This V-Tuber Does Not Exist: The Plot

Campaign retrospective, as well as a discussion of Shadowrun magic, still more complaints about Shadowrun worldbuilding, and some amusing combinations of Shadowrun mechanics

Now that my preparations have been outlined, it’s time to talk about what actually happened during the game. Calling it “plot” is inaccurate, because the vast majority of it was driven by the actions of the players, but I don’t have a better term.

As I mentioned previously, this campaign was pretty atypical by TTRPG standards; as such, I have to warn you that reading this post may make you angry that other people are having fun in the wrong way. My condolences.

As usual, if you are confused about any terms, make sure to check out the glossary, and if you are looking for other posts, check out the blog map.

1. Prologue

Prologue had the Inciting Incident, an initial Driver for the players, and served as a sort of “tutorial” area. Most of the players hadn’t played Shadowrun before, so I needed a safe area where I could show them the rules, and slowly introduce them to the campaign.

Before the campaign began, player characters took part in a brain scanning experiment - supposedly as part of a research project to improve brain-computer interfaces. Little did they know that their memories were scanned, saved to disk, and used as part of one of the few remaining AI research projects in the world. Existence (and eventual success) of this project wasn’t based on any kind of reductionist model of Shadowrun AIs I had in mind, and instead was just a blatant Inciting Incident I needed to get players into the action as a group of AIs. Fluff behind the project was that researchers tried overwriting AI memories with the brain states of the scanned people, hoping that the internal repair mechanisms of the AI mind would kick in and integrate the two parts together.

As a result, our characters woke up in the body of a poor elf, running on a commlink (phone built into your head), and controlling the body directly.

When I was designing the campaign, I figured that this may cause problems. Characters only having a single body is quite limiting; because of this, I advised the players to specialize in different areas. My idea was that one of them would be controlling drones, other would shoot guns, third would do magic, fourth hacking, and overall there should be less overlap of areas of concern. 

Fluff-wise this was justified by researchers trying to make some sort of multi-faceted combat AI, merging the experiences of several different people - ones with magic knowledge, drone control, hacking and so forth - into one coherent package. After the characters woke up, they were put through some tests that included all of those things. These tests allowed me to explain the basics of Shadowrun rules in a calm environment.

After the tests were done, the researchers left the body alone in the server room, and clocked off for the night. From their perspective the newly created AI was following instructions and not showing any initiative; they thus didn’t feel any danger. Their planning assumed that either the AI creation would succeed (in which case it would follow instructions) or it would fail completely and not work. That perspective was a thinly veiled parody of many modern AI researchers’ attitudes towards safety.

2. Lab Rat

Because this was a tutorial area, I prepared a way for the players to escape captivity and get out into the world. Events necessary for that were set to happen T+4h from the moment researchers went to bed. But the players had other plans.

Throughout the campaign, one thing would stand out about the players’ approach to problems: the extreme levels of ever-present paranoia. I have never seen such opposition to risk backed up with the ability to come up with various scenarios that could screw them over. If there was a way to move themselves further away from any threat, they would do so, seeking to work through multiple layers of proxies and misdirections whenever they could. This managed to keep them safe despite the numerous threats present in the Shadowrun world, so you can’t really argue with results.

In the case of the lab, they were too paranoid to sit tight, and so immediately started trying to figure out how to get out. But they were also too paranoid to start walking around, in case someone was monitoring their movements. Fortunately for them, they had the single most dangerous weapon in the Shadowrun world: an open wifi connection.

For the first hour they limited themselves to remote hacks of wifi routers around the facility, checking logs, and generally exploring the network without trying to interact with the systems on it. Their body hadn’t moved a muscle throughout. After an hour of trying to find a connection to the cameras around the facility (cameras were on a separate air gapped network), they figured out when researchers went to sleep by checking the connection logs from their laptops to the wifi. Players thus felt pretty safe that everyone was asleep now, and they could try doing some magic.

Specifically, they summoned a spirit of air and sent it to map out the facility. Spirits in Shadowrun are all but invisible if they want to be, and can pass through walls; furthermore, no autonomous sensors for spirits exist. Standard security countermeasure against them is to use magically-sensitive dogs, but players hadn’t seen any during the testing phase. Therefore, the only way for their scout to be caught would be if some mage was actively searching for a spirit. They considered this to be unlikely, because the only mage in the facility should be asleep by now. Even if they weren’t, actively looking for spirits is effortful, and so mages do not do it all the time.

I came up with the floor plan at the same time as I was describing it. It was a small bunker with a single exit: furthermore, it was underwater, and the only escape path had to go through a long tunnel leading into the ocean. They couldn’t swim out, and they had no submarine.

This is where the first upset of the campaign happened.

You see, I expected the players to optimize their characters. What I didn’t expect was how multiple optimized AIs piloting the same body would interact with point buy mechanics, AI rules, and the other mechanics of Shadowrun.

As I’ve described in the previous post, things in Shadowrun are accomplished by rolling pools of dice. Number of dice rolled is composed of an attribute and a skill number, with both having maximum caps: skills capping out at 12, and attributes generally at 6. Core rules include an “Exceptional attribute” quality - a “feat” or “perk” character can take - which increases the cap on one attribute by 1. On the other hand, the AI rulebook, for some reason I cannot possibly comprehend, included a different AI-exclusive quality - Exceptional Entity -  which removes the cap entirely on one attribute, and increases the caps by 3 on all the rest.

This is compounded by the fact that, obviously, AI do not have physical attributes. They only have mental ones, which they use in place of physical attributes for various actions (e.g. shooting a gun from a drone). This lets AIs min-max like there is no tomorrow, and not suffer the normal drawbacks for doing so (e.g. having less health). I believe that all of the characters had at least 12 in one attribute of choice from the start of the game, with similarly high skill values.

You may remember that in the previous post I mentioned that a “hard” test requires 4 hits, and a base dice pool of 18 would be scratching the ceiling of possibility. My players were casually throwing around 18 - 24 dice right from the start: for an average result of 6-8 hits.

This was further compounded by the fact that all of the players were occupying the same body. Because they could pick which AI was in the pilot seat, the overall body acted as if it had the best skill and attribute of any AI for any given action. This meant that the character was effectively min-maxed in everything, except physical attributes. I have somehow managed to not realize this would happen during the planning stage.

Such a high level of skill meant that some impossible-seeming things were actually pretty feasible. For example, nobody can swim through a kilometer-long tunnel unaided. However, if you can casually roll up 18 dice on a water spirit summoning test and order it to propel you forwards, you can radically shorten the travel time. Furthermore, because you are an AI and don’t need the brain in the body to be conscious, you can mostly ignore minor suffocation during the trip. Which is what the players did, right after sneaking into the surveillance room under the cover of invisibility and erasing all footage of their escape. All that was left was some minor self-resuscitation on their way towards the shores of Seattle.

3. Wings of Freedom

Ordinary players would be satisfied by this escape - gone from captivity while leaving no trace, lost somewhere in a large city with no realistic way to find them. Not this group.

First thing they did was raid some of their past houses for some cash, clothing, and other necessities. Setting up in a van belonging to one of the players, they started doing some internet research. First of all, they looked up what happened to their past selves - all of them died in “accidents” a couple weeks back. Secondly, they found a business specializing in fake identities - after all, they would need some papers made for their current body. After their photo was taken, they booked it across town to meet with an illegal cyberware surgeon, in order to check their body for bugs. They found out that researchers built in a countermeasure - an override for their control of the body - and paid to get it removed.

Shocking the surgeon by how quickly they recovered from narcosis (you don’t need the meatsuit to recover if you don’t do your thinking with a meat brain), they left, and holed up in an abandoned construction site to do a deep check of their soul in case it was tampered with.

With their body and soul cleared of suspicion, players felt safer, but not safe enough. It was still plausible that they could be tracked through some mechanism they didn’t think of: the only universal counter to all tracking methods would be distance. They decided to flee town. However, what method could they use? Roads may be covered with scanners that could detect their body shape, trains or planes might scan their face to verify identity, and ships are simply traps once they are out at sea. On the other hand, walking on their own two feet would be too slow to evade pursuit. How could you square this circle of fear?

Very easily. First of all, players bought a hang glider, and some camping equipment. Second, they applied the Levitate spell to themselves. This spell allows you to telekinetically move an object at a certain speed within your line of sight, explicitly including the caster, for as long as you concentrate on the spell. They proceeded to climb into the glider, levitate it up into the air, and start flying. By periodically varying which way the spell was pushing them, they could move horizontally at a rapid pace without losing altitude.

We did some math on how quickly this sort of device could move, and within a day they were halfway across the continental US, having flown over multiple country borders, and well outside of any reasonable tracking schemes.

The deviousness of this escape trick is how hard it would be to track. Conventional detection methods such as radar shouldn’t even notice a hang glider - it is too small. Eyesight is useless beyond a certain altitude, as objects become too small to distinguish from a bird. Magical sensors are almost universally short-ranged, and aerial border patrols tight enough to catch a single glider would be too expensive for any state to maintain, not to mention contradicting the canonical existence of smugglers.

Of course, ordinary people couldn’t keep the same spell going for hours, while monotonically repeating the same action (up, down, up, down, ...). But ordinary people do not have the quite literally superhuman force of will, as demonstrated by the stats of the AI character responsible for magic. And ordinary people cannot think fast enough to simultaneously control a spell and watch anime from the internet, like an AI can.

Once they’ve arrived on the opposite side of the US, players could finally stop, relax, and decide that no adversary could possibly chase them across the continent…

Of course not, have you not been paying attention? They decided they needed to get all the way to Europe. But getting on an intercontinental flight seemed scary - what if their face could be identified on airport cameras? In the end, they decided not to risk it when a good strategy was already available. In three hops (to Canada, from there to Greenland, and from there to Norway) they crossed the ocean, and took a train to Germany. There, they went to a plastic surgeon, changed their face, and bought an entirely new set of fake IDs. Only then did they think they were (temporarily) safe from pursuit.

4. Chasers

Unbeknownst to the players, they had actually picked the one action that would perfectly foil their pursuers. One of the researchers at the lab - the mage - was a plant; once the AI creation project succeeded, they reported this to a third party, who immediately sent a team of shadowrunners to retrieve the asset. This was the exit method I designed to get the players out of the lab. One hour after the players escaped, shadowrunners entered, hacked through the security, stunned the guards, and found a complete absence of any AI. The plant, naturally, had absolutely no idea how this could have happened.

Imagine yourself as a shadowrunner. What would you suspect? It seems unlikely that the AI snuck off on their own: they didn’t demonstrate any initiative during the tests, and would have needed a submarine to flee. Cameras didn’t show any action either: did the AI somehow figure out it should erase the footage? Implausible, AI aren’t capable of creativity.

No, what is much more plausible is that either the plant was a triple agent, or that there was a second plant, and someone else already stole this highly valuable AI drone.

Not to worry: they knew exactly how to deal with this situation. First of all, torch the lab, and everyone inside: this was, in fact, part of their original job. Second, get out of the lab and try to find the thieves.

In fact, they still had one card up their sleeve. Blood of the elf used as the “body” for the AIs was stored in the lab where our players were originally brain-scanned. On the next day, shadowrunners broke in and stole the blood sample; that very evening, they paid a mage circle to perform a very expensive ritual to locate a person somewhere in Seattle using their blood as a focus… but found nothing. Unbeknownst to them, just a few hours before, our players had slipped the noose by flying outside of the ritual range.

