Sunday, 18 April 2021

Scorched Earth NPC Relationship Table

Integrating player-written character backgrounds into your campaign is real work. Sometimes they just write so much, who can keep all those names straight?

Why struggle? Let the dice be your friend! Every time they bring someone up from that dog-eared sheaf they keep bringing to game night, just roll d6:

d6Whatever happened to so-and-so?
Moments after you last saw them, they joined up with the enemy. Life as an entry level hobgoblin is hard, but it also has a kind of simplicity. They don't really think about the past. Cudgel, rusty helmet, raawr, you know?
They died, pretty much right away. Their surviving relatives and friends remember the PC as a bit of a dick to them in their final years.
One day, their home just fell into the earth, and they haven't been seen since. Coincidentally, the hole is the entrance to that dungeon you were planning on using next!
It was never reciprocal, and the NPC is honestly having trouble remembering the PC's name. I mean, it's great you thought we we were close or whatever, but I think you read too much into it. Please don't make it awkward.
Nobody can remember the NPC. Who? Are you sure? There's no magic at work, just an overactive imagination.
After a brief, high fever, they erupted into a brightly colored fungal mat. Nobody's been able to clean it up, the stench is unbelievable.

Friday, 9 April 2021

Some Mosaic Strict Sets Appear!

Forty eight days ago, I wrote up Mosaic Strict, a somewhat light-hearted set of compliance rules for making modular parts of RPGs. It asked, what if there were no mechanical interconnections between various parts of a game at all?

Since then, several folks have actually made Mosaic Strict rule sets!

EDIT: Since this post was published, more have appeared. I'll track them in a spreadsheet.

A Lovemaking

The first was Paul Czege, who wrote A Lovemaking, rules for how characters negotiate whether to get it on. Released March 1, 2021.

I'm Thinking of Ending Things in the Dungeon

Next was Alfred Valley, with what must be the shortest Mosaic rule set so far, a two-sided business card on how to decide whether to hang up the spurs and leave the dungeon. Released March 4, 2021.

Turn Up Your Nose

Third was Dan Maruschak with a one-pager on how to empathize with the snobbish upper classes when they're forced to deal with people 'beneath them'. I imagine this could be played inverted to good effect. Released March 9, 2021.

Matthias of Liche's Libram wrote this rule set about how to interpret the significance of characters' dreams. Released March 14, 2021.

Cook Off

Matt Bohnhoff's first rule set is about cooking under pressure! Whose dish will win the day? Released March 15, 2021.

The Magic of Names

I missed this one on Twitter when it was first announced, but there's now an Ursula K LeGuin-inspired freform magic system! By Thomas Manuel, March 30, 2021.

Quick Combat

Last on my list is this quick-play combat system, also by Matthias of Liche's Libram. Lose an eye in no time! Released March 31, 2011.

Fishing Trip

Matt Bohnhoff then made this chill mini-game about stepping aside from the adventure to catch some fish. Released April 1, 2021.

I'm aware of a few more cooking.. but to my knowledge they aren't released yet. Did I miss any?

Sunday, 14 March 2021

The Thwarted Course

Word has spread as far as Birevia—masons are needed, or those willing to learn, to repair an ancient bridge. The silver is flowing, but why is nobody returning?

The Thwarted Course is my take on a classic setup: the ogre-troubled bridge. Here, however, the ogre isn't an isolated hermit, but the penultimate link in a dreadful chain of exploitation.

This location has had a long, long gestation, I think I started on this map seven months ago! Here's the illustrated bridge on its own, ready for you to either print and show to your players, or to restock however you like:

If you're using this location in your own setting, the bridge fits almost anywhere you have a steep-sided river in a sparsely populated area.

If you're using this with the setting from the Trilemma Compendium, the bridge stands in the far northwest, in hex 0107, upriver from Witchknuckle in northern Claimsun. If adventurers investigate the road, it should eventually connect to a Martoi cyst of some kind—a ruin at least, possibly a fortress.

* * *

Thanks to my generous patrons for supporting this adventure-making project, long have you waited for this installment!

Saturday, 20 February 2021

Nothing at the Bottom: MOSAIC Strict RPG Design

This blog post describes an experimental RPG design principle, Mosaic Strict. I'm defining this principle because I'm curious what kind of games result from applying it. (I have no idea!) I'm defining it very carefully because we live in a fallen world and shared understanding is fleeting.

MOSAIC is a set of criteria that might be true of an RPG text:

  • Modular
  • Optional
  • Short
  • Attested
  • Independent
  • Coreless

If all the criteria apply, then that text is Mosaic Strict. If only some (or none) of the criteria apply, then it's not Mosaic Strict. There are no partial points, it's all or nothing.

