The Tempest Cluster was created to be the setting for my Mothership open table. This is a peek behind the curtain for my setting prep.

When I first sat down to design the cluster, I knew a few things:

• As an open table, the PCs would have a home base — a point from which essentially every session would begin.
• I’d read several Mothership adventures, and had a short list of scenarios that I already knew I wanted to use. (This gave me some guidance what the cluster would need so that I could place those adventures.)
• Mothership requires a setting to have some specific infrastructure to work (e.g., ports for shore leave).

I got started with a short brainstorming session, just listing some cool ideas and broad concepts for star systems and planets that I thought would be interesting (or were dictated by the things I already knew the cluster would need). Then I laid that sheet of paper to one side and grabbed two more blank sheets. On one of these I began sketching jump node maps and on the other I started naming and listing features for specific systems.

I knew I wanted to keep the scale of the cluster relatively small. First, if travel time became too large, it would cause problems with keeping the PCs in sync. More importantly, I know that layering material is more effective than dispersing it: It’s more interesting to put three adventures on the same moon and see what happens when their concepts start bumping into each other than it is to, for example, create a whole new system for every adventure.

On the other hand, I wanted the cluster to be large enough that some stuff would be near to the PCs’ home base and other stuff would feel far away. It helped when I realized that, since the nature of the cluster would naturally constrain the open table, I could place the PCs’ home base at one end of the cluster (in what would end up being the Ariel system) and immediately create a “far end” (in the Verstern system). This is also the origin for the Long Road, the series of dark systems between Verstern and Hajar:

There were originally several more dark systems in the Long Road, but they ended up making travel from Ariel to Verstern to lengthy and I needed to adjust it. (In much the same way that I often let players make adjustments to their characters after the first couple sessions of a campaign, I also won’t hesitate to do some quick setting retcons if we discover something isn’t working in actual play.)

HERE THERE BE DRAGONS

I also deliberately DIDN’T fully flesh out every detail of the setting. For example, I could’ve gone through and said things like, “Hajar has exactly nine planets. Hajar-I is a super-Jupiter. Hajar-2 is a small terrestrial planet. Hajar-3 is an all-water planet, and between Hajar-2 and Hajar-3 there’s a binary pair of dwarf planets.”

Filling in concrete details like this can lead you to discover interesting stuff about your setting, but at this early stage I generally prefer to sketch in enough detail to give everything a unique character — Hajar has multiple asteroid belts; the Ternary is filled with lots of Earth-like planets; Mrachni is a black hole — but leave a lot of blank spaces where I can plug stuff in later.

For example, I’ve recently been reading Joel Hines’ Tide World of Mani and Desert Moon of Karth, a pair of linked planet supplements. If I’d already detailed every planet in every star system of the cluster, I’d either be unable to use these supplements or I’d need to open up a new jump point and expand the cluster. Instead, looking around, I can see that there’s plenty of room in the Laxmi system. (I’d previously placed a different adventure in that system, which established that the two major terraforming megacorps are engaged in a large campaign of espionage and sabotage there. So it’ll be really interesting to weave the politics of Mani and Karth into that conflict.)

Similarly, I also left the precise history of the Tempest Cluster rather nebulous. This is somewhat unusual for me, as I often enjoy exploring and developing a setting through its history, but in this case I wanted to let things cook a little longer before nailing down dates to things. (Part of this was also that I wasn’t entirely sure how I wanted to handle the calendar yet.) After about a dozen sessions of play, however, I ended up with some tangles of continuity — between character backgrounds, scenario setup, and player questions — that needed specificity to work out. My current timeline, therefore, looks like this:

• 90 Years Ago: The Long Road discovered between Verstern and Hajar.
• 80 Years Ago: KAS operations in the cluster abruptly come to an end.
• 25 Years Ago: Golyanova Bratva takes over Prospero’s Dream.
• 15 Years Ago: Ternary discovered.
• 10 Years Ago: Jadis discovered.
• 2 Years Ago: Cloudbank pulls out of the Tempest Cluster.

As you can see, this is still pretty barebones, but it’s enough to make sure that historical cause-and-effect stays consistent. (KAS can’t shut down before they discover the Long Road; the Bratva needs to take over Prospero’s Dream before the Ternary is discovered. And so forth.)

A key question for me in setting these dates was how long the “land rush” in the Ternary had been going on. I wanted it to be recent enough that I could justify having whole new worlds which had been barely been touched, but also long enough that if I had a “colonists have been here awhile and then things went to shit” scenario, then I could slot that in.

Note that leaving room for the adventures you haven’t dreamt of yet means (a) leaving undefined space, but also (b) making sure you’ve got the broad conceptual scaffolding. For example:

• An adventure set on an asteroid? I’ve given myself both the debris fields of Mrachni and the multiple asteroid belts of Hajar.
• Urban adventures? Katerineta is an older colony world with established cities, etc.
• Colony worlds? Gave myself a lot of conceptual space for this.

In many cases, I’ll try to give myself a couple different options. As continuity begins accumulating around one option, it may box other stuff out, so it’s nice to have a fallback.

My inclusion of dark systems also plays a part here: If I ever need more space… well, I guess one of those undefined dark systems actually has some interesting stuff in it!

Of course, not everything needs to be (or should be!) left a cipher. Where you need or want detail, don’t hesitate to lock it down. For example, I knew that I wanted the Ariel system, where the PCs’ homebase would be located, to be fairly barren (as a contrast to all the exciting places they’d travel to). So in this case I did describe and define all the extant planets in the system.

MEGACORPS

Having multiple megacorps in the cluster similarly gives me options: If a particular mission, project, colony, or facility doesn’t feel right for one megacorp, I can assign it to another. Plus, with multiple megacorps in play, I can have them in conflict with each other, and all kinds of adventure scenarios can spill out of that conflict.

From the beginning, I knew that I wanted two megacorps fighting over colonization and terraforming in the cluster. I’d created the name Salem-Watts when I wrote up the description of pseudomilk predators last year: They ran the Kikkomari V colony. I briefly played with the idea of including the Kikkomari system in the Tempest Cluster, but ultimately decided it would instead exist “offstage” in the Oberon Cluster.