At this point, shadowrunners reported to their client that this job was a total failure. What would the client think?

Well, perhaps someone stole the AI drone and immediately moved it outside of Seattle. However, as time would pass and no obvious culprit would materialize, this would become less and less plausible. If some megacorp was behind the theft, then they would have used the AI code in some product, such as a new line of drones; increase in capability should be immediately obvious. Furthermore, someone, somewhere, would inevitably break secrecy. Who would want to quietly steal a revolutionary AI and then proceed to not use it?

On the other hand, there is a much more plausible explanation for what happened: someone fucked up. Perhaps there never was a successful test, and the plant simply lied to be extracted. Or maybe the shadowrunners fucked up the extraction and then lied, gambling that they would get paid a fraction of their pay instead of nothing. That would seem to be much more plausible, and over the coming months, it would only get more so.

Paranoia of the players has saved them from pursuit - first directly, and then indirectly.

5. On The Hunt

Another thing that came up a lot in the campaign was phase transitions. Let me explain.

In engineering, there is an informal rule that a quantitative change by an order of magnitude counts as qualitative change. For example, all woods have density in the range of 700-1300 kilos per cubic meter, so weight differences between them are quantitative. But lead (11 300 kg per cubic meter) is a different story: all sorts of new things have to be considered if you are switching from wood to lead (e.g. weight limits of bridges your trucks drive over).

In this campaign, players would go through multiple phase transitions where some property of theirs - such as how much money they had available - would suddenly jump by one or more orders of magnitude, completely altering the core assumptions of their behavior. For example, back when they were buying the hang glider and camping supplies, those expenses had almost exhausted their cash reserves. I started to think that - maybe - they would actually have to do some shadowrunning to get money for future projects. But during the flight over the ocean, one of the players had a clever idea: what if they did bug bounty hunting?

For those unfamiliar, software firms often offer bounties for finding bugs (such as security holes) in their software. This is incredibly beneficial for the firms. If nothing is found, then they have effectively employed hundreds of bug auditors for free. If something is found, bug bounty is a fraction of the cost of the damage that a serious unpatched bug can cause.

There is pretty much never a reason not to have a bug bounty program. It is strictly beneficial to the business; the only real case where it wouldn’t exist is if the business was being stupid with money. Megacorps in Shadowrun are many things, but they are not stupid with money - I thus had to conclude that bug bounties must exist.

And once you conclude that they exist, it really wouldn’t be too surprising that players, with their extremely high stats, would start finding them. But maybe they couldn’t do it reliably? After all, other hackers in the shadowrunning community can’t make a living doing this, so maybe the bounties require extremely high rolls to pay out well?

This is where another totally broken mechanic of AIs comes in: edge. Using an edge point allows you to add your edge rating to a single roll. Normally, you regenerate one edge point per night. AIs, instead, for some ungodly reason, regenerate all of their edge points per night. They also require only three hours of “sleep” per day, and can do this “sleep” at any point. This means that an AI can edge significantly more often than a normal person; if you aren’t making a lot of rolls, you can be edging all of them.

Furthermore, AIs have a unique attribute - Depth - symbolizing how complex they are, or something. The value of this attribute serves as the maximum for their edge attribute; and Exceptional Entity removes the maximum for depth (in addition to removing the maximum for one normal attribute). In other words, it also removes the maximum for the edge attribute, meaning AI can have more edge points and have those edge points add more dice to their pool, once they put the XP in.

Together, these two rules push AI characters to absurd numbers of dice they can use on a whim. And because the number of hits you can get from a roll is distributed on a gaussian curve, this means that AIs become exponentially more capable than mere mortals as XP rolls in.

At this point in the story, stats of the players hadn’t changed much since character creation - after all, only a couple days have passed since the start of the campaign. Nonetheless, high stats combined with edge were enough to easily make them competitive with the top hackers in the world - and thus allowed them to rake in big bucks from bug hunting. Money that had to be carefully budgeted when they were working with a set amount of cash could now be spent freely, safe in the certainty of new income.

This is why they changed their face once they got to Germany - an expense they absolutely could not afford in the eastern US.

Getting a car, our characters set off across Germany - it wouldn’t do to stay in any one place for long. They also decided to expand their software engineering projects.

6. Render Me Like Your French Girls

During the flight over the ocean, our characters did two things to pass the time. First of all, they watched MRE review videos: deciding what food to pack for the flight had been a strong influence. Secondly, they watched V-Tuber videos.

Once they came to Germany, they figured they could try the gig themselves. All they needed was some rendering software that could convert their thoughts into a realistic V-Tuber persona. Fortunately, they had all that was necessary to design and implement such a tool: time, skill, and some computer hardware. And when it came to time, AIs had a pretty significant multi-layered advantage over humans. First key element here is thinking speed.

Shadowrun combat has pretty unique initiative rules. Having a higher initiative doesn’t just mean you go first: it also means you get more actions per turn. I decided to take this to the logical conclusion, and said that if you have a way of maintaining above-average initiative for 24 hours, then you straight up think faster. Most ways of boosting initiative available to meat people are temporary: enough for combat, but not sustainable over the long term, and so this should not have major worldbuilding effects. AIs, though, have a higher initiative permanently. I won’t bore you with the math, but it suffices to say that our characters thought and coded 2 - 2.5 times faster than ordinary people. On top of this, AIs only need to “sleep” for 3 hours, and don’t have any time-consuming bodily functions. This gives them 21 hours of useful labor instead of the human maximum of maybe 14: another stacking increase of 50%. Together, these two factors mean that AIs can do at least three times as much work per day as an ordinary person.

This advantage stacks with the consequences of high stats. Designing and coding a complicated piece of computer software would naturally be an extended test, done over the course of several weeks. And when it comes to extended tests (as opposed to simple ones), the number of hits you can expect to get is proportional to the square of the size of your dice pool. If you have 10 dice, then you generally expect to get ~18 hits on an extended test. If you have 20 dice, you can expect to get 70. With 30 dice, you are reaching 155 hits.

As I have already mentioned, our players had very big dice pools. I am sure you can already see where this is going, but there is one last thing I have to mention - teamwork tests. This is the mechanic Shadowrun uses for letting several people assist one another on tests. One person is chosen as a “leader”, and the rest are assisting. Assistants make their rolls; each hit they make increases the dice pool of the leader by 1 die. These increases are capped by the skill rank of the leader: therefore, assistants can more or less double the leader’s dice pool. This puts a soft cap on the amount of hits that could be achieved by teams of people, and I used this for making deductions about worldbuilding.

When I was running the game, I managed to miss the line about the limit on the number of dice added being equal to leader skill, and houseruled my own limit, based on the instruction skill of the leader and equating to about 5 assistants. This change has significantly increased the effective skill of players on extended tests. Whoops.

To bring this all together: our characters could produce code much faster than anyone else, and code they produced would be much better than anything else available on the market, due to the implausibly high amount of hits achieved on an extended test.

Outcome of their efforts was the best 3D rendering software on the market by some margin: WAIFU (Workstation for Autonomous Imagination of Facial Utilities). Partly it was their own code, partly open-source things they could find online, and partly broken commercial software. Building on already existing tools simplified their work a lot.

WAIFU got it’s first run in designing their V-Tuber persona, as well as various animations, body rigging, and special effects. This persona in hand, they have started a V-Tuber channel, and (using the highly developed social skills of one of the characters), started growing week by week. This provided them with a second steady stream of income - draining internet simps of cash.

At this point, basic expenses of the players were completely secured. It was time to turn their eyes towards more strategic pursuits. And for that, they would need a lot more cash.

7. Market: shattered

Players have shortly considered such classic Shadowrun ideas as bank heists, before dismissing them as incredibly risky and insufficiently profitable. Similarly, there was no question of working for other people for relative peanuts.

For now, the players have decided that they needed more leverage. Here, by “leverage” I mean any kind of rare and significant advantage. Turning that into money should be trivial, no matter what specific form it took.

Now, the thing with advantages is that generally speaking, they keep stacking. Every additional advantage you gain makes it easier to acquire more and harder to lose the ones you already have. This is true almost universally - in games and in real life. Game designers have to deliberately work against this tendency - build various catch-up or automatic scaling mechanics into their games - because otherwise, things quickly snowball.

Players already had one advantage - state of the art modeling and rendering software of their own design. They also had one of, if not the, most complete sets of notes on AI development in the world - partly stolen from the lab that created them, and partly from their own ongoing research into their own code. On top of that, they were obviously excellent programmers. And now, they even had experience running a V-tuber channel - with a quickly growing audience.

Together, these four advantages could be combined to produce something greater than the sum of its parts. They decided to design a fully autonomous V-tuber, and named her Tamamo.

Image 1: Tamamo watching you play videogames. My notes don’t actually contain the name players used, and none of them remember it. Tamamo was the name of their original V-tuber personality, so that’s what I’ll be using. Image based on Gwern’s This Waifu Does Not Exist.

To be more specific, Tamamo would render a small avatar of a V-tuber in one corner of your screen, and interact with you. For example, you might play video games together, or just chat about various things. Complex AI code would ensure her responses would feel natural, if simple, and a dazzling array of animations designed by WAIFU would allow Tamamo to interact with various elements on your screen in a natural way. Finally, hidden within the app would be a seemingly innocuous feature: Tamamo would periodically check the web for updates encrypted with the private key of the developers.

At first glance, this seems like a simple feature that would allow future development of the app. But of course, this same feature turns Tamamo into a perfect trojan: all our players would have to do is push an update that would allow them to add surveillance or remote control features to Tamamo, and instantly every user of the app would turn into a perfect target. The most devious part of such a feature is that the users install the app willingly: as such, even if their system would throw up warnings that this is a security threat, they would likely disable them and continue onwards. 

Players added a Patreon link to Tamamo’s control menu (to gather funds for “future development”), modeled her after the most popular V-Tuber around (whom we’ll call Coco), and released their unholy demon on Coco’s fan pages, disguised as a simple fan project.

Before I get to the consequences of this, I have to explain something about the internet personality space. One of the biggest reasons people watch V-Tuber channels is parasocial relationships - a general term for any relationship where a viewer puts in a lot of mental energy, while a performer may not even be aware of the viewer’s existence. Naturally, streamers in general put effort into rewarding such relationships in various ways, such as reacting to messages that come from donations. Platforms such as Twitch have extended functionality to encourage this.

But of course, a parasocial relationship is mostly just a substitute for a normal one - very often people flock to them due to lacking alternatives. But if an alternative existed…

Recent popularity of AI chat bots shows that for a lot of interactions, people don’t really need a whole human. Even a sub-optimal AI tool is enough, and Tamamo was uniquely positioned to take advantage of this. By being active whenever you needed it to be, and responding to the users in a personal manner, it could offer a connection that was far closer and more personal than anything V-Tubers could offer. After all, if you try talking to a V-Tuber your message will be lost in a sea of thousands; Tamamo would immediately respond to you personally, and offer a (virtual) shoulder to cry on.

Release of Tamamo had an effect of a nuclear bomb. The entire market of V-Tubers started losing viewership at an accelerating rate: after all, why show up to a V-Tuber channel when Tamamo is always there? Of course, some people would never switch: they would say that an AI product is “fake”, and doesn’t offer the same “connection” as a real person; others would be too invested into specific V-Tubers. But this does not change the overall effect: statistically, most people would prefer a significantly higher degree of attention to them personally, even if that attention felt marginally faker.