(If you use a hash tag for some reason, please use #MosaicStrict capitalization so screen readers handle it properly.)


Mosaic Strict RPG texts are meant to be used together with other RPG texts; each text only describes a portion of the rules that will probably be in use. One might describe an initiative system, another how combat works, another how gutter mage rituals work.

Anyone trying to have a pleasant evening of gaming will probably need to use several of them.

Modular: Any game text that explicitly claims to be
an entire, complete RPG is not Mosaic Strict.

This doesn't mean that all Mosaic Strict texts necessarily work well together! Any given pairing might be hot garbage. That's fine, because they're all..


Mosaic Strict RPG texts are all optional; you don't have to use them. Each play group will decide for itself which one(s) they are using. If a game text describes itself as mandatory, or necessary for the use of a certain gaming experience, it is not Mosaic Strict.

Optional: Any game text that describes itself
as necessary for play is not Mosaic Strict.

Please note that the modular and optional tests aren't about the rules of a text and whether they're "really" modular or optional; this rule is about how the text describes itself.


Mosaic Strict optional rules are concise enough to fit on a two-page spread, no more than 1500 words.

Short: Mosaic Strict game texts are no more than 1500 words.

This includes any and all words that are part of the publication, such as words in illustrations, the document's title, subtitle, headings, subheadings, byline, copyright notice, Mosaic Strict attestation (see below). If the text is longer than this, it's not Mosaic Strict.


If the game text doesn't explicitly say that it's Mosaic Strict, it isn't.

Attested: Mosaic Strict texts say they are Mosaic Strict.

This rule exists for three reasons:
  1. Mosaic Strict game designers, if there ever are any, have a slim chance of finding one another if their work is labelled
  2. It might help confused readers of your Mosaic Strict optional rules understand why it refuses to use the term saving throw for no good reason
  3. It makes the acronym work


Now the easy stuff is out of the way, here's a tough one, the rule that is really what Mosaic Strict is all about:

Independent: Mosaic Strict texts do not refer to the mechanics
or quantified state in any other game text.

By mechanics I mean any procedures (do this, then do this next, roll these dice, use dice at all) that tells you how to play. By quantified state, I mean any use of numbers, tags, attributes, binary or multi-state conditions (alignment, prone, not prone) to characterize what's going on in the game world.

Mosaic Strict texts don't refer to any rules whatsoever from other documents. Mosaic Strict texts do not build on one another, they don't assume you're using alignment or levels or that you have a Strength stat described in another document, none of that. There are no mechanical connections whatsoever between Mosaic Strict documents.

This means that a Mosaic Strict game text that describes a turn order for combat cannot assume that characters have Dex scores, or that the game is using 2d6. The combat turn order document could define those things

Now, game worlds have things that can be counted or quantified: money, a PC's height in inches; light switches flip on and off. A game text can refer to quantified things that exist in a setting and still be Mosaic Strict.

A two-page spread about being a wizard could say that you need $13,333 to join the magic society and that you must devour 13 white cats to be able to cast a tree-climbing cantrip up to three times a day and could still count as Mosaic Strict; these are all in-game quantities.

A game text that references any mechanical quantity from another text is not Mosaic Strict.


This isn't a separate rule, but a consequence of Independence that I need to be really clear about: There are no core rules and no character sheet at the bottom of Mosaic Strict, no standard interface of compatibility.

Coreless: assume nothing else is in use beyond free-form play

A game text could define a character sheet and still be Mosaic Strict, but any other document that requires its use is not Mosaic Strict.

A game text could define a universal resolution mechanic and still be Mosaic Strict, but any other document that references that resolution mechanic is not Mosaic Strict.

There's no central document or required rule at the heart of it all, only free-form play role-play and whatever assumptions a particular group brings with it.

A Few Clarifications

Q. Is this gatekeeping bullshit?
A. No! Or at least, I hope not. I intend this in the spirit of the 200-word RPG contest: a very specific set of rules to see what comes out of it. My first few conversations about this convinced me there's a powerful temptation to do almost what I'm describing, slithering back into much more normal game design, so I want to be really, really clear about what I'm talking about.

Q. This sounds stupid, how does this make a better game?
A. I doubt it will!

Q. Is there any Mosaic Strict actual play so I can see what the fuck you're talking about?
A. No. The Mosaic Strict criteria are meant to apply only to RPG game texts. They don't apply to play styles or culture, design aesthetics, genres, or anything else—just the text of the RPG.