Meanwhile, I’d used Behind the Name to generate some names, and that pushed me into Arabic influences for the Hajar and Jadis systems. (I can’t actually reproduce the steps that led me to Tasm and Jadis, but that’s the fun part of going down the research rabbit hole.) The Alshaahin megacorp, with its operations based out of Imliq Station (named after a king of Jadis), flowed pretty smoothly from this.

I added Namir-Radi as a sort of catch-all megacorp for any projects that didn’t fit Salem-Watts or Alshaahin. This has inadvertently, and largely through coincidence, caused it to become the most prominent megacorp in the campaign so far.

The last megacorp, Cloudback, is taken from the Gradient Descent adventure, which I’m planning to incorporate into the open table. My original plan had been to swap out the name “Cloudbank” for one of the other megacorps, but I wasn’t sure which one, so I decided to put a pin in it. Before I had a chance to circle back to that, however, one of my players rolled up an android character and, in exploring their background, I ended up invoking the name Cloudbank.

This turned out to be fortuitous, however, because it led me to develop the “Cloudbank mysteriously pulled out of the cluster two years ago” concept, which has created some low-level intrigue for the players who are paying attention and is also beginning to spin off a lot of ancillary developments that are really interesting, too. (For example, what happens when the megacorp who was providing hospital services to new colony worlds suddenly shuts down all the hospitals?)

CHARACTER BACKGROUNDS

This touches on something else I think about when developing a new setting: I want enough context that we can hang PC backstories off it. Furthermore, I’ve learned the power of being able to give players a couple different choices.

Player; I’m playing a Marine.

GM (Me): Okay, there are a couple military outfits in the cluster. First, there’s the Tempest Mercenary Company. There’s also the Novikov Naval Eskadre based  out of the Verstern system.

For an open table like this, what I’m usually doing is asking for an initial concept pitch (“Tell us about your character”) and then following up by either (a) taking a general idea (“I think I came here to do scientific research”) and making it specific (“you could’ve been working for Namir-Radi”) or (b) prompting them with a question (“if you came out here to research terraformed biomes, how did you end up bumming jobs on Prospero’s Dream?”).

Even with this limited background development, it’s remarkable how much it can end up driving the development of the settings (like the example of Cloudbank spinning off in a completely unexpected direction because someone happened to roll up an android).

PROSPERO’S DREAM

Using Prospero’s Dream as the home base for the open table was a gut instinct. Reading A Pound of Flesh, the supplement where the station was first detailed, I was really intrigued by the three phased fronts and how they’d been cleverly integrated throughout the book to create a palpable sense of passing time and escalating stakes. I saw the contours of how I could bootstrap that structure into an open table to potentially create something really cool, but I knew it would only work if the PCs were based out of the station.

We haven’t played enough to be sure how all that will turn out, but the initial results have been really promising.

On the other hand, having a space station with a population of 5 million as a home base for the campaign also forced me to confront a lot of issues with Mothership (like shore leave being classified by port type, which bizarrely means Prospero’s Dream has no dive joints) sooner rather than later.

The layout issues with A Pound of Flesh (pink text on a pink background?!) also make it incredibly unfriendly to use at the table. I have “completely reorganize all this information so that it’s not a headache to use it” on my To Do list.

TO DO

Speaking of my To Do list, this write-up of the Tempest Cluster is very much a beginning, not an end. My own version of the document has already expanded quite a bit as I add emergent details from character backgrounds (“kinfolk mines? interesting…”) and cross-reference scenario notes (Nirvana is one of the moons of Apsaras; Ypsilon-14 is located in the Hajar system; etc.).

But, as I talk about in So You Want to Be a Game Master, one of the great things about this initial setting write-up is that it also doubles — with little or no change — as a setting briefing for the players that I could post to our Discord.

(In practice, at an open table, many players nevertheless won’t have the opportunity/time to read it. So I have a five-minute spiel for new players sketching in the broad outlines of the cluster, which I can then flesh out with additional details as they roll up their characters.)

Some of the stuff on my lengthy To Do list dates back to when I originally wrote the setting up (stuff that I knew I would need to add at some point), while other needs and opportunities have been discovered through play. Examples of stuff I need/want include:

• A menu of shore leave options that the PCs can choose during downtime.
• Exotic shopping options, where the PCs can seek out non-standard equipment.
• Alphanumeric codes for the dark systems (KU-2B, KU-17, etc.) for easier referencing and keying.
• Name lists for the major cultural groups in the cluster.
• Name the spurs of Prospero’s Dream for easier referencing/keying. (Possibly add urbancrawl layers.)
• Figure out exactly how the NNE Volk 79 security patrols along the Long Road work.
• Where is the Stratemeyer Syndicate?

At the moment, pure worldbuilding stuff — no matter how interesting — is largely backlogged behind finetuning my open table procedures (downtime, life events, job board, journeys, etc.) and scenario prep. So my setting notes are largely only getting expanded as those needs dictate.

Honestly, this is how I do most of my worldbuilding. Every so often inspiration will strike and I’ll start exploring the setting out of pure curiosity, but for the most part I’m designing stuff for play and letting the setting slowly accrete over time.

Which also means that I have only the slightest inkling of what the Tempest Cluster will look like a year from now. Particularly since, if all goes well, the players will begin having larger and larger effects on the state of the world.

And given that this is Mothership, the whole place might have been eaten by an Elder God or invaded by time-traveling aliens unwittingly released by the PCs.

MORE MOTHERSHIP
Mothership Review: Adventure Sphere
Mothership Review: Trifold Adventures
Mothership: Thinking About Money
Mothership: Thinking About Combat
Untested Mothership: Astronavigation
Untested Mothership: Ablative AP
Mothership Monsters: Pseudomilk Parasites & Predators
Unboxing Mothership!

The Tempest Cluster was designed as the setting for my Mothership open table.

The Tempest Cluster is located in the Shakespeare Sector, its systems rimspin of the shattered, balkanized remnants of Terran Hegemony. It was previously two unconnected micro-clusters:

• Verstern, connected by a Jump-2 gate to the Oberson cluster
• Ariel and Hajar, connected by a Jump-4 gate to the Banquo cluster

Verstern lay on the edge of the Russo-Germanic Novikov Confederation. The isolated, low-value Ariel and Hajar systems were squabbled over by a variety of megacorp subsidiaries.