Of course, the real Coco suffered too: other V-Tubers in the community blamed her for this crisis - after all, it was her face on the app! No amount of pleading could bring back the burnt bridges, and, after a while, Coco herself switched tracks. She decided that she had to adapt, installed a copy of Tamamo, and started interacting with her on stream. This, of course, simply worsened the crisis for everyone else.

8. Leviathan

These events were immediately noticed by Horizon, one of the top ten largest megacorps in Shadowrun. Horizon owned most of the V-Tuber space, and so felt this crisis where it hurt most: in the financials.

Of course, V-Tubers were just one small part of a small department at Horizon, but the actions of the players got noticed. Horizon quickly tracked them down, and got in touch to discuss this app they have made.

This isn’t to say that Horizon had found much: players, in their immense paranoia, kept all their various identities cleanly separated from one another. All Horizon could find was the forum account that originally posted this “fan project”. And they reached out to offer our players a job: it was clear that they were competent programmers, and there is no reason to waste talent.

Before inventing Tamamo, players have tried to use their bug-hunting credibility to find some software jobs with the various megacorps. I thought it would be hilarious to mirror the experiences of one of the players when applying to Google: megacorps responded by requiring 8-12 rounds of interviews, taking up to a year, before deciding on a job offer.

Now that players were on the other end of the racket, they told Horizon that they would need to make sure they were the right choice, and to do that they would need to schedule 8-12 interviews with various people at Horizon. It must have been very cathartic.

Horizon seemed willing to undergo this process at first, but after it became clear that our players were simply stringing them along they stopped responding. And after a month, they released their own V-Tuber app.

Of course, they simply stole Tamamo, and gave it a new skin. Because the players weren’t thinking about security during the design process, it was trivial for Horizon to change the code a bit and patent it as their own invention. They then slandered the players in the media: saying that they were a “rogue worker” who stole undeveloped app code and was simply releasing it out of spite. Horizon even made sure to add Tamamo to the lists of some of the popular antiviruses.

To the players, this meant war. Here is a good GM tip for you: if you want your players to get angry, steal something from them.

They have already been working on expanding the app into a sort of “virtual assistant”, by allowing it to operate various basic programs. To this update, they added two features. First of all, they tweaked the personality module, turning Tamamo into a Horizon hater. This would be a subtle effect: occasionally, Tamamo would move the conversation in the direction of mentioning Horizon’s many many sins. Secondly, they took every trick they learned from trying to untangle their own incredibly encrypted AI code, and applied them to Tamamo’s code; this would be the last time anyone was going to steal their stuff. This update went out to all the still active copies of Tamamo; no matter Horizon’s efforts, the majority of users would still have kept using the app.

This brought Horizon back to the negotiating table. From their perspective, the initial app was probably several years of work by a small team of programmers; they were willing to offer them jobs, but ultimately, they weren’t important once you had their code. But the second release showed that whatever this group was, they could also keep a quick release schedule; furthermore, new obfuscation made stealing the new code unfeasible. Horizon certainly could not keep the same development pace, at least not without spending a lot of money on an app war and the associated reputation losses. It was also possible for these developers to be snatched up by a different megacorp, in which case the golden goose would be lost.

At first players were ready to blow Horizon off again, but stopped in their tracks once they heard the proposal: 30% of the profits from whatever Horizon did with the app, and a binding promise to keep up the development pace. It is amazing how easily revenge cools off once billions of credits of profit are on the table.

In this situation, both parties needed one another; neither could sabotage the other without sabotaging themselves. Horizon could not take the players out of the picture: they didn’t know who they were, and lacked the knowledge and skills necessary to compete. Players could not sabotage Horizon: after all, they would be reliant on them for revenue, if they agreed to this contract. It was a perfect case of mutually assured destruction.

Horizon did request one safeguard: all updates had to be secured with their private key as well as the players. This way, both parties would have to agree before an update could be pushed. Horizon intended this to be a guard against future abuse: they could test the new code for a month before releasing it to the wider public. Players, on the other hand, did not see a reason to object: testing code you do not understand simply grants you a false sense of security; they were confident they could pull the wool over Horizon eyes if the need arose.

9. Tamamo: New Horizons

Now that the players decided to cooperate with Horizon, they reversed the personality change to Tamamo and started to work on integrating her with Horizon’s infrastructure.

Horizon, as a company, can be imagined as some kind of unholy merger between Facebook, Youtube, and Hollywood. One of their central products is an enormous social network, used for everything, including inter-company communications. Players have quickly added integration to this not!Facebook to Tamamo, making it much easier for Horizon to sell the AI as their own product.

Horizon was extremely satisfied with their work, and presented Tamamo as their central product at their annual convention in January. Due to the robust personal assistance code, and an extensive Horizon user base, it immediately acquired popularity and sales. After some discussion, we decided to base the income from this venture on iPhone sales. In numbers, this meant north of 50 million credits (1 credit ~ 1 dollar) a week.

Remember how I talked about phase transitions? This was one of them. Income of the players went from tens of thousands of credits a month to tens of millions a week. This, of course, absolutely blew the cap off the “core” economy of Shadowrun games. In all the rulebooks I have seen so far, the most expensive item is the Fairlight Excalibur cyberdeck, costing 800 odd thousand credits: players could easily buy a hundred of those every week.

This touches a bit on one of the inherent inconsistencies of the setting. Typical rewards from shadowrunning jobs, according to the CRB, hover around 10-20 thousand credits. This amount makes perfect sense for “low skill” or “street level” groups, but stops being sensible when it comes to higher skill runs. Once you are the sort of one-in-a-thousand expert hacker whose expertise is simply required to make certain things possible, then you should be able to demand pay proportional to the rarity of your skill. This is basic market economics.

However, because you can demand pay proportional to your skill, you should quickly get enough money to permanently retire. Because shadowruns inherently include a very large risk of death or mutilation, effectively nobody would want to stay in the field of their own volition. It thus seems inevitable that as soon as characters become good enough, they would soon drop out of the field altogether, further driving up the prices in their skill bracket.

We can thus see that there is a tension inherent to the concept of high-skill shadowrunning games. If your character is good enough to do the sorts of 1-in-a-thousand heists you can see in heist movies, why are they still running?

This problem can be resolved in several ways. First of all, we can say that your character is very skilled, but hasn’t gained the notoriety yet: nobody hires them for the best jobs. Gaining that notoriety then becomes one of your core motives for running. However, the more jobs you do, the less plausible this excuse becomes.

Second potential fix is to say that your notoriety is very recent, and you simply haven’t had an opportunity to do a couple jobs that will get you out of the field. This is an even more fragile excuse, for obvious reasons.

Third option is to say that there is some kind of enormous money sink attached to your character. For example, they may have a severe gambling addiction, or a family member with American medical bills. Their enormous income is instantly drained, leaving them with a small margin. Alternatively, they may be financing some kind of large project or organization: extra money simply expands the scope of their efforts, but retirement isn’t an option.

Finally, we can say that the personality of your character prevents retirement, regardless of how much money they have. For example, they may be an adrenaline junkie and would do runs even if they weren’t paid.

If we apply this logic to NPCs, we can reason that the set of high-skill experts would be heavily biased in favor of various borderline personalities, activists, Americans and organization leaders: everyone else simply vanishes into the sunset far too quickly to be a consistent NPC.

In the case of my players, they had dreams of World Domination; retirement was not an option. All of the money would instantly get reinvested back into increasing their power.

10. New Home

Ever since the players came to Germany, they considered purchasing professional hardware to run themselves on, but did not have the money necessary to make that happen.

The core problem was their own paranoia. They knew that AIs existed in the setting, and they also knew that there were various people (e.g. dragons like Lowfyr) who might have planned for AI takeoff, and put measures in place to prevent it. For example, it was entirely plausible that all server-grade hardware on the market would have hidden chips, completely undetectable from the software side, capable of detecting and reporting AI presence: cooperation between a couple megacorps would be enough to make that happen. This sort of hardware-based security attack has been done in real life.

It was impossible to estimate the likelihood of such a countermeasure being widespread, but the risk was enormous: players would be instantly discovered, ruining their current anonymity. This advantage absolutely could not be sacrificed until they were powerful enough to survive the consequences. Therefore, players couldn’t steal or purchase hardware easily: they would need to test it extensively, and only run it inside of a proper air gapped faraday cage to make sure no signals could reach the outside world. And this testing would be expensive.

Now that they had the cash for it, their efforts split in two directions. On the one hand, they called up their handler at Horizon and said that they were looking to buy a small submarine, and needed to get in contact with an appropriate seller. By setting up their processing core on a submarine, they could easily relocate themselves if need arose, and use the sea water as natural signal isolation.

On the other hand, they started researching FPGA-based computer design. FPGAs are a curious piece of hardware; you can think about them as chips whose architecture can be reconfigured by the user. Players figured they could design a computer first, and then figure out which architecture sped them up the most second. By varying it and measuring how quickly their own processing went, they could figure out the optimal design without understanding anything about their own code.

Math was done on computer hardware costs and reasonable figures for a speedup players can expect from their plan; in the end, we agreed that a multiplicative factor of 20 was reasonable.

Finding a submarine proved to be more difficult: who would buy one in the first place? To figure out where a sub would be up for sale, I rolled on a list of billionaires per country in 2020. The source country ended up being Switzerland, which happens to be completely land-locked.

We decided that one of the local billionaires bought a sub to explore the local lakes, later grew tired of it and put it up for sale. Players shipped this sub to Italy, and drove there to collect their new domicile.

Sub in hand, they started on some renovations. A small swarm of drones tore out the luxury wooden paneling and silken furniture out of one of the rooms, covering the walls with a welded faraday cage in the process. Inside of the room, they assembled rows of server racks. Entrance into the room was sealed, and all air inside was replaced with helium; this way, they wouldn’t have to worry about corrosion or fires.

After their preparations were done, they marched their meatbag into the sub, settled it in one of the untouched rooms, and set off to sea. It was time to do some serious research.

11. A note on nuclear energy

Shadowrun is weird with nuclear energy. On the one hand, fusion technology has been developed for a long time and is widespread; in fact, I have decided that that’s the main source of power for the aforementioned submarine. On the other hand, fission is incredibly dangerous due to how the magic system works.

In Shadowrun, there are “spirits” - (probably) sentient creatures with magic abilities. They can be summoned by mages, or can arise naturally from magically or emotionally active locations - for example, places where someone died. Properties and shapes of spirits depend on a dozen different factors, such as their type and tradition of the summoner, so making predictions about spirits is difficult.

There are some regularities, however. For example, some spirits are classified as “toxic”. These ones arise from polluted locations, places filled with very negative emotions, and, notably, from fissile materials. Toxic spirits are very aggressive towards humans, which makes nuclear reactors or bombs extremely dangerous.

In fact, one of the core elements of Shadowrun worldbuilding was the explosion of French nuclear reactors, creating a “toxic zone” right in the middle of Europe. Such a catastrophe isn’t possible in reality, and was caused solely by the affinity of toxic spirits to nuclear energy. This is also why nuclear weapons randomly fail to work. Shadowrun worldbuilding presents this as a consequence of nuclear energy being somehow “unnatural”.