This means that if a bunch of people get together, grab three Mosaic Strict documents (one for being a wizard, one for stealing things, another for driving too fast in a jalopy) but bring along their 5th edition play assumptions and experiences and use a d20 for everything, the criteria of Mosaic Strict have nothing to say about their play experience. If a group is playing Call of Cthulhu mashed up with cool take on slime infection in a Mosaic Strict two-pager, we don't say that their session was Mosaic Strict or not.

All play experiences incorporate elements from the players that are not in the rules. Mosaic Strict only applies game texts, not play experiences.

Monday, 15 February 2021

Realistic Kickstarter Goals

Using your entire project budget to set your Kickstarter funding goal will make your goal too high. The financially responsible funding goal only covers your remaining costs.

With The One Ring making a big splash on Kickstarter this month, there's been renewed talk of fair funding goals. Free League set its funding goal at 100K SEK (about $12,000 USD), which is unlikely to be enough to cover the costs of producing a full-color, 300+ page book full of art.

Is this too low? Are they gaming the system?

Maybe not.

Kickstarter (or perhaps just Kickstarter culture) rewards projects that make their funding goals quickly, which encourages low goals. Funding on the first day (or hour!) is particularly celebrated. It's tempting to set a funding goal that's much lower than the project's budget in order to be able to hit this funding goal very quickly. That success can be used for marketing oomph, building more buzz and sales in a virtuous cycle.

That is.. if you get enough backers. If you don't, have you basically gambled financial disaster to buy some marketing juice?

Two Crucial Funding Levels

There are two crucial funding levels, one of which we talk about all the time: break-even point. This is the point a which your project earns enough to cover its costs, and makes its first dollar of profit.

The other funding level is one I don't hear about much, and that's the "go/no-go" funding level. How many sales do you need from your Kickstarter to make continuing to production a good idea?

Intuitively, these are the same amount. Why would you go forward with a project that's going to lose money? Unfortunately, this assumption will make you pick the wrong funding level, one that increases your risks.

Sunk Costs

The flaw in this thinking is that by the time you Kickstarter, you've already invested a lot. You've probably at least got a first draft, you've put time into play testing, you've spent time lining up the production team, working out some of your logistics, and you've bought enough art to at least give your Kickstarter a chance.

On Kickstarter launch day, there's no way to "unspend" this money.

If your project is unprofitable, you can't go back in time and not do all that pre-work. This has a huge effect on how you set a financially responsible funding level.

What you need to look at is the various scenarios going forward from that point in time, and setting a funding level that guarantees you'll be in better financial shape than you are now. That funding level only covers your remaining costs! Why is this?


To answer that, let's look at hypothetical indie RPG project. I've picked March 1st as my KS launch date, and by that time I estimate that I will have sunk $5,000 into the project. (Let's say this is three weeks of full-time work at a 'livable wage' of $30/hr, plus $1,000 in assorted freelance help such as KS banner art.)

Having carefully thought out my remaining costs in time and freelance work, I project I'll need another $5,000 to finish the project. (Art, layout, copy-editing, time spent finalizing shipping logistics, coordinating freelancers, etc.)

I've priced the book at $30 (including shipping), and it costs me $20 per book to print, package, and ship to backers. The per-book margin is therefore $10.

This means that my break-even point is 1000 sales. That raises $30,000, which covers the $10,000 in fixed costs, plus the $20,000 in variable costs of delivering 1000 books.

However, my go/no-go point is much less, only 500 sales. Why is this?

Several Disasters

Let's consider several crappy outcomes. One is that I don't get a single backer. If this happens, I won't fund, and I don't go to production. I pull the plug on the project, and (crucially), I don't spend the additional $5,000 to complete the book.

This means that my total costs are the $5,000 I already put in (and a broken heart).

Now let's imagine that I got 900 backers. That's not enough to break even, because my total costs are $28,000. However, my revenue is $27,000, which means my final position is $1,000 in the hole. That's considerably better than $5,000 in the hole, which is where I'd be if I set my target above 900 backers. I'm $4,000 better off by going to production with 900 backers.

Clearly, 1000 is not the optimal funding goal.

With a funding goal of 500, as soon as I get 500 backers, my additional fixed spend on art to complete the project ($5,000) is exactly covered by the margin of my book sales (500 x $10). I'm no better or worse off than pulling the plug. At 499 backers, I'm slightly worse off ($5010 in the hole), at 501 backers, slightly better ($4990).

This is the balance point that matters, when my per-book margin equals my remaining fixed costs.

One way to look at this is to plot my project's final profit based on various funding goals and actual backer levels.

Here is the ideal profit line.