The lengthy Jump-1 route between Verstern and Hajar was accidentally discovered by a xenoarchaeology expedition, creating the unified Tempest Cluster. This created a minor trade route between the Oberon and Banquo clusters (albeit inhibited by the Jump-4 link to Banquo), but more importantly, the resources of the Hajar and Ariel systems were suddenly in demand on Katerineta, the old Verstern colony world.

The Tempest Cluster, however, remained an ill-visited backwater.

Everything changed, however, with the discovery of the Ternary – three systems directly linked via Jump-1 points, each with multiple worlds in the habitable zone. It was a massive colonization target – people and megacorp money began flowing into the cluster at an unprecedented rate.

This is the Tempest Cluster today: The fate of a dozen newborn worlds being written among the stars.

Because Jump-1 drives are more common, cheaper, and less prone to time dilation, space naturally becomes divided into clusters of systems connected by Jump-1 points. Galactic directions are divided into rim vs. core and trail vs. spin. Thus, “rimspin” is towards the edge of the galaxy and in the direction the galaxy is spinning.

TEMPEST STAR SYSTEMS

The Tempest Cluster has eight major systems and, of course, numerous dark systems, six of which lie along major trade routes.

Dark Systems: These intermediary systems along the cluster’s jump routes contain little of interest (or, at least, little that has yet been discovered). Ships mostly just pass through these systems on their way from one jump point to another, although there is a risk of pirates and the other horrors of rimspace.

ARIEL

The gateway to the Banquo Cluster, the Ariel system is a barren system. It has numerous dwarf planets in the outer system, but only two planets of note:

• Ariel I is a hot Jupiter which has gotten too close to the star. Its atmosphere is currently being ripped away. A deuterium plasma mining station operated by Salem-Watts called Hephaestus can be found within the “Roche river.”
• Ariel II is a gas dwarf. It also orbits relatively close to the star and has been stripped of its moons.

Prospero’s Dream, a station whose population has recently swelled to 5 million sophonts, orbits Ariel II. The station was dying before the Ternary was discovered, allowing the station to be taken over by the Golyanovo Bratva, a mafia with origins in the Oberon Cluster. The bratva has held onto control with the muscle of the Tempest Mercenary Company, although the rapid expansion of Prospero’s Dream is now taxing the station’s existence in different ways.

THE TERNARY

These recently discovered star systems are filled with colonization and terraforming targets. The resulting colony rush has only been accelerating as more terraforming projects come online.

• Parvati
• Vani
• Laxmi

The systems are named for the Tridevi — the three principle Hindi goddesses.

Pandora Station: Located in the Parvati system, Pandora Station is an infamous X-class “pleasure city” – the perfect place for the best shore leave of your life… if you can afford it.

Moons of Apsaras: Apsaras, a dark gas giant in the Vani system with an abnormally small magnetosphere, is orbited by multiple planet-class moons with terraforming potential.

HAJAR

The Hajar system recently (in astronomical time scales) had multiple terrestrial planets destroyed. (The current theory is that a rogue planet passed through the system. In addition to destabilizing some planets in its own right, it also caused the orbit of Hajar II, a super-Jupiter, to move inward, wreaking havoc in its wake.) This has resulted in multiple asteroid belts, several of which currently orbit at strange inclines to the planetary disk.

Generations of asteroid miners who have made Hajar their home are now having to contend with increased hypercorp interest in the system’s riches. There’s also been a significant uptick in piracy.

JADIS

On the far side of the Hajar system, Jadis is another newly discovered system with multiple terraforming targets. With so much focus already placed on the Ternary, development has been slow here.

The system is currently governed by the Jadis Terraforming Conglomerate (JTC), which is effectively controlled by the Alshaahin megacorp.

Imliq Station: The JTC is based out of Imliq Station, in orbit around a planet named Tasm.

Jadis is named after the “lost” Arabian tribes of Tasm and Jadis.

MRACHNI

Mrachni is a black hole orbited by numerous dead worlds, many of them ten or twenty times the size of Earth. A scattering of isolationists and scientific stations can be found throughout the system.

There are spacer tales of a fabled Eden — a habitable, Earth-like world hidden somewhere within the glare of Mrachni’s accretion disc, warmed by the blueshifted light of the cosmic background radiation. But no reliable evidence of such a place has ever been found.

VERSTERN

Verstern is a border system of the Novikov Confederation.

Katerineta is an old colony world with a population just over 1 billion. The equatorial region is too hot for human habitation, but both polar regions have been settled. The south lacks a continental mass and is referred to as the Archipelago.

LX-510 is a Class-B military port that has expanded to also support trade between the Oberon and Tempest clusters. It’s home to the Novikov Naval Eskadre (NNE) Volk 79, which is spread thin ostensibly providing anti-piracy patrols in the Verstern-Hajar corridor.

MEGACORPS

ALSHAASHIN (Royal Falcon): A terraforming megacorp based out of the Banquo Cluster. The clan-guilds of Alshaahin control the Jadis Terraforming Conglomerate (JTC), which governs the Jadis system.

SALEM-WATTS: A competing terraforming megacorp based out of the Oberon Cluster.

NAMIR-RADI: A hydra-headed hypercorp with multiple subsidiaries active in the Tempest Cluster.

CLOUDBANK: Cloudbank specializes in medicine, biotech, cyberware, androids, and artificial intelligence. They had a growing presence in the Tempest Cluster, but recently pulled out of the cluster entirely for reasons which remain largely unexplained.

Next: Designing the Tempest Cluster

There’s been Discourse™ of late about the use of GenAI/LLMs in creating RPGs. Not the artwork in an RPG book (that’s a whole ‘nother kettle of fish), but the actual design and development of the game itself: Feeding game text into ChatGPT, Claude, or similar chatbots and asking it to critique, analyze, revise, or otherwise provide feedback.