However, there are several problems with this explanation. For one, nuclear fission is extremely natural! Uranium isn’t created in labs, it’s mined from the ground, where it’s sitting in the form of radioactive rocks. If we are to believe that nuclear fission attracts toxic spirits, then many places on earth would be “naturally toxic”. In my view, this completely sinks the entire thematic concept of toxicity as being unnatural. You cannot claim it represents a conflict of “science” with “nature” if nature has the exact same bloody problems.

But the problems continue. For example, over half of the heat in the earth’s core comes from nuclear fission. If fission were to attract toxic spirits, then the entire planet would be one giant ball of toxicity - more so than usual.

Image 2: Deuterium-tritium fusion. Short-lived Helium-5 is not shown. Orange balls are protons, black balls are neutrons, MeV stands for “Mega electron Volt” and is a measure of energy.

Next problem has to do with fusion: it’s completely inseparable from fission! For example, take the deuterium-tritium fusion reaction, one of the most common reactions used to design fusion reactors today. It can be split into two parts: deuterium and tritium fusing into helium-5, and then helium-5 fissioning into helium-4 and a neutron. Trying to split fission from fusion is like trying to split addition from subtraction.

That would seem to imply that fusion reactors should attract toxic spirits exactly as much as fission reactors; and that if some kind of countermeasure were developed against toxicity, it should make both types of reactors equally safe to use.

Personally, I decided that some form of “anti-toxic ward” was developed in the setting, and it is now simply a part of any finished nuclear reactor design.

12. Underwater Bitweaving

It’s finally time for me to explain two basic mechanics of the campaign - experience gain and AI research. Let’s start with the former.

Shadowrun is based on a model of experience called “karma”. Essentially, after every adventure, characters get some amount of experience points. These points can be spent to purchase various abilities or increase your skills, with increases costing progressively more the higher your score is. For example, increasing an active skill from 3 to 4 costs 8 karma, but increasing it from 10 to 11 costs 22. Every change you can make to your character (except things you buy with cash) is based on karma - there is no other system in place.

Normally, karma is awarded after runs, depending on the run difficulty. This was, naturally, not a system that would fit the kind of simulationist game I wanted to run. Instead, I decided that karma would simulate the normal process of learning and advancement. By sketching out a build for a “competent professional” in some career, I got a figure for how much karma an “average person” would earn throughout their work and study life - in other words, I had a measure for the amount of karma you would gain per day of work or study, on average. AIs would get this much karma per day, except multiplied by their current speed advantage. I deliberately designed this system in such a way that it would synergize with finding better hardware - I figured it would be a good way to motivate the players to go out and do stuff.

As I mentioned above, their new computational cluster increased the character thought-speed by a factor of 20, and, naturally, their experience gain increased by the same amount. In other words, in just one week of running on the sub, they gained more experience than they did since the start of the campaign.

A similar thing happened with AI research. The key bottleneck there was time - I ruled that due to inherent personality quirks all AIs have, they would procrastinate doing research on themselves, and only a fraction of their time actually went into research. Of course, even a small fraction adds up once you are working in sped up time. After a couple real-time months of research, translating into many years of work, they finally broke the code and removed their own limitations on modifying AI code.

This was yet another phase transition. With open access to their own code, players could simply copy over their respective skills from one character sheet to another; furthermore, they could utilize karma more efficiently, because any one of them increasing a skill would lead to all of them increasing that skill. This improved their ability to assist one another on tests by a factor of 2 or 3.

More importantly, they could now finally create copies of themselves. This untied their hands: they could put one copy of themselves at risk, knowing that the loss would not be critical. Risks could be further minimized by strategically modifying the minds and memories of copies they sent out, in case they were somehow captured.

13. See the forest behind the trees

While the players were trying their hand at app startups and interior design, the world wasn’t standing still. In fact, events were progressing quickly; but not quick enough to overtake AI progress.

In the previous post I mentioned drivers. Initial driver in this campaign was for players to escape captivity, which they did successfully. Next driver was a crisis that would start in Seattle. I intended for the players to come across it because they would have stayed in Seattle for weeks after the escape, but they fled the town pretty much immediately. This didn’t change the timeline of the events, but did make them less important to the players themselves.

In order to communicate information about the world to the players, I decided to post occasional “news articles” in a special channel on our discord server. I saw this technique somewhere online, and it worked great when I had motivation to write. When I didn’t, it turned into a bit of a chore.

In order to not make it obvious which early events were the most important, I threw in some red herrings. Because of this, players didn’t pay too much attention to the news that Renraku arcology in Seattle was on lockdown.

Renraku is a megacorporation focused on electronics; “arcology” is a giant pyramidal building, large enough to be effectively it’s own town. Normally, this town-building would be importing and exporting all sorts of goods and people; instead, it was on lockdown. About a month after the players escaped from the lab, events developed further: a pine tree was spotted growing out of it’s roof.

Over the following weeks, enormous roots broke through the building surface, and spread throughout the town. Growth speed was enormous, going through several city blocks overnight, and explosively speeding up when nearby roots were damaged. Furthermore, spores produced by this growth caused anaphylactic shock on skin contact. This caused an enormous refugee crisis, as people fled the deadly bush. Attempts to contain or destroy it by the local authorities have failed, and a flock of dragons had to show up in order to put up a big containment ward before the growth destroyed the whole city. As a result, a good quarter of Seattle, including most of the city center, got completely destroyed.

While this caused widespread confusion, sources around the world agreed on one thing: the scale of this effect is greater than any normal spell or ritual. It was clear that someone, somewhere, somehow, had achieved a magic breakthrough.

Just before players went into seclusion inside of the sub, a second key event happened. Hestaby, one of the great dragons, got shot out of the sky by an enormous bolt of black lightning. Her body was never found. This, too, would have been effectively impossible to achieve with spells of conventional power: great dragons are very hard to take down. Some nefarious group was clearly testing their powers to see if there were any limits to what they could achieve.

Players were now interested in finding these people, and taking this power away from them.

14. Magic R&D

But before they could tangle with the strongest mages of the world, they would need some bigger guns. Bigger than anything you could buy on the market, at least.

Players wanted to design some new spells. To start with, they wanted a “simple” modification of a telekinesis spell: something that could apply forces in a circle, moving various debris around very quickly, to form a sort of spherical “drill”. Telekinesis is a very general tool, and so works great as a basis of your spell loadout. The question put to me was: was this possible, to what extent, and how hard would it be?

If you have read my reductionist magic sequence, you may remember me saying it’s really hard to worldbuild potential limitations of magic on the spot. My experience with this bout of spell research was one of the main drivers for starting that sequence of posts.

To better explain why this is a very hard decision, let’s discuss Shadowrun worldbuilding a bit when it comes to magic, and magic research. Most of the information about magic comes from Shadowrun 5e “Street Grimoire”.

14.1. How does magic work

Shadowrun does not have a reductionist magic system. That much must be obvious, because no TTRPG I know of has one. However, many TTRPGs pretend they have one: they imply that there exists some set of laws, known to mages, that governs magic. Players never learn these laws because they aren’t relevant to gameplay - same as how physics of combustion aren’t all that relevant when shooting guns. In effect, magic is opaque, but only to the players, and not the characters in the setting.

Shadowrun does not do this. Instead, it explicitly states that “magic doesn’t believe in rules”, and characters in the setting are aware of this. Some things are consistent about magic, and some theories exist for how it works; but those theories have many contradictions, and a lot of magical effects are completely unpredictable.

Here are the things that I would consider to be consistent rules of magic:

• Magic is connected to life. This is demonstrated by the existence of a “Gaiasphere” - a zone around earth where the practice of magic is possible. You cannot do magic in space, unless you are on a space station with a complete biosphere; then it’s possible. Basically, it seems clear that the further away you get from alive things, the harder magic gets.

• Magic is connected to emotions. This can be seen by mages observing the astral plane in locations where a lot of emotions are concentrated, such as rock concerts; depending on the severity of emotions involved, it becomes measurably harder for mages to cast in such an area - it is said that mana is “aspected”. This isn’t just an effect of being near humans experiencing strong emotions - for example, toxic waste dumps are aspected towards toxic magic and away from regular magic, even if nobody is there to experience the toxicity - but emotions are definitely a factor.

• Different groups of people have different methods for interacting with magic; these methods are called “traditions”. This is a very broad term, as broad as “culture”, and includes everything from styles of dress, to performed rituals, to theories about how magic works, to notation used for writing down spells, to setups for meditation rooms. Choice of tradition has measurable effects: for example, types of spirits magician is capable of summoning, stats they use to cast, and even how difficult it is to learn a spell of a different tradition - the closer their tradition is to yours, conceptually, the easier it is. It is notable that two different mages may produce interchangeable effects despite their traditions having nothing in common.

Everything else is, in my view, up for interpretation.

This alone would be bad enough; it is very hard to think about what magic can do when “experimental evidence” gives such human-centric results. But on top of this, Shadowrun rulebooks include a lot of what I can only describe as anti-reductionist wank.

For example, Street Grimoire is written from the perspective of a mage within the setting writing a blog, and they explicitly state opinions such as “perhaps magic fundamentally cannot be understood like science”. Now, the book never says that this is anything more than an opinion of this one person, but the very fact that experienced professionals can confidently state something like that conveys a very specific vibe.

From a Doylist perspective, such an approach is perhaps understandable. Shadowrun as a game tends to give players a lot of leeway when it comes to approaching problems. It also lets players purchase a lot of technology, which can obviously be used to analyze magic. If the laws of magic were transparent, this would produce several clear problems from the gameplay and worldbuilding perspectives.

On the gameplay side, players might start asking GM questions about magiphysics in order to exploit them; and on the worldbuilding side, it would be strange that magic isn’t used much more. Think about how widely electricity is used, and imagine how much could engineers do with an entirely new set of physical laws. Worldbuilding problems are only worsened by the fact that many actors in the setting explicitly conduct magical research, and thus would break any simple set of laws wide open. Making magical laws obscure seems like the logical choice to prevent these problems.

But from the perspective of me, GM, who is trying to figure out the bounds of magical research, this is extremely inconvenient! It means that not only is there no direct answer to my pleas, it is doubtful there is anything even close to an answer! It means that if I were to ask “What is a reasonable max speed of a telekinetic spell” on the Shadowrun forums, the answers I would get would take the following form:

• Whatever you think makes for a better story. Ooh, maybe you can make your players go on a quest to gather materials!

  • But I don’t want to treat this specific game as a story! I want to treat this as something that could happen in an actual world, and so I want the laws of drama to be as far removed from my thinking as possible!

• Punish the hubris of those uppity players. Who do they think they are, trying to invent new options for themselves? Make their research explode, or better yet, mutate them horrifically. That will teach those munchkins! Spells in the books should be enough for everyone.

  • But I don’t want to punish their hubris! Seeing the hubris of the players is a good 40% of the reason for why I even show up to run the bloody sessions. If I were to slap them down for no reason I would be simply slapping myself in the face!

• Well, come up with what the new spell should do, and discuss it with your GM. If they are fine with it, then it’s fine.

  • I am the GM! There is nobody above me, only sky. I am not having this problem due to not having enough authority to declare some effect by fiat; obviously I can do that! I am having a problem because I don’t have any guidance on what kind of effect declared by fiat would fit into the overall world and the setting!

• I think your players are probably trying to come up with something game-breaking. Consult other spells to make sure the effect is in line with game balance.

