its filled with hundreds of racks all filled floor to ceiling with blinking light machines which all look identical.
  • Understanding which rack is critical.  Luckily, here the Decker can lean on her financial services IT resource.
  • Then, right before trading opens, they need to find the right ports on the right routers in the right racks.
  • And through all the firefights, the 10M cable is good.

For argument’s sake, this all happens, because named characters.

Shadowrunners break into a fortified datacenter, hide in the walls until the right time and, using Decker-supplied maps, figure out which rack and which piece of equipment in which rack.  A series of big buts with caveats.

A pinch of a plastic clip.  A small snick.  The new cable slows down all AIs attached to that particular subnet. They exist in slightly slower time. Their flow of data curtailed. Their minds sluggish.

The Decker’s AI has extra nanoseconds.

Invisible war explodes in the datacenter.  AIs leap on crippled AIs.  Those AIs, including the Decker’s, takes advantage of this new development in their endless war to gain more nanoseconds for their masters.  Billions slosh invisibly through the datacenter.  Some servers pop with surges. The place smells of ozone.  Newly stronger AIs escape from their datacenter imprisonment and make their way on to the Matrix.

AIs place trades on markets.

Markets soldier on.

Getting the Hell Out

The Shadowrunners are millionaires.

But, here’s the next problem for the Shadowrunners: they have to escape the datacenter.  The must work their above problems in reverse.  And, to escape detection, they must escape at night.  That’s a long time to have security cameras down… they’ll need to bring them up and take them down again as they leave.

The Shadowrunners can simplify their escape plan by leaving the cable clipped to the router.  Even if the Shadowrunners were never detected getting in or out, someone will notice their AIs are not performing up to code.  They will perform diagnostics, find the network is sluggish, and call in remote hands to inspect the physical datacenter (legally).

But meanwhile, the Shadowrunners will cause havoc in world markets, let the AIs loose on the Matrix, massively short a company when its report comes out, destroy some jobs with crypto-corporations, and make millions and millions of nuyen…

Or they could have day-traded on the Decker’s inside information and fed the AIs information to make them happy and fat. But what fun is that?

Image Credit: Art by Jaydot Sloane of Vanity Games –


  1. Ivan Kvapil says:

    I am enjoying this new dive into Shadowrun, has to be one of my favorite game systems ever made

  2. I greatly enjoyed this article,and the imagery of financial war.

    > But to make her trades successful, it tells her, she needs extra nanoseconds.

    I don’t really understand, why it’s *necessary*. I get why it’s a nice advantage, but I don’t get why the success of the entire scheme hinges on it. I get why it’s good from gameplay perspective: the mission is awesome and makes for a much better closure. But do we have any better justification?

    Let’s look. So, you have this insider report. How do you make money with it?

    1) You are mundane murder hobo: just short Mitsuhama’s stock. You don’t have much money, but you have something. Sell 100 stocks today (price: 100 Nuyen), buy them back tomorrow (price: 70 Nuyen), get a 3000 Nuyen profit.

    Result: small profit, nobody notices.

    2) You are rich corporate AI: do the same, but on a much larger scale. Other market players notice this, but they are unable to prevent you from profiting. Today, you already sold stocks at 100 Nuyen. Tomorrow, when the report is made public, the price is going to be lower than today (and you don’t even need to predict the future equilibrium price of Mitsuhama to know that). Tomorrow, you are going to buy low, *no matter what happens inbetween*.

    Result: millions of profit. Market crashes, ripples felt anywhere. Price of fish skyrockets. Funny things are happening at the mortgage credit market.

    3) You are rich *and smart* corporate AI: spend some time thinking. Mitsuhama’s stock will crash, what happens next? They will downsize their research department. Their Project X likely will be postponed or canceled. The stock of their competitor, TetraCorp is going to rise. TetraCorp is the largest industry of Nigeria. Nigerian Naira is going to rise. What do your fancy machine learning algorithms predict about that? Could you arrange a margin call to some of your opponents?

    Place orders accordingly.

    Result: trillions of profit. Some nations go bankrupt, some turn unexpectedly rich overday. The ownership of Vault-Tec changes hands. President of Germany barely escapes assassination attempt.


    None of that require nanoseconds. The schemes aren’t interactive. The only place where low latency makes a difference is a moment when you send your first orders: you need to buy while price is still low and sell while price is still high. You need to short Mitsuhama’s stock before your competitors do the same. But, assuming the cat remains in bag, you have a day to do it.

    Yes, extra nanoseconds would be a nice addition, but they always are a nice addition. They could be a separate their own right.