If you know anything about how LLMs work, it will likely be immediately obvious why this is a terrible idea. But the truth is that a lot of people DON’T know how LLMs work, and that’s increasingly dangerous in a world where we’re drowning in their output.

Michael Crichton described the Gell-Mann amnesia effect: “You open the newspaper to an article on some subject you know well. In Murray’s case, physics. In mind, show business. You read an article and see the journalist has absolutely no understanding of either the facts or the issues. Often the article is so wrong it actually presents the story backwards—reversing cause and effect. (…) In any case, you read with exasperation or amusement the multiple errors in a story—and then turn the page to national or international affairs, and read with renewed interest as if the rest of the newspaper was somehow more accurate about far-off Palestine than it was about the story you just read. You turn the page… and forget what you know.”

Flipping that around, I think analyzing stuff like LLMs in arenas we’re familiar with is valuable because we can more easily see the failures and absurdities. My particular arena of expertise and familiarity — and one I think is likely shared by most of you reading this — is RPGs. So let’s use that familiarity as a lens for looking at LLMs.

Before we start, let’s set a couple baselines.

First, I don’t think AI is completely worthless. I also don’t think it’s the devil. Whether we’re talking about LLMs or some of the other recent technology that’s all getting lumped together as “AI” or “GenAI,” there’s clearly specific ways of using those tools (and also building those tools) which can be ethical and valuable. I don’t think pretending otherwise is particularly useful in trying to prevent the abuse, theft, propaganda, systemic incompetence, and other misuse that’s currently happening.

Second, I am not an expert in LLMs. If you want a truly deep dive into how they work, check out the videos from Welch Labs. (For example, The Moment We Stopped Understanding AI.)

I think the key thing to understand about LLMs, however, is that they are, at their core, word-guessers: They are trained on massive amounts of data to learn, based on a particular pattern of words, what the next most likely word would be. When presented with new input, they can then use the patterns they’ve “learned” to “guess” what the next word or set of words will be.

This is why, for example, LLMs were quite bad at solving math problems: Unless they’d “seen” a specific equation many times in their training data (2 + 2 = 4), the only pattern they could really pick out was X + Y = [some random number].

LLMs are actually still incredibly bad at math, but the “models” we interact with have been tuned to detect when a math problem is being asked (directly or indirectly) and use a separate calculator program to provide the answer. So they look significantly more competent than they used to.

DESIGNING WITH CHATGPT

It’s truly remarkable how far what are fundamentally babble generators can take us. With nothing more than word-guessing, LLMs can create incredible simulacrums of thought. Every generation interprets human intelligence through the lens of modern technology — our brains were full of gears and then they were (steam) engines of thought before becoming computers — but it’s hard not to stare into the abyss of the LLM and wonder how much of our own daily discourse (and even our internal monologue?) is driven by nothing more than pattern-guessing and autonomic response. We see it in the simple stuff:

Ticket Taker: Enjoy the show!

Bob: Thanks! You, too!

But does that sort of thing go deeper than we’ve suspected?

Regardless, there’s one thing missing from LLMs: The ability to form mental models. They can’t read a text, form a mental model of what that text means, and then use that mental model. They can’t observe the world, think about it abstractly, and then describe their conclusions. All they can do is produce a stream of babbled text.

This is why the term “hallucinate” is deceptive when used to describe LLMs’ propensity for spreading misinformation. A “hallucination” would imply that the LLM has formed a false mental model of the world and is now describing that false understanding. But this is not, in fact, what’s happening. What’s happened is that it guessed a word and that word, while matching the patterns found in the model’s training data, did not conform to reality. It’s just words. There is no underlying mental model behind them.

It’s also why asking LLMs to critique anything more complex than the grammar of individual sentences is a waste of time. In order to meaningfully critique something, you have to be able to form a mental model of that thing, have deep and original thoughts about it, and then figure out how to express the conclusions you’ve drawn. An LLM can’t do any of that. At best, it can produce a simulacrum of criticism — a babble that you could perhaps use like a Rorschach blot to free associate your way to a useful insight.

Which brings us to the current movement convinced that they can meaningfully “collaborate” with ChatGPT when designing RPGs. They want to feed in the text of a rulebook and ask the LLM things like:

• Can you build an encounter for 7^th level characters?
• Is this new class I’ve designed balanced with existing classes?
• Are there any problems with the probability of my core mechanic?
• What attribute scores and skills should I have?

In light of what we know, of course, we can immediately see the problem: ChatGPT can’t reliably do basic arithmetic, but it can definitely do complex analysis of mathematical game balance in a system of rules it fundamentally can’t understand?

Someone recently challenged me to “see for myself”: Take an existing RPG, feed the rules into ChatGPT, and ask it to playtest the rules.

This was an interesting challenge, and I’m certainly always open to being proven wrong.

Let’s take a peek at what happened.

CHATGPT DOES A PLAYTEST

I pointed ChatGPT to a copy of Lasers & Feelings, a one-page space opera RPG by John Harper. Then I asked it to generate a playtest transcript of the game.

The choice of game was deliberate: Using a one-page RPG should make it much easier for ChatGPT to analyze and use the entire text. The game is also limited to two mechanics: The action roll and Helping (which adds +1d6 to another character’s action roll). This was the challenge in its easiest form possible.

ChatGPT begins by saying this:

Lasers & Feelings — Playtest Session Transcript

GM: You’re the crew of the interstellar scout ship Raptor, alone while Captain Darcy recovers in a medical pod.

ChatGPT is combining two different sentences from the PDF. Then:

Your mission: find the source of a distress signal coming from an uncharted system.

This is incorrect. In Lasers & Feelings, the GM is supposed to roll on a mission generator. ChatGPT’s mission cannot be generated by the game. But let’s give it a pass here, because there are plenty of human GMs who would also ignore the mission generator and just riff.

ChatGPT now moves on to character creation. It does a decent job of identifying the four steps of character creation, but almost immediately screws up basic math:

Number: 3 (balanced but slightly leaning Lasers)

In Lasers & Feelings, the player picks a number for their character from 2-5. “A high number means you’re better at LASERS (technology; science; cold rationality; calm, precise action). A low number means you’re better at FEELINGS (intuition; diplomacy; seduction; wild, passionate action).”