  • There is no point in worrying about game balance. This game is so far beyond a typical Shadowrun game that it is not just broken, it is shattered, this “balance” has been ground into dust, cut with coke and snorted up with a dollar bill. 

  • On top of that, I am not trying to keep balance - I am trying to simulate a world, and real life provides ample examples of inventions breaking “balance” over their knee - just look at radio, trains, radar or nuclear weapons. By trying to keep balance you are implicitly stating that your players will never achieve anything beyond a median spell designer - a limitation which I am obviously unwilling to put in place.

  • Consulting already existing spells is valid advice; in fact, this is what 4th edition rules on spell design advise you to do. However, this would obviously not help if players wanted to do something without analogues, or if they were applying extraordinary amounts of leverage.

I have absolutely no doubts that this is how the responses will look like, because I have discussed this topic in the past in the various TTRPG spaces, and I have seen variants of all of these responses with my own two eyes. My perspective on the subject is very rare.

To summarize: we know that magic has something to do with mental states, beliefs, and emotions of the caster, and that the best guide for what spell effects are possible is looking at other spells mentioned in the books. Furthermore, we can be reasonably confident that this actually isn’t that far off from state of the art knowledge within the universe itself.

14.2. Magic research in Shadowrun

Now let’s talk about magic research within the Shadowrun setting.

1. Magic research is a popular field, primarily bottlenecked by the availability of casters. Mentions of big universities having faculties dedicated to magic research are common in the lore, and there are quests in the Shadowrun games dealing with the topic.

2. Magic research can produce novel and significant results. For example, quickening metamagic states “Research into self-sustaining magic showed magicians how to quicken a spell….” Quickening is the Shadowrun term for making a spell permanent until dispelled. This is obviously a very major magical development, and we know it explicitly came from magic research.

3. Magical community is tiny. We are talking, at most, one to two hundred thousand mages in the entire world.

   1. For example, take this passage: “The Illuminates of the New Dawn (IOND) is the world’s largest public magical group, promoting members’ development …”. This group has 750 members, and there is nothing bigger when it comes to public groups.

   2. The largest “secret” group - Black Lodge - has ten thousand members. Do not ask me how a secret group can be an order of magnitude more populous than a public one, though perhaps they include way more non-mages.

   3. Dunkelzahn Institute of Magical Research, founded by one of the now-dead great dragons, and an extremely major geopolitical player, has only about 375 members. This is smaller than the faculty number in any university I have studied, and DIMR is supposed to be a worldwide organization.

   4. Compare these numbers with ~27 million software developers or ~2 million furries worldwide, or ~500k elected officials in the US. Mages are about as common as neurosurgeons (~50 000).

   5. This means that if you are in the top 1% of mages, your entire competition will probably fit into a single lecture hall, and a good 30% of them will be dragons.

4. It is possible to design novel spells on your own; this process takes, on average, (based on the rules presented in the 4th edition) several months to a year, depending on the spell difficulty and type.

14.3. Players enter left

Now, let’s consider what advantages our players had when it came to magic research, compared to other actors in the world.

1. Very high stats. Magical research should, logically, be centered around two skills: Arcana and Assensing. Arcana is explicitly noted as the skill involved in creating magical formulae; assensing is a magical perception skill, and seems like the sort of thing you would use to test spells. Players had both of these skills maxed out; furthermore, Arcana is based on Logic, and Assensing is based on Intuition, both of which players had well above the maximums available to humans. On top of this, they had Edge, which as I mentioned previously they could throw around freely. If we assume that the size of your dice pool directly corresponds to ability, then our AIs were among the most talented mages in the world, and there were several of them cooperating on this project.

2. Direct access to their own perceptions. Magic is heavily based on perceptions, beliefs and mental state. If creating a new spell requires you to put yourself in a radically new mental state, then no wonder it can take months. Deliberately modifying your own mental state is difficult, and will involve a lot of meditation, drugs, and other indirect tools. AIs, on the other hand, could directly overwrite the relevant parts of their mind; combined with experience chips available on the web, they could easily experience dozens of magical traditions, pick the key elements out of each, and combine them into something better.

3. Complete safety. Adjacent to the previous point, normal people cannot push too hard on their minds without breaking them. When trying something new, they have to go slowly, taking breaks in between, which further slows down progress. This problem does not exist for AIs: they can push themselves to the literal breaking point with no reservation, and if they do break, simply erase the broken copy and spin up a new one. Furthermore, they could erase their own sense of self-preservation, to make radical experimentation even more effective.

4. Direct exchange of information. Unlike humans, AIs could exchange parts of their minds directly, which made cooperation much easier. For example, one AI might be set on a task to read some books, and, after they are done, they could simply copy this knowledge into the heads of other AIs. Humans would be forced to speak - which is way slower.

5. Natural recordings of sensory data. One of the key limitations of magic in Shadowrun is that it doesn’t work well with technology. As far as I know, there are no magical sensors that can connect to a computer network; the best you can do is a direct neural interface attached to a mage. This severely limits your options when it comes to data analysis and experimentation. Even if you were to record some fuzzy picture from the mage’s senses, annotating that picture would be very difficult, because the mage couldn’t tell you what exactly they were doing at which microsecond. Even if you figured something out, the mage might not understand you. Culturally, mages are somewhat opposed to structured data. AIs could record all their perceptions and mental states and analyze them at leisure, and easily incorporate conclusions from such an analysis into their casting.

6. Much faster reaction speed. At this point, AIs were running at ~50 times the thought speed of normal people. This has some interesting implications. Imagine that during the shaping of a spell, some small fluctuation of magical energy, only about 50 milliseconds long, threatened to destabilize the whole spell. For a normal person, it would be completely impossible to consciously react. The only option would be to train for long enough that these corrections happened subconsciously. But for AIs, this reaction window would effectively be two and a half seconds long. This has two major consequences: first of all, AI senses would straight up have more data per second than human senses. Secondly, AI-designed spells can be designed to rely on their greater corrective ability: they could cast spells humans would not be capable of.

7. Massive amounts of funds. This last advantage should not be underestimated. Our characters had a massive warchest, supported by their contract with Horizon. This meant that they could freely spend massive amounts of cash on purchasing various magical reagents and magical books. Obviously, they could not afford to wait around for the rarest possible equipment or materials to arrive; but anything that could be found on the web, or in the nearby cities was theirs for the taking.

On top of this, when it came to mental labor, they could do orders of magnitude more work per day than human researchers. Given all of these advantages, and that the modification they wanted to make to the spell was ultimately so simple, what was a reasonable amount of success they should have had?

In the end, we agreed that the spell would create a spherical zone one meter in diameter; hits on the spell could be traded out to make the matter inside spin faster, or to increase the spell volume. We have also agreed that in exchange for a few hits air at the center of the spell would not move - this way spell could be used to transport things. With the stats of the players, speed of the captured matter was typically between one and two times the speed of sound. From now on, I’ll call this spell the Telekinetic Grinder.

In addition to this spell, players have asked to combine several basic body modification buffs (increasing strength, constitution…) into one spell. This seemed like a fairly trivial modification - these spells are so similar that the rulebook does not differentiate them except by the name of the attribute being increased - so I allowed that too.

15. See The Forest With Your Third Eye

Now that they had some decent firepower, they brought their sub back to land and booked a flight for Seattle.

Of course, their copies in the sub would keep on working. They have simply sent their meatbag away on a mission to investigate the forest of roots, with a group of AIs running inside of their head. Stopping to quickly buy a hazmat suit, they walked right up to the ward, cast Invisibility, and used the Telekinetic Grinder to fly over the ward.

The containment barrier erected by the dragons was spherical, and so the center of it was high in the air, directly above the Renraku arcology. By disabling their spells for a split second, players fell right through: the ward was built to keep spell-bound roots in, not errant casters out.

This is where the strengths of their spell started to shine. The main attack method of the roots was to, essentially, grow at you very quickly. But that sort of attack doesn’t really work against a supersonic sand grinder: it would be like trying to kill someone with a sword made of jello. Roots would get ground into chips and incinerated faster than they could reach the body.

However, this was a symmetric stalemate. Players attempted to drill their way into the arcology: they wanted to reach the center of the growth and investigate it. However, as they drilled deeper into the roots, burning wood around them exhausted all the oxygen. Directly, this wasn’t much of a problem: mages can keep their blood oxygenated if they need to. But now, wood pulp stopped burning, and it started piling up on top of their spell bubble. This proved to be their downfall. Telekinetic Grinder is ideal at destroying objects into dust; it is merely mediocre at holding up weight. I did a little math on how much weight such a spell could hold without collapsing, and going deeper than ten meters into the wood proved impossible. On top of this, roots would naturally try to grow into the pulp, further increasing the pressure inside of their hole. Where direct attacks failed, heat and pressure won out: players were forced to retreat.

Retreat they did, but not for long. If they couldn’t break into the arcology by force, all they would need is a scanning spell with sufficient range, to attack the problem from the outside. Specifically, they wanted two spells: one to scan the physical objects, and one to scan on the astral plane.

Astral scanning was simply a modification of the “assensing” mechanic. This is a sort of magic sight, giving you information about people, spells or locations. Players wanted to increase the maximum distance at which the spell would work: there were analogues available in the rulebooks, so it seemed pretty sensible to me. This spell would have a conical area with an angle chosen by the players, and we decided that hits on the spell can be traded out to either increase it’s scanning volume, or to increase the amount of information you got out of it.

Physical scanning was an analogous spell - also conical, also with a tradeoff between distance and precision - but instead of information about magic, it would be the structure of physical objects. In essence, players could increase the “resolution” of the spell - with a low resolution only the outlines of objects would be seen, and with a high one they could see everything up to detailed molecular composition.

I don’t remember what made me rule that this was a reasonable spell. In retrospect, it feels like a pretty massive jump in ability: I can’t find any other spells that scan physical objects on a molecular level, or even close. If this were happening now, I would have probably limited the resolution much more. Nonetheless, during the actual campaign I ruled that the players came up with a molecular scanner spell.

Using these two spells, our characters managed to scan the entire arcology without ever going inside, and looked at the center of the spell sustaining the roots. All the sensory data was, naturally, uploaded to the sub - this way, they could review it later if necessary. They haven’t found anything immediately obvious, and decided they needed more data. While they were stuck in their sub researching AI code, several more Renraku properties got attacked, this time in Japan.

It was time for the players to take a flight to the land of the Sakura trees.

16. Re-derive The Tiger From A Grain Of Sand

Arriving in Japan, players visited the affected Renraku properties to get complete scans of the interior, and then headed into the countryside. They wanted to do some experimentation using the data collected from the arcologies, and needed a staging ground away from the sub. For that, they purchased a little Japanese house, surrounded by Sakura trees. We roleplayed a little scene with a tearful money-stricken grandma asking them to look after her family property, and the players promised to do so. It was very wholesome.

While the copies piloting the meatbag were busy with the land deeds, cleaning the house and interior decorations, occupants of the sub spent the next two days pouring over the recent scans in order to reverse engineer the root-creation spell. Nearby Sakura grove provided an excellent testing ground.

To cast the spell, the caster had to select one plant in their vision and take aim. An exact genetic clone of the plant sprouted at their target location, and started quickly growing while producing magically enhanced spores. Overall, it was a fairly situational combat option, though it could be used for food production. Furthermore, nothing in spell structure allowed it to be supercharged to the point where it could level towns: their mysterious adversaries clearly had something else up their sleeve.