With a funding goal of 500 backers, there is never a point where I'm more exposed than my sunk costs of $5,000. $5,000 in the hole is as bad as it gets. At the same time, as soon as I get even one extra backer past my funding goal, my position is improving.

Let's compare that with other funding goals:

With a funding goal of zero (dark blue line), I've made the choice to spend the additional $5,000 in art costs whether or not I get any sales. This exposes me to some additional downside, as much as $10,000 in the hole if I get no backers.

With a funding goal of 250 backers (red line), I have some protection against the worst, but going into production with 250 sales means my final position is $7500 in the hole, considerably worse than 500 backers.

Now look at funding goals above 500 (e.g. green, orange, teal). Those goals protect me from deepening the hole, but they make it harder to start recouping my costs: it takes many more backers before my project lifts off that -$5,000 profit line. If my funding goal is 1000 and I only get 750 backers, my project doesn't fund. If my goal had been lower, I could have recouped some of my sunk debts, even though my project wasn't profitable overall.

Unconscious Signals

What's interesting about this is that it matches my own gut feel choices while planning my own Kickstarters. I had funding levels that were considerably below the total project budget, because I knew I was going ahead regardless.

With plans to use print on demand, once the book was actually written, there was very little reason not to go ahead with production, especially since post-Kickstarter sales also become available.

Of course, if you have a series of unprofitable books, you need to look at your operation pretty carefully. The time to consider the possibility of turning a profit is before you start sinking time and money into something. When you're hovering over the Launch Kickstarter button, however, very different calculations rule the day.

Sunday, 14 February 2021

Sci-Fi Where Art Thou?

Recently, I wondered to myself, "What's the 5E of sci-fi?" If one was going to create science fiction adventure content to supplement the most popular system, what would it be?

Class A Modular Ship by Galen Pejeau

Surely there's something d20-based floating around, but it took me a while to remember Stars Without Number. Why is that?

I grabbed the Q1 2020 Orr Report from Roll20's blog to see what their stats said. I did a quick classification of the top ~95% of campaigns (there's a pretty long tail), and here's what I got:

Here it's pretty clear—fantasy rules the roost by an astonishing margin. More than 60% of all campaigns were fantasy. There are plenty of campaigns that don't declare a system, but after that there's a solid chunk of Call of Chthulhu.

Sci fi is 1.9% of campaigns?!

Talking about this on Twitter, a number of people wondered:

  • It this just because of the 5E effect?
  • Is there a bunch of sci-fi gaming classified as horror? (e.g. ALIEN)
The answer to both questions is no. Here's 5E and all Uncategorized campaigns removed. Horror (as I said) is basically all various edition of Call of Cthulhu (with a homeopathic quantity of something called inSANe).

When I look across sci-fi, cyberpunk, and science fantasy (e.g. Numera), it's no surprise that the top systems are Star Wars (various), Starfinder, and Shadowrun. What's more surprising is those three heavyweights don't even account for 20 games out of 1000 on Roll20.

  • A number of definitely sci-fi systems like ALIEN and Coriolis, are lumped into system categories like "Year Zero". Other systems presumably have some sci-fi component, like FATE, Savage Worlds, and all of PbtA are lumped into system categories. Even so, that's only 2% of Roll20 campaigns in total, much of which is other genres besides sci-fi.
  • Roll20 campaign data may not be representative of overall gaming patterns. Anything that doesn't really benefit from maps and tokens may simply be using Zoom and Google Sheets, and not messing with VTTs at all.
If you're looking for sci-fi, Zine Quest 2021 is going now, and has several great-looking Mothership supplements!

Sunday, 4 October 2020

The Bombing of the Quick Stop Gas Station Bench

One of the ingredients that's coalescing for Bunkers & Billionaires is a completely surreal mission landscape. Rogue intelligence agents and terrorist cells do things that seem to make no sense, until you understand that they're the pawn moves in AI-powered late stage capitalism.

(What is Bunkers & Billionaires? It's the working title for a near future dystopian RPG that doesn't yet exist.)

I think it's easiest to illustrate with an example.

The Bombing of the Quick Stop Gas Station Bench

At 8:32pm, Nov 7, 2035, three self-professed members of "QMove" jump out of an electric van and blow up the graffiti-covered park bench in front of the Quick Stop gas station with a pipe bomb. 

One of the bombers returns to the scene on foot to quietly post live video of the Welland Corner fire department putting out the flaming wreck of the bench.

Later, on the way back to the fire station, the fire truck stalls out, frozen by a remote anti-theft command sent from head office at the request of a numbered corporation, the new owner of the fire truck as of 9:11pm. An argument breaks out