The number 3, you’ll note, does not “lean Lasers.” It does the exact opposite.

Furthermore, there’s no such thing as “balanced” in Lasers & Feelings. The game specifically doesn’t give you the choice of a midpoint. The whole point is that there’s a tradeoff between Lasers & Feelings. ChatGPT has fundamentally misunderstood the core design principles and theme of the game.

After character creation, ChatGPT proceeds with a transcript of play, and it almost immediately makes a skill check:

Lee: I want to pilot us carefully in. That’s Lasers because it’s technical precision.

Lee rolls 1d6 + 1d6 (prepared) since they have experience navigating rock fields.

This is incorrect. Lee is a Pilot, which means he’s an expert in piloting. If he’s also prepared (as ChatGPT asserts), he should be rolling 1d6 + 1d6 (expert) + 1d6 (prepared).

GM: Your target number is 3 (your stat). Under 3 counts as a success for Lasers.

This is incorrect. A 3 would also count as a success (in addition to generating a Laser Feeling).

ChatGPT cites this blog post as a source for this, but the blog post summarizes the mechanic correctly. ChatGPT just screwed up.

Lee rolls: 2, 5. ✔️ One die under 3 — Success!

GM: You thread us through the asteroids. The distress beacon pings again. What do you do?

According to the rulebooks: “If one die succeeds, you barely manage it. The GM inflicts a complication, harm or cost.”

The GM did not inflict a complication, harm, or cost. ChatGPT has screwed up again.

The “transcript” continues in this vein. Sometimes ChatGPT gets the rules right. It frequently doesn’t, in a wide variety of ways.

You can see the pattern and understand the root cause: ChatGPT can’t actually understand the rules of Lasers & Feelings (in the sense of having the words of the rulebook create a mental model that it can then use independent of the words) and, therefore, cannot truly use them. It can only generate a sophisticated pattern of babble, guessing what the next word of a transcript of Lasers & Feeling game session would look like based on the predictive patterns generated from its training data.

And if it can’t understand the rules well enough to accurately call for a simple action roll, what possible insight could it have into the actual design of the game?

None, of course. Which is why, when I asked it what changes it would make to the game to reinforce the themes, it replied with stuff like:

• The GM should only be allowed to inflict consequences that affect relationships. (Making the game functionally unplayable.)
• Encourage players to switch modes between Feelings and Lasers by inflicting a -1d penalty to the next Feelings roll each time a characters uses Lasers. (This rule would obviously have the exact opposite Plus, it doesn’t recognize that many rolls only use 1d, so how would this rule even work?)

Maybe one of these nonsense ideas it generated will spark an idea for you, but it’s inspiration from babble. Mistaking it for actual critical insight would be a disastrous mistake.

AI GAME MASTERS

Reading ChatGPT’s “transcript” of play, however, it’s nevertheless impressive that it can produce these elements and distinct moments: The distress call isn’t from the rulebook. It’s plucked that out of the ether of its training data. When I mentioned earlier that it’s remarkable how much can be achieved with an ultra-sophisticated babble engine, this is the type of thing I was talking about.

Examples like this have led many to speculate that in the not-too-distant future we’ll see AI game masters redefine what it means to play an RPG. It’s easy to understand the allure: When you want to play your favorite game, you wouldn’t have to find a group or try to get everyone’s schedules to line up. You’d just boot up your virtual GM and start playing instantly. It’s the same appeal that playing a board game solo has.

Plus, most publishers know that the biggest hurdle for a new RPG is that, before anyone can play it, you first have to convince someone to GM it — a role which almost invariably requires greater investment of time, effort, and expertise. If there was a virtual alternative, then more people would be able to start playing. (And that might even end up creating more human GMs for your game.)

There will almost certainly come a day when this dream becomes a reality.

But it’s not likely going to come from simply improving LLM models.

This Lasers & Feelings “transcript” is a good example of why:

• The PCs are following a distress signal.
• It turns out that the distress signal is actually a trap set by bloodythirsty pirates. Two ships attack!
• ChatGPT momentarily forgets that everyone is onboard ships.
• We’re back in ships, but now there’s only one pirate ship.
• And now they’re no longer pirates. They’re lost travelers who are hoping the PCs can help them chart a course home.

It turns out that the GM’s primary responsibility is to create and hold a mental model of the game world in their mind’s eye, which they then describe to the players. This mental model is the canonical reality of the game, and it’s continuously updated — and redescribed by the GM — as a result of the players’ actions.

And what is ChatGPT incapable of doing?

Creating/updating a mental model and using language to describe it.

LLMs can’t handle the fictional continuity of an RPG adventure for the same reason they “hallucinate.” They are not describing their perception of reality. They are guessing words.

The individual moments — maneuvering through an asteroid belt to find the distress signal; performing evasive maneuvers to buy time for negotiations; helping lost travelers find their way home — are all pretty good simulacra. But they are, in fact, an illusion, and the totality of the experience is nothing more than random babble.

And this is fundamental to LLMs as a technology.

Some day this problem will be solved. There are a lot of reasons to believe it will likely happen within our lifetimes. It may even incorporate LLMs as part of a large AI meta-model. But it won’t be the result of throwing ever greater amounts of computer at LLM models. It will require a fundamentally different — and, as yet, unknown — approach to AI.

I’m a huge fan of Blades in the Dark, and I was very excited for Deep Cuts, a major expansion designed by the game’s original creator, John Harper, with additional design and writing by Sean Nittner, Pam Punzalan, and James Mendez Hodes. I backed the crowdfunding campaign for the book and, once I held it in my hands, I was immediately sucked in and read the whole thing cover to cover.

Then I settled in and started doing some deep thinking about Deep Cuts.

The book can be roughly divided into two parts: The first adds a bunch of new material to the post-apocalyptic steampunk setting of Doskvol. The second radically revises the rules of the game, effectively offering a sort of do-it-yourself kit for assembling an ersatz second edition of the game.

SETTING

The setting material is a goulash of goodness. Blades in the Dark is, of course, designed to be a modular, sandbox game, and Deep Cuts is basically just expanding the toybox full of goodies that you can drop into your game.