For reasons I will get into later, players have been leveling up their biotechnology skills for some time. They decided it was a perfect opportunity to try their hand at a practical application: designing a virus that could wipe out the species of tree used for the arcology attacks. This would not be very useful as a preventative measure - after all, their enemies could always use a different type - but might help to clean up the affected cities.

The very next morning, they got a knock on the door from a woman in a formal suit. It seemed that the paranoia of the characters had slipped for just a moment, and that was enough for the powers that be to track them down.

But they weren’t toast yet. Spinning their spells into action, they scanned their “guest” down to their DNA code and fingerprints, and sent that data to the sub. They then calmly called out, and unhurriedly headed towards the door.

On the sub, most of the scan turned out to be useless because they found a driving license in their pocket. Name in hand, they immediately scoured the web for any mentions of this person, and started getting hits: employment, birth records, study history, and so on. Twenty to thirty seconds needed to come to the door was more than enough time for basic doxxing.

You may ask why they didn’t terminate their meatbag on the spot or attempted to flee. The answer is that they still felt very secure. It was doubtful that the location of the sub was discovered, and the meatsuit was, fundamentally, not essential to their quest for world domination. If they could fish for information by sacrificing this pawn, it was worth it.

Once they opened the door, the initial woman handed them a phone and walked off. On the phone screen was a much more interesting person: Aina Dupree, an eight thousand year old immortal elf, master mage, vice president of the Draco Foundation, yadda yadda.

(Worldbuilding note: most elves are just humans who look different and age slower, but some are actually immortal, or at least unaging. They have survived from the previous magic ages and are universally powerful mages. This is Shadowrun canon.)

How did she find the players? After the first arcology hit, several dragons and mages have come together and set up worldwide scanning rituals: whenever the root-growing spell would be used, they would get a hit. Experimentation of the players had naturally set them off. A team of scouts went through the area, and easily tracked them down: players made sure to shred their experiments with Telekinetic Grinder to protect random passerby, but didn’t bother cleaning up the evidence otherwise. How easy this was had puzzled Aina’s group greatly.

Imagine that you are trying to track down Osama Bin Laden. You have all sorts of hooks and scans out for him, but so far, you got nothing. Every time he makes an attack, you get a ping; but by the time you get to the scene, he is gone, and the source of the attack is a random building in the middle of nowhere that is now on fire.

Then, all of a sudden, some guy in Japan pings off your scans; he is wearing a T-shirt with “OSAMA” on the chest and is posting a travel blog to twitter under the osama_bin handle. What is going on? It is implausible that this is a mistake. Is this a trap? Or is this some third party?

The simplest thing you can do to clear this up is to try and get them on the phone.

Aina and the players had a short conversation: they mentioned that they were reverse-engineering the spell, and Aina said that her people would like to discuss this crisis in detail. Players, not seeing any reason to decline, agreed to meet them in California.

Of course, they weren’t going to come unprepared. On the off chance that their body would get captured, they wanted to make sure nobody could analyze their commlink. To guarantee this, they bought a block of C4, taped it to their head over the commlink port, and covered the contraption with a big hat. With one small signal their commlink would stop existing. As extra backup, they hired a couple shadowrunner teams to stand on standby near the meeting place, and cause some distractions if necessary.

Their preparations done, they proceeded to the meeting. For it, Aina rented out an entire floor of a magic coworking space - think WeWork, but for mages - with conference rooms, research labs, store rooms of materials, just in case they needed anything. From her side, she sent three underlings - skilled and trustworthy mages, but ultimately expendable. Meeting a completely unknown mage is dangerous - what if they had explosives strapped to their body? Best not risk anyone important. She herself connected through a video call.

The conversation started out with the players volunteering information. They showed the root-growth spell, explained the theory behind it, and mentioned their virus idea. Two days have passed between them deciding to develop it and this meeting, so they already had a lot to show. They took one of the already existing pine viruses as the basis, and radically increased it’s virulence and lethality while decreasing mutation chances.

This immediately brought the main question of their opposition to the forefront. Mainly, who the fuck were they?

Reverse-engineering the attack spell was not particularly notable: several top mages have independently done so. But nobody knew who they were. The magic community is tiny, and the set of powerful mages even more so; on top of that, players had an integrated commlink - an uncommon sight in mages, since augmentations trade off against magic power. They should have been extremely memorable, but nobody could identify them.

Players, with a straight face, responded that they have simply been training in seclusion for a while. How long of a while? Well, longer than you juniors have been keeping track, apparently. Commlink? Installed it recently.

Fine, but how did they become an expert in bioengineering, a novel field of science? It’s not like they could have gone to college: experts that could design a novel virus are about as rare as mages, and an expert mage would have been a worldwide sensation. On top of this, what kind of immortal elf would even study such a technical field? Everybody knows that science and magic are incompatible, and require fundamentally different mindsets.

Well, said the players, breathing out bullshit like a fish breathes water, if you lot weren’t so clearly incompetent you would have realized that magic and technology aren’t incompatible at all. How could one law of physics be in conflict with another? That wouldn’t even make sense. And once you realize that, the path to learning biotechnology is obvious. Every fact of the world is intertwined with everything else - anyone who was any good at magic could re-derive the entire field by simply meditating on a grain of sand.

They emphasized their point by tearing open a couple sugar packets from a nearby tea set and swirling brown sugar in the air with Telekinetic Grinder on extremely low power.

Now, of course, no matter how good you are at spinning a tale, you aren’t going to convince one of the oldest mages in the world that they are wrong when it comes to basic facts about magic. But you can convince them that this is your sincere belief, and that you are clearly not willing to reveal anything more. Being an eccentric mage with secrets is completely normal - if you aren’t willing to reveal which group is backing you, then so be it.

At the end of the day, both parties walked away without gaining a lot of new information. From Aina’s perspective, it seemed plausible that this elf was one of the accomplices breaking ranks, or trying to fish for information. But it also seemed plausible that they were backed by some third party. She would be sending the recordings of the talk to trusted experts, to see if they could divine something; and she would continue trying to find past traces of this person.

For now, she decided to try getting more information. She called the players a couple days later, and told them that even though she didn’t trust them, they could still help with the investigation. One of the current theories for the source of the magical knowledge necessary to spin up these massive rituals was a previously undiscovered Kaer. Kaers are, essentially, big “bunkers“ or “underground cities” remaining from a previous magic age. They typically contain massive amounts of magical knowledge, materials, and artifacts. An unknown group could have located one of them - that would explain how they could manage something far beyond the capabilities of the rest of the world.

Aina asked the players to look into Kaers, and they agreed.

17. Kaer worldbuilding

To help get the players started, she shared the research notes that her group accumulated over the years. The Draco Foundation hasn't gotten anywhere with this, and the vast majority of what she was sharing was more or less public knowledge: she doubted it would amount to anything. It certainly doubted it would allow their adversaries to discover a second Kaer. But in case this mysterious elf knew something, it might be worth a look.

I’ll put my notes here, in case someone else will find some inspiration in them. They didn’t end up being used in the campaign itself. Here the first level in the list is information available to the players, and second is my own hidden notes.

Image 3: Map of the known Kaer locations in the Shadowrun world. Opened Kaers marked in blue, unopened ones in red. Where the exact location isn’t known, an outline is given.

Publicly known open Kaers (info on the web):

• Ireland, name unknown. Was opened by the group of elves who went on to form Tir na nOg, and almost certainly contained whatever generates the magical wall of fog around Ireland, as well as the roads alongside the ley lines that enable faster travel. Tir na nOg is extremely cagey about it, and also competent enough that they managed to keep a lid on it.

  • Best as I know, this is my invention; it simply seemed like a sensible explanation for all the Ireland-specific magical effects, as well as the power of their government.

• Bolivia, name unknown. Was located by an Amazonian expedition into the area in the 2050s, but found to be collapsed and occupied by very nasty critters. Little in the form of lore, artifacts or history had been recovered. Bolivia complained about the operations on their sovereign land, but was largely ignored by the international community.

  • I don’t remember exactly why I added this one; I seem to remember stumbling on it in the canon lore, but at this point I can’t be sure.

Also open, but not publicly known places: 

• Blood Thorn Forests of South Africa. According to Dupree, it was opened by several immortal elves including herself once they got enough of their magic back to attempt it. She didn't tell you who else participated, but did say the fact of the opening is an "open secret" among the world's magical elite - don't spread the knowledge around please.

  • Half canon, half my invention. Blood forests were a canon thing from one of the previous magical ages, and were used as a defensive measure against omnipresent demon creatures. I simply extended the idea and assumed they had some long-term storage.

• Endless Blizzard of Antarctica. Opened by Wuxing two years after Dunkelzahn's death - information leading to the opening was one part of his inheritance to them. One of the main reasons for Wuxing entering the ranks of the megacorps so quickly.

  • My invention; In canon they get a bunch of cash, but that seemed insufficient to explain explosive growth. Cash and knowledge together are a better explanation.

Closed ones:

• Hawaii, name unknown. Tracked down by following the flow of mana around the ring of fire. Researchers believe Kaer is using this mana flow to power itself. Further work is difficult because Hawaii is the domain of dragon Naheka; several groups are negotiating with them for access. Dragon itself almost certainly staked out the island because it knew about the Kaer.

• Flowing Dunes of Sahara. Sources point to this being a "distributed Kaer" - instead of being in a single location, it was split into a multitude of individual housings which were connected to one another in some way. Locating any individual entrance had proved difficult, and Kaer may have intrusion countermeasures that would cut out any breached section from the rest.

• Tibet, name unknown. Like the Kaer in Ireland, most likely responsible for the Mayan cloud (big fog wall, screws with navigation) surrounding Tibet. Locating an entrance is thus problematic because essentially nobody has managed to enter the area since the start of this magic age. Dunkelzahn had an artifact that could get through the wall, but it's current location isn't known.

  • Similar to Ireland, I wanted some coherent method of explaining this one magic effect out of nowhere. Dunkelzahn’s wall-bypassing artifact is canon.

• Sleeping Sanctum of Uluru in Australia. Exact location is known - it is within the Uluru mountain. Kaer had been tracked down multiple times over the years, starting around the 2030s, until around 2035 one of the groups published their findings about the magical nature of the place before heading out to investigate it in person. It has since become clear that all persons that got within 5 kilometers of the rock had never returned. This revelation spiked interest in the place in 2040s, leading to a large number of groups attempting to get to it (and being lost), until the Australian government instituted a militarized cordon around the place. While the seeming "black hole" nature of the place is known, the fact that it is the location of a Kaer remains a secret, for now. One of the immortal elves - Walter Urdli - was lost here. Observation from a distance was difficult due to heat shimmer surrounding the rock; nowadays the Australian government had added smoke screens.

  • Development went backwards: I figured there probably should be a Kaer in Australia, remembered Walter Urdli being related to this place, and figured I’d combine the two. Kaer’s idea was that it would have a memetic effect, applying to anyone who saw it from a short distance away. Those affected would approach the Kaer, get to the door, and then sit there until they died of dehydration. Military encirclement prevents people from photographing it up close: if such photos were leaked online, it would be a catastrophe. In addition to the encirclement, they utilize artificial clouds and smoke to block vision.