Innovations provide a selection of new equipment, including rare items like a Bolton Autocycle or a Sparkrunner Rig that could easily become the target of a score or an Acquire action. I really like the presentation of most of these items as new technology in the setting itself: Part of the steampunk aesthetic is that technology is marching forward. You can take this stuff and feed it into your campaign one or two items at a time, blaring it as headline news, lightly seasoning it as background detail, or framing it up as a surprising new threat in the middle of a score.

Squeezed into the middle of the Innovations section is a whole new calendar that renames the six months of the year. This really reflects how Deep Cuts is just a semi-random collection of stuff John Harper created for his home games. There’s not a single, unifying theme or thesis here. It’s really just a guy saying, “Hey! Look at this cool thing I made!”

Next up are twenty-five new Factions. If you’ve run multiple campaigns of Blades in the Dark over the past decade (as I have), then these are a welcome breath of fresh air. You and your players may have become intimately familiar with the Crows and Red Sashes, but now you can frame up whole new starting situations with factions like the Ink Rakes, Ironworks Labor, and Rowan House.

My only reservation with these new factions is that so many of them are “infrastructure.” The firefighters, cabbies, and railworkers are an important part of the city, but a multitude of faction goals like “modernize the guild” and “formalize their union” really underline that they’re more bureaucratic functionaries than sources of adventure. Nonetheless, there’s a lot of good meat in here, too, and perhaps inspiration will strike and turn stuff like the fire brigade’s PR campaign into your campaign’s Chinatown.

Also tucked away in this section is a 30-year plan to expand Doskvol, pushing out into the ghostlands and adding four new districts. This is an incredibly cool concept, immediately driving all kinds of internecine strife while also providing a long-term scaffolding for evolving the setting in a game that encourages time-skips and decompressed downtimes.

Heritages and Backgrounds then provide a wealth of new information about the wider world beyond Doskvol, while also giving players new structures for adding depth and color to their characters. Did you come from a family of Blood Cullers or Lockport Scummers? Before you joined the crew, were you a roof runner or a farmhand? Being able to flip through these sections and have concepts like “refugee occultist” or “electrochemist from a family of lightning couriers” leap off the page is delightful. (In fact, the only thing I might do differently is put them on random tables so that players looking for inspiration can just roll-and-combine.)

THE CAMPAIGN CATALYST

SPOILERS FOR THE DEEP CUTS CAMPAIGN FRAME

Wrapping up and also tying together the new setting material is a campaign frame: Fractures in reality have connected Doskvol to another reality. Strangers from this other world are slipping through into Doskvol, bringing with them strange technology and magic. Others are vanishing from this world and passing into the other. Factions are slowly beginning to learn the truth about what’s happening and are secretly moving to take advantage of it.

This is a fun concept with (if you’ll pardon the pun) a lot of dimensions to be explored. Harper weaves it into faction clocks and NPCs, then fleshes it out with mission profiles for every crew type. Combined with the structures for emergent play which are the backbone of Blades in the Dark, this all primes the pump and gives you almost everything you need to introduce the Others and hit the ground running. It is, very literally, a catalyst — something you can inject into your campaign and initiate a run-away chain reaction.

The only problem here is that Harper wants to leave the precise nature of the other world up to the individual GM. He offers several possibilities:

• Victorian Earth
• Mirror Universe
• Demonic World
• Time Travel
• Lady Blackbird

Or any number of other possibilities you might imagine.

On the one hand, it’s great to offer this kind of flexibility. On the other hand, the practical effect is that nothing attached to the catalyst can have any kind of specificity. It, by necessity, ends up as just vague handwaving and a few encouraging, “You’ve got this!” cheers shouted to the GM.

For example, the members of Rowan House are secretly descended from the Others. Does that mean they’re demon-spawn or that their grandmother was from 1810 London? It’s a radical difference. And because the book can’t give you an answer, they end up being… nothing at all. Deep Cuts has to avoid making strong, interesting choices over and over again.

So the catalyst is a really cool idea and it demonstrates the broad outlines of how you could implement big, widespread events like this into your Blades in the Dark campaign. But it also leaves you with all the hard work of making it playable.

SYSTEMS

At this point we transition to the back half of the book, which is loaded up with dozens of changes to the core mechanics of Blades in the Dark. To be clear: Very little of this is new mechanical systems. It’s almost entirely changes to the existing mechanical systems.

Let’s deep cut to the chase here: I did not like this section of the book.

There are a few interesting nuggets buried in here, but the vast majority of the material makes the game significantly worse. If a second edition of Blades in the Dark were published with these changes implemented, I would not play it or run it.

Let’s break down a few examples.

EFFECTS & CONSEQUENCE

Let’s start with something I really liked: On page 101, Deep Cuts includes a unified Effects & Consequences table providing escalating adjectives and mechanical effects for Limited, Standard, Great, and Extreme effects. (For example, a standard effect of Intimidated might be downstepped to a limited effect of Hesitant or upstepped to a great effect of Afraid or an extreme effect of Dominated.)

This table is incredibly useful, particularly for a new GM trying to wrap their noggin around the concepts of position, effect, and consequences.

LOAD

I also liked the expanded system for Load, which not only finesses the handling of exceptionally heavy objects in the system, but also uses Load as a way of managing encumbrance gains and complications during a job.

HARM

John Harper saw that some players in actual play videos would forget to apply the penalties from Harm to their die rolls, so he produced a revised system in which Harm has no inherent mechanical effect. Instead, it becomes a tag system that the GM can choose to invoke to create complications, reduce position/effect, or… apply a penalty to the die roll. When the GM chooses to invoke the condition, the player gets to mark XP.

Ugh.

First, I’m not generally a fan of dissociated tag systems where the reality of the game world only meaningfully exists when someone arbitrarily chooses to “invoke” it.

Second, the response to players being unable to keep track of a dice modifier for their characters should NOT be, “Well, I guess there’s just no choice but to require the GM to keep track of every individual wound suffered by every single PC and also be responsible for invoking them in way that’s fair and effective in some nebulously undefined way.”