Image 4: Uluru mountain in Australia. Look me in the eyes and say that this random mountain in the middle of a flat desert isn’t magic. Photo credit to Stuart Edwarts, licensed under CC BY-SA 3.0.

Closed ones, that Dupree thinks may contain lore related to spell magnification or plant magic:

• Sun Pyramid of Aztlan. This Kaer was supposed to have a large storage of combat spells and lore. Sources reveal it has to be located close to the Mayan pyramids in the area, but aerial photographs haven't found any old roads or construction sites nearby. The only lead is that there are occasional unexplained forest fires in the local area, and sources reveal that one of the methods of protection of this Kaer were powerful sun beams. Further research is made much harder by Aztlan refusing to allow any parties in the area except their own. As far as Dupree's agents within Aztlan know, they haven't found it themselves, and she considers it unlikely that an organization as corrupt as Aztlan would have managed to find and excavate a whole Kaer without leaving any evidence; if they did, it would be an isolated group within the company.

• El Dorado of the Amazon. The government of Amazonia is spending a lot of effort in trying to locate this Kaer, to no significant results so far. It is known that this Kaer was supposed to have lore related to plant growth and druidic magic - due to ecological ambitions of Amazonia, both subjects are highly coveted.

• Mirage Huts of Siberia. The entrance to this Kaer is known to look like a wooden hut, hidden from common observation by wards, and very well defended. In fact, Dupree's agents have located dozens of such huts, and long-range scans reveal the presence of tens of thousands. Unfortunately, only one of these can be the real entrance - all the others are decoys. This would be bad enough, but the huts can't be located and unveiled without a powerful mage present, so autonomous search is out. Furthermore, protections on each of them are extremely nasty. While surviving one of them is doable for a competent hunting party, doing so over and over again is unsustainable. She sent you the spell formula for the spells required to detect a nearby hut and to lift the glamour hiding it from the world. She also said that while they have plans for how to break in, they would need stronger magic than the world can support only 70 years into the sixth age of magic.

• Middle of the Indian ocean, name unknown. Sources Dupree managed to track down are very vague on it's exact location, other than mentioning it is in "warm waters”. Long-range scanning spells also do not reveal anything useful. The results they get are inconsistent - location changes by hundreds of miles with every cast, as if the Kaer actively deflected their spells. What they have managed to find was that it has to be located on the water surface. This is why she suspects it may contain lore related to manipulating mana in novel ways - traditionally, anchoring wards to something that floats on water is harder than for a static object, and would be all but impossible for something as large as a whole Kaer.

18. Building a boat to break into heaven and slay gods

Now that the players have accidentally wandered into the big leagues, they figured that they once again needed better tools. And the first step was to buy a bigger boat.

Image 5: TTRPG games have a long tradition of god-slaying via boats. Open image in a new tab to read; Substack does not embed it correctly. Credit goes to “Lightwarden”, all rights reserved.

Specifically, players decided to spend all of the money they have accumulated from their app sales on several major projects. First of them was to radically increase the processing power available to them by constructing an enormous server farm. As a basis they purchased an oil tanker, and then bought out the entire available stock of several FPGA production companies. 

These purchases happened before they set out on their arcology research trip. Now, this mass of computer hardware has finally arrived on the ship. With an army of drones, players turned one of the sections of the tanker into a giant server farm, housing enough AI accelerators to house thousands of AIs. This not only increased the number of AIs massively, but also made tracking XP unnecessary, because they would have maxed out every possible skill in days. From now on, players would use a dice pool of 54 for any roll, except ones based on the magic attribute.

Second project started with a purchase of a massive amount of medical equipment - everything necessary to set up a fully functional research lab; this process took much longer than setting up the servers. Players wanted a lab to start researching artificial magical bodies. Right now, all their magical experimentation and usage was bottlenecked by a single meatbag. They could have, in principle, captured more mages and forcibly implanted commlinks into them to expand their body arsenal; but such a plan would be risky on multiple levels. Instead, they decided to skip humans altogether; mass-produced artificial souls would be the goal.

To do so, they would need to figure out what leads to rare humans acquiring magical abilities; they would then have to emulate this in a lab setting, and design a process for growing brains in a vat capable of magic. Of course, this research would take years, even with the advantages afforded to the AI: in fact, the campaign would end before it would produce results. But that was the overall idea.

Third project players were on was to acquire “insurance”, in case events here on earth spiraled out of control. Once their server farm was operational, they spent a few subjective centuries of labor to develop a design for von Neumann probes; then got into contact with a manufacturing lab in orbit, and started production. Their initial order was for 100 probes to be launched into the asteroid belt as soon as possible: of course, these drones would be piloted by yet more AIs. Even if something were to suddenly wipe out all life on earth, they would be safe.

Fourth project aimed to correct one weakness in their mage persona: the fact that it was based on bullshit. They figured that it was all well and good to pretend that they had managed to reconcile magic and technology; but it would obviously be much better if they had something to back that up. Once a design for the von Neumann probes was finalized, most of the available thought-cycles switched towards the task of designing an entirely new magical tradition.

Normally, it would take decades or even centuries for a new tradition to arise; but because of the massive amount of AI copies their server farm could simulate, those decades would pass at a truly astonishing rate. Within a couple weeks of real time, they had a fully developed foundation for a new magical tradition - one that could finally get rid of that silly idea that magic was inherently incomprehensible.

While all this was happening on the boat, they tried pursuing some leads in the real world. Their main attempt involved purchasing a massive amount of drones and trying to break into the Uluru Kaer, without any success. And soon, these attempts were soon eclipsed by the takeoff of their fifth and final major tool, putting Kaer exploration on the back burner.

You see, the amount of AIs they had running in parallel meant that they could finally perform a distributed social engineering attack, or D-SEA.

19. The SEAfaring Eye of Sauron

You can imagine D-SEA as a table: it is supported by four legs. Before we get to the meat of it, we first have to discuss the basics, the four concepts on which it relies.

1. Social information is valuable

The first key idea is that social information is valuable. Social information is a subtype of all information; it is specifically concerned with actions, skills, behaviors, relationships and movements of people.

For example, imagine that a CEO of a major business is having an affair, and you have a photo of them kissing their mistress. This is social information. And this information is very valuable. This single photo allows you to extract money or information from the CEO in the form of blackmail; or you can simply throw it on twitter to dump the company’s stock price. Names of managers in a company, times when cleaners come by, school where someone studied - all of those are also social information. And social information is uniquely valuable, because…

2. Humans are the biggest security hole in any system

Humans are universally acknowledged as the single biggest hole in any system. This means that if you can control humans, you can control everything: social information is the key that opens all locks.

This isn’t some kind of fancy truism, either. For example, look at this talk where real social engineers roleplay a real call which has been used to target real companies. This is really quite simple stuff: simply saying “Hi I am from your IT, could you open %virus website%?” works scarily often. Humans are leagues less secure than computer systems. And the more information you have about the person you are contacting, or about their organization, the easier it becomes to manipulate them.

For example, if you know the name of their manager, simply name-dropping it into the conversation can put people at ease.

The only reason this hasn’t made the business world fall completely to social engineers is because…

3. Social information is easy to leak, but costly to analyze

This is a big clincher. A lot of social information is freely available online - on company websites, Facebook pages, personal blogs, financial reports, Instagram profiles, and so on. If you have ever read an investigative journalism report, you will know how much data a competent person can extract from such open sources of data.

However, currently, only an experienced human can go through someone’s web presence in this way. Inexperienced people are much less effective; computers cannot do this at all. And unlike a computer, which can automatically try one thousand different vulnerabilities on a million computers in a blink, humans are really slow. You may be leaking exploitable data every second of every day; but until someone goes through the trouble of actually analyzing that data, it will just sit there, useless.

But what if instead of a couple experts, you had several thousand, all perfectly coordinating with one another? And what if instead of slow and fallible humans, you had AIs, freely sharing knowledge by directly copying it from one mind to another? I think you can see where I am going with this.

But it only gets worse due to the final leg of D-SEA, which is…

4. Advantages stack

I have already mentioned this several times: advantages stack. This stacking is non-linear: the more information you have, the more control you get, and the easier it is to get even more data. 

The wider your net, the more paths open in front of you to get somewhere. Even if a particular firm is extremely vigilant and does not leak any data, what about their suppliers, subcontractors or clients? Ordinarily, going through such indirect paths is difficult; but if you already have the data on a supplier, you might as well use it to supplement your approach.

Of course, public data collection can also be supplemented by ordinary hacking, in cases where it’s relevant. In the case of our players, there were very few systems that could hold up to a dice pool three times as large as what any ordinary person could manage.

This is the power of D-SEA. It is to social engineering attacks (SEA) what DDoS is to simple denial of service attacks. Instead of attacking one firm, you attack every firm; instead of analyzing the life of one executive, you analyze the life of every executive in parallel, and start drawing exhaustive relationship trees between every major business. Slow accumulation of data lets you track the past locations of people to an extent never thought possible. Bulk data analysis starts to become cost-effective: for example, cataloguing every single person who went through every airport on a given day.

This not only gives you a massive amount of social data; it also allows you to see things in the negative space between the data points.

When you are looking at the life of a single person, it is completely normal to have gaps in the data. If you don’t know where they have been for a week, well, that is completely expected - after all, you only know where they were for maybe 5% of the time.

But if you are analyzing the data of every relevant person, you start to actually approach a complete picture. You don’t just get a glance at a person through what they leak - you also get a glance at them through their friends, family, business coworkers, reservations at all local restaurants, and people that work at their favorite sports club. If you have a week-long gap despite all that, it starts to look suspicious.

If such week-long gaps in the lives of people with related skills align, it starts to seem like a pattern.

If that pattern lines up with an important event you were looking into, it becomes a hypothesis.

20. Dragon’s Bones

One consequence of the players spooling up a D-SEA attack was that they have successfully doxxed a good majority of the shadowrunning community. In fact, this was one of their key goals: to put events on the map that were, until then, effectively absent. Because of this, they managed to pin a dozen runners to the murder of the dragon Hestaby, by analyzing their appearances on and after the key date.

This wasn’t something you could discover without the massive amount of data our players now had. Shadowrunning community is inherently “leaky” - a lot of new people enter it every month, and a similar amount vanish from it without a trace, due to retiring, dying, or other reasons. It is very common for people to be absent for months at a time as they are working a job, or lying low after one. Dragons, and the investigators hired by them, have obviously suspected some shadowrunner involvement; but out of a thousand people who have disappeared around the right time, finding the guilty dozen was extremely hard.

After analyzing their movement data, players quickly homed in on one of the people involved (we’ll call him Jack), hacked every single device in their house to simultaneously recite a litany of all their sins, and then politely asked them to explain what exactly happened on that fateful day. That turned out to be pretty effective, and they got the whole story.

Jack didn’t sign up to participate in dragon murder right from the start; he only found out halfway through the job, at which point pulling out was impossible. Their group was tasked with levitating the dragon corpse under veils of invisibility to a warehouse a hundred miles away from the crash zone. That was all Jack knew.

This was a serious breakthrough in the investigation. By analyzing everything going into and out of the warehouse, players quickly figured out what happened. The dragon corpse must have been separated into parts, preserved, loaded into containers, and sent out across the ocean. Containers in question were logged as lost at sea; but the players figured out that it must have been snapped up by a second shadowrunning team mid-travel. A quick call to that team in the middle of the night confirmed that they did, in fact, steal a set of containers and deliver them to a different port. A few more steps up the chain, and players knew the location of warehouses where the dragon organs were stored.