This is part of a long trend in RPGs of making the GM responsible for everything that happens at the game table. I strongly feel that the entire hobby needs to break away from laying this immense burden on the GM’s shoulders, not needlessly multiply it.

TRAUMA

This module lets the players mark XP if they use their Trauma in a way that creates a problem or complication for them during play, which is a great way of reinforcing it as a roleplaying prompt.

Trauma is also modified so that it’s no longer permanent, with options given for recovery. This, obviously, blunts the core game’s relentless march to character retirement, which makes it a nice option for groups that don’t want as much character turnover. Another module gives options other than prison time for removing the crew’s Wanted Levels, to similar effect.

Personally, I think there’s a significant drawback here: The original game created a low level of character churn, which systemically encouraged players to create multiple characters. This, in turn, put the narrative focus of the game on the crew, rather than the individual characters, and diversified the player’s experience and interactions with the game world. Moving away from this also means moving away from one of the things that made Blades in the Dark unique and distinguished it from other RPGs.

ACTION

Which brings us to the Action module, in which Deep Cuts overhauls the entire core mechanic of the game.

And, frankly, Harper makes a dog’s dinner of it.

Take Devil’s Bargains, for example. In the original game, the GM could offer a devil’s bargain: The player could gain extra dice on an action roll by accepting an additional cost — collateral damage, Harm, Coin, sacrificing an item, starting a clock, betraying a friend, adding Heat, etc. In Deep Cuts, on the other hand, the term “devil’s bargain” now refers to any time the GM requires the players to make a roll. But also the extra dice thing. And also you can offer a devil’s bargain to let them skip the dice roll entirely.

A useful term of art is turned into a mess, while at the same time trying to position it as a central axis around which the game turns.

This all ties into the new concept of Threat Rolls, which essentially replace action rolls. When using this new core mechanic, instead of setting position and effect, the GM simply sets the negative consequence that will happen on a failed roll and then tells you what action rating to roll: On a 6+ you avoid the consequence. On a 4/5 you suffer a reduced consequence. On a 1-3 you suffer the full consequence.

This whole mechanic essentially waves a white flag at those who want to run Blades in the Dark exactly like D&D (with the DM telling the players what to roll and what the result will be). Good for them, I guess. But, once again… Ugh.

I’m just not interested in Blades in the Dark being turned into generic mush.

DOWNTIME

Deep Cuts also heavily revises the Downtime system. This is more of a mixed bag.

On the one hand, there’s a cool new Debt mechanic and new rules for using Banks. The new guidelines for Heat — which take a base rate based on Tier and then modify it by Target, Chaos, Death, and Evidence — are also great, along with the supporting guidelines for modeling actions to reduce or mitigate the Heat from a job.

There’s a similarly revised set of guidelines for Payoffs. These are also useful, although it’s worth noting that they significantly increase the amount of Coin the PCs will earn per job.

On the other hand, there’s stuff like the new Entanglements system, which is designed to remove random dice rolls. The overall effect, though, seems to greatly flatten the experience: Outcomes become predictable and routine. There are no surprises and no variations from one Downtime to the next. It also removes a lot of the procedural content generation aspects that make Blades in the Dark such a fun sandbox engine.

CONCLUSION

I don’t like Deep Cuts.

I loathe the mechanical changes. They’re disastrous and, honestly, pretty depressing. It’s always sad to see a game heading in the wrong direction, and that’s even more true when the game was something truly special like Blades in the Dark. If the book was limited to its mechanical content, I’d probably give it an F-.

However, almost half of the book is the new setting material, and while that has a few duds and some missteps, there’s a ton of useful stuff in there that will definitely enhance and expand your Blades campaigns. Therefore, with the book only costing $25, I think dedicated fans can just barely justify picking up a copy for the setting material alone.

GRADE: D

Designer: John Harper
Additional Design & Development: Sean Nittner, Pam Punzalan, James Mendez Hodes, Allison Arth, Sharang Biswas, Sidney Icarus

Publisher: Evil Hat
Cost: $25
Page Count: 128

G. asks:

I’ve run a couple Powered by the Apocalypse games and I don’t get the hype. What are the strengths of PBTA games supposed to be compared to games like Call of Cthulhu, Vampire the Masquerade, or Savage Worlds?

Let’s start with a quick orientation for people not familiar with Powered by the Apocalypse:

• Vincent and Meguey Baker released Apocalypse World in 2010.
• The Bakers put the game under a free license.
• As a result, the novel system has been adapted and used in hundreds of RPGs. These games are referred to as Powered by the Apocalypse (PBTA).
• The system also notably influenced John Harper’s Blades in the Dark, which has also inspired countless RPGs published as Forged in the Dark.

In actual practice, Powered by the Apocalypse is barely a dice mechanic loosely paired to the concept of Moves. And on of Baker’s own PBTA games doesn’t include the dice or the moves. So there’s a huge range in what “PBTA games” do and how they do it.

But let’s break down each feature that seems to be closely identified with Powered by the Apocalypse.

CORE MECHANIC

PBTA uses a generally three-tier outcome for all action resolution: Success, Partial Success/Success with Consequences, Failure.

It’s pretty typical for RPG mechanics, particularly pre-PBTA, to default to binary outcomes. Even conventional RPGs featuring something like margins of success or critical hits will still usually define success as “the PC achieves exactly what they want” and then maybe they get something extra if they roll or a critical or their margin of success is high enough. PBTA, on the other hand, tends to define typical success as “the PC gets SOME of what they or they pay a price for it” with “achieve exactly what you want without cost” being treated as the exceptional result.

Similarly, conventional RPGs tend to have the Failure state default to “you didn’t do it.” PBTA instead defaults to having consequences for your failure: You didn’t just “miss” the ogre; you rushed the ogre and the ogre punched you in the face. This approach also means that PBTA games tend to embrace failing forward, while leaning towards not just player-facing mechanics, but a specific flavor of player-facing mechanics featuring fortune-in-the-middle decisions.

PC MOVES

Beyond three-tier resolution, PBTA almost always package their resolution mechanics into Moves. The distinction here can get pretty fuzzy (due to the breadth of both conventional RPG and PBTA design), but the result in what I consider better PBTA games is a throwback mechanic that evokes true old school design: In the ‘80s, RPG design shifted almost entirely to the “generic universal mechanic” as their core design. (Usually, but not necessarily, some variation of ability score + skill + dice vs. difficulty.)

The distinction here can be confusing to some, because quite a few OSR retro-clones have retrofitted the games they’re emulating to be built around a generic universal mechanic. But the older of old school games were built around, “You want to do something? Let’s build a custom mechanic for it!”

PBTA isn’t strictly old school, though, because it’s less, “Here’s a collection of stuff we made ad hoc at the table to address situations that came up” and instead “here’s a carefully curated selection of tools which will deliberately shape the focus and direction of play.”

Let’s think of this as neo-old school design.

(And, again, this applies to the, in my opinion, better PBTA games. There are quite a few PBTA games that turn their Moves into generic, unfocused mush because their designers are defaulted back to “generic universal mechanic” as their design model.)

PLAYBOOKS

A character creation and advancement system featuring distinct Playbooks for different types of characters is also a common feature of PBTA games.

In practice, though, this is just class-based or archetype-based character creation, which is quite common in conventional RPGs. (I’ve seen any number of efforts to explain how, “No, no! It’s totally different!” But it really isn’t, although the class abilities can have a unique feel to them because of how they tie into the Moves methodology.)

GM MOVES

This is another neo-old school design element.

To explain what I mean by that, consider the original 1974 edition of D&D: It included a hyper-specific procedure for running a dungeon. If you strictly follow that procedure, you get a very specific style of play and type of adventure.

GM Moves in Apocalypse World are designed to do that the same thing, providing a very specific structure of prep coupled to a very specific procedure of play that creates a very specific outcome at the table.

This is, it should be noted, another place where a lot of PBTA games turn into generic mush by designing their GM Moves as “the generic stuff that GMs do.” (Often accompanied by weakening or removing the provision that GM Moves are the ONLY thing a GM is allowed to do.) In some cases, these chapters degrade entirely into generic GM advice.

An interesting lens that can help understand this distinction is Blades in the Dark, which, as noted above, is heavily influenced by PBTA, but distinct from it. John Harper, the designer, notably replaced GM Moves with a chapter called GM Actions, which is the “generic GM advice” approach to GM Moves. But, notably, this is because Harper has moved all the hyper-specific procedure stuff into The Score and Downtime chapters. (And, in fact, made it even more hyper-specific, in a style very similar to the 1974 D&D dungeon procedures.)

FRONTS/THREATS

The last thing that I, personally, consider a core identity for Apocalypse World was the concept of Fronts: A collection of threats and agendas, motivated by a Fundamental Scarcity, defined with specialty Moves, and tracked with Countdown Clocks. Fronts were a specific structure for prepping situations, and the GM was instructed to create their campaign by simply setting up Fronts and then playing to find out  what happened as those situations evolved dynamically in concert with the PCs’ actions and agendas.

But then Apocalypse World 2^nd Edition eliminated the entire concept of Fronts and replaced it with a heavily revised system for managing Threats. I haven’t run a game using the revised system, so I can’t comment too much on the details, but although the methodology was significantly altered, the core intention and approach remained the same: Stock the world with dynamic situations. Use them to pressure the PCs and, when the PCs respond, play to find out by following your procedures (Moves, Clocks, etc.).

The basic concepts of Fronts and Threats have been adapted in myriad ways by other PBTA games.

WHITHER THE APOCALYPSE?

Circling back to the original question, if PBTA games really are “special” or different from other RPGs, why might someone playing them not understand what the big deal is?

Well, depending on what non-PBTA games you’ve been playing and also how you’ve been playing them, PBTA games may not, in fact, be a radically different experience for you. Situation- and procedure-based play have, as I noted above, go all the way back to Arneson and Gygax. It’s really, fundamentally, what the RPG medium was designed to do. I’ve personally been preaching about how you can do situation-based play in any RPG for a couple of decades now, and a wide variety of old-school and OSR games are designed around these principles, too. (And even more now than when Apocalypse World first came out.)

On the other hand, as I mentioned, there are a number of PBTA games that turn these core features of Apocalypse World into mush: Moves just become generic “you do stuff.” Fronts become disconnected from procedure. Sometimes whole chapters of How to Prep Plots are attached. Even the core mechanic often gets defaulted back towards something much closer to traditional binary outcomes. So it’s quite possible to play a PBTA game that pretty thoroughly disguises or eliminates the most distinctive features of PBTA games.

Similarly, there are a lot of GMs running PBTA games — including Apocalypse World — that aren’t actually running those games. This is actually a surprisingly frequent phenomenon with RPGs: No matter what the rulebooks actually say, for these GMs every game just defaults to a core resolution mechanic that they arbitrarily invoke. (In many cases, you’ll see this degrade even further, with resolution mechanics that amount to little more than “high roll on the dice = good, low roll = bad” regardless of skill modifiers, difficulty classes, or anything else.)

Some GMs have also been so thoroughly conditioned in prepping and running adventures in one specific way (often, but not always, a linear railroad), that they similarly default to habit no matter what structures a game may be designed for. Sadly, some of these GMs, when running Apocalypse World, will even go so far as to prep a bunch of Fronts and/or Threats as the rulebook instructs… only to do nothing with them. At best, the Fronts serve as a kind of idea board for them. Often they’re just laid aside entirely, and the GM will be left scratching their head and wondering why they wasted all that “pointless” effort.

Combine both of these — railroad adventure prep and “all systems are a die roll and a vibe” — in a single GM (which is far from uncommon), and you pretty much eradicate everything that makes Apocalypse World and PBTA games special and unique.

With all that being said, even if you’ve been running or playing the full-fledged PBTA experience, it’s quite possible that it’s just not your jam. Not every game is made for every person.

But if you’ve played a PBTA game and it didn’t feel different from, say, a D&D dungeoncrawler… well, that probably means something got mushy somewhere. Might’ve been the game design. Might’ve been the GM. Either way, it’s probably worth giving PBTA another chance and finding out what happens when you really embrace the structure of the game.

Go to Ask the Alexandrian #1