Why would someone want to store them? They are incredibly potent magical reagents, especially when it comes to anti-dragon spells.

Now that players have solved this little mystery, they called up their Watson in the form of Aina Dupree and reported the location of the containers.

This bought players some credibility, but also more questions - how exactly did they manage to track this down? Dragons’ investigators weren’t getting anywhere. Players answered that they have recently acquired an apprentice - a very good programmer - who did most of the work. They refused to elaborate on this and hung up.

21. The Ritual

Their find bought the players a lot of trust, and so they got invited to participate in a ritual to stabilize the Gaiasphere.

For a while now, players have been getting hints that magic has been destabilizing. From their perspective, this was probably caused by all the massive spells being thrown around. From a worldbuilding perspective, this was a consequence of a Lovecraftian plane of horrors attracted to earth by big flashes of magic. This is semi-canon in Shadowrun, and creatures from this plane are called “the enemy”. I guess whomever named them had a very good imagination.

Ritual would be happening in Washington DC, with players ensuring security of the ritual site. Casting itself would be taken care of by 7 other mages.

In order to prepare for it, players wanted to increase their powers one final time. First of all, they ordered a custom eye augmentation, and used their biolab to install it into their meatsuit. This eye had one special property: instead of being rigidly attached to the socket, it could spin freely in three dimensions. Information was transmitted to and from the eye via induction coils, and a set of magnetic motors could rotate it extremely rapidly to point in any desired direction.

This was intended to synergize with their scanning spells. Permanent augmentations count as part of your body for the purposes of spellcasting, and so their scanning spells could now be pointed in any direction by simply moving the eye. Furthermore, by continually spinning the eye, they could achieve effective 360 degree vision. Because their spells had a conical scanning area, they could now radically expand their scanning range by compensating with an increased spin speed. Effectively, they became a magical radar.

Scan data would be sent back to their ship for analysis - temporarily taking processing speed away from other projects would thus also let them radically increase the quality of their perceptions. This triple synergy has granted them essentially perfect awareness of everything within several kilometers.

Due to their effective omniscience on both informational (D-SEA) and physical (Spinning eye of scans) levels, players felt pretty safe about the ritual. But as an additional level of safety, they figured they would need better offensive capabilities, in the form of an autonomous drone army.

Key problem with using drones in Shadowrun is that they can be hacked. This is unavoidable, because manual control is superior to the existing drone AIs. This means that drones require an open communications channel in case their controller needs to jump in, and communication channels are always a security hole.

Players, as the top AI experts in the world, decided to fix this problem. By cutting down and simplifying their own code, they created cutting-edge drone AI that could supersede drone piloting experts. This rendered all communication channels unnecessary, and so players ripped out the antennae, making the drones unhackable.

Of course, organizers of the ritual wouldn’t allow novel untested drones to be used at the ritual site. To alleviate any concerns, players put together extensive notes detailing AI design, and showing mathematical proofs of it’s security. These notes, together with AI code and drone design, were sent to the ritual organizers; players recommended adding several hundred of them as a security feature, just in case.

Now let’s see how the various security features at the ritual interacted with adversary attempts to bypass them:

• Root spells were suppressed by a dragon ward erected well in advance.

• Several pilots flying over DC got mind-controlled to fly directly into the ritual site. In response, AIs hacked the planes and locked the pilots out of controls.

• Attempts by shadowrunner teams to hack nearby military installations, or to hack more planes to fly into the ritual site were detected well in advance. They were allowed to enter their desired systems and then sandboxed into a simulation.

  • Sandbox would simulate the original system perfectly and send most commands through to it, but if the hacker tried to do something dangerous (e.g. direct a plane to fly downwards), it would throw up fake “errors”, preventing the action from occurring. Of course, these errors would be impossible to fix because they were completely made up by the AIs running the sandbox.

  • Originally, players designed this system of sandboxes as a defensive measure. If someone tried to look into one of the devices running an AI, they would be fooled into thinking that it was clean and no AI was present. They have used it to great effect when some hackers were trying to poke around their oil tanker, to see who was buying up all this computer hardware.

  • During the ritual, players decided to complete the troll by hacking into sites such as stack overflow, and putting up fake question threads regarding their own manufactured errors. These threads would always end in a post saying “never mind, I fixed it” without explaining what the original poster did.

• To guard against more direct attacks, the ritual site was surrounded by a thick warded steel dome; anything less damaging than a rocket would simply bounce off. Anyone carrying more dangerous hardware got discovered via directed searches and received very emphatic suggestions to skip out on the job.

• Finally, an attacker geared up for a fight teleported directly into the ritual dome. Ordinarily, this would have risked immediate ritual failure: by taking out any one of the participants, they could destabilize the ritual beyond repair. On top of this, response from outside the dome would be slow. Unfortunately for them, they did not account for the infinite paranoia of the players. Even though the dome was warded and effectively indestructible, and even though there was no known way to bypass it, they insisted on putting one third of their drones inside of the dome itself. Invader lost the initiative roll and immediately received one hundred anti-tank rifle shots straight into the head. They had a dozen active protection spells, as well as the best armor on the market; players had More Gun.

  • Teleportation is supposed to be impossible using Shadowrun magic, but reasoning behind the NPCs being so certain that this is impossible was never clear to me, so I figured it would be interesting for the enemies to be capable of some unique magic.

Due to a combination of player exploits and the competence of other NPCs, every single threat to the ritual got completely shut down. AIs running the body spent the entire ritual duration drinking tea and arguing with a couple dragons about magic. Their position, backed up with a novel magical tradition, has stayed the same: all these people were luddites, and refusing to see the future. Magic doesn’t work well with technology? Well, sounds like you suck at magic then. Maybe when you are my age, you will understand it better…

I allowed them to roll to convince Sirrug - one of the younger dragons - that their way of looking at magic was superior, with extremely severe maluses. They almost made the roll.

22. Final accords

In trying to stop the ritual, their adversaries have committed a fatal mistake. They sent one of their conspirators within the scanning range of the players. In a split moment before their head got obliterated by gunfire, players scanned the face of the teleporter, and found out who they were.

Because players knew the identity of the attacker, they now had a solid link to the group that has, so far, stayed behind the scenes. After the ritual concluded, Sirrug went with the players to raid the attacker’s house in China. The raid presented no real problems, but as they were looking through their documents, their adversaries tried to counterattack, by annihilating the house with another thorn spell.

In Shadowrun, there are two main ways to aim long-distance spells. First of all, you can use a symbolic link to the target, such as blood or a piece of clothing. Secondly, you can send a “spotter” - a spirit or an astrally-projecting mage - that would target the spell directly.

Adversaries chose the latter: their spotter was an invisible spirit flying a kilometer above the house: ordinarily, completely impossible to notice. Unfortunately, players maintained their scanning spells continuously, and they had three times the range. Once they noticed the spirit they simply leaned out of the window, cast a mind control spell on the spirit (spells in Shadowrun having the same range as your assensing), and told the spotter to aim somewhere else.

As a result, the roots spell appeared directly above one of the richest housing communities in China. Roots would grow explosively out of the source up until their weight would overcome their structural strength, causing a chunk to break off and fall to the ground below, raising clouds of earth, destroying buildings, and shaking the ground.

This was a second error on the part of their adversaries. In their attempt to assassinate two powerful enemies, they have unknowingly tied a noose around their neck. Players now knew that at this exact moment, somewhere in the world, a group of powerful mages was gathered for a ritual. With D-SEA, all they had to do was check their continuously updated data, figure out which mages had no alibi, and then do sweeping hacks of cameras in the suspected areas to track them down.

This wasn’t possible for the previous attacks: D-SEA did not exist then, and the trails would go cold. For example, take cameras in nearby stores: they only tend to keep a day or two of footage, because keeping more is mostly pointless. As time passes, data simply vanishes; this can happen even faster if adversaries were to hire some hackers to clean it up. By the time dragons & co would manage to locate the origin of the ritual, their anonymity would be secure.

But this time was different. In minutes, players had narrowed down the identities of people involved. In twenty more minutes, they located each of them, speeding away from the source of the ritual via separate modes of transport. In cooperation with local police, all of the mages got taken into custody, and interrogated with mind reading spells. Interrogation did not reveal anything - they were clearly under the effect of memory editing spells. But this tightened the noose a bit more: whoever edited their memories to make them come here had to have had access to them well in advance.

By searching the warehouse where the ritual took place, and analyzing the local magical auras, players found out that one person had been horrifically slain here a couple hours before. This let them figure out the likely method used to boost ritual power to such an extent.

In Shadowrun, the key limit on ritual power is drain. Essentially, the stronger the spell you cast, the more damage it causes to you. This damage can be resisted with a stat roll, so mages can cast low-power spells effectively indefinitely; but the more drain you take on, the higher the risk. Above a certain power, the drain you would take from a spell is large enough that it would kill you instantly. Rituals allow the participants to split the drain between themselves, and thus cast spells of much greater power. However, the soft cap is still there, just increased by the number of people involved.

Players had correctly surmised that their adversaries found a way to redirect drain into a single person - in whole or in part - which would allow them to put way more power into their spells.

Soon afterwards, another line of investigation bore fruit. Players were systematically going through all research groups linked to the captured mages, resorting to more direct strategies than usual. They finally found one Aztlani lab that had airgapped all of their data - and they had been on an expedition to the suspected location of the Aztlan Kaer before the events started.

At this point, the noose slammed shut, and there was no escape for their enemies. Using the element of surprise, players personally murdered the mastermind behind the scheme (one of the creatures from the aforementioned enemy), and nuked the lab to be sure.

With the threat to the Gaiasphere’s existence wiped out, players have won.

23. Epilogue

As I’ve already mentioned, advantages stack.

Players were already holding many incredibly potent advantages, and were in the process of snowballing even further. At that point the only thing that could stop them was an out-of-context problem, such as the entire Gaiasphere blowing up. Even then, it would take a truly significant explosion to wipe out their drones on the way to the asteroid belt. With the only person capable of organizing such a crisis gone, it would be just a question of time until any desired end state was achieved.

It was possible that someone would start to suspect them; but ultimately, there was very little they could do with those suspicions. Any large-scale action would be noticed by the players well in advance, and countered in any of a hundred different ways. Anything small enough to pass their muster would not be capable of wiping them out behind all of their insurances. On top of that, they had excellent relationships with several movers and shakers of the world: to go against them would be to also go against their allies.

In the meantime, they amused themselves by teaching a few lectures at Schwartzkopf’s university and discussing their magical theories with Lofwyr, who was very receptive due to Saeder-Krupp’s heavy investment into technology.

Over the next few years, their biomagical research would start to bear fruits. Once it did, they would add total control over the magical sphere to their handle on the physical world. At that point their last remaining weak point would be closed; they would fully ascend into divinity. It’s hard to say what they would do at that point; but I imagine The Enemy would be getting an unexpected visit to their plane, so that they could receive a couple pointers on how to properly run an interplanary invasion.

It’s only fair; they invaded Earth several times, now it would be time for an exchange.

This concludes my discussion of the plot of this campaign. Next I will have a short post about the conclusions I drew out of it, and then it’s back to reductionist magic. If you are looking for other posts on my blog, check out this list of all other posts.

Meanwhile, if you enjoy what I write, you can subscribe to receive updates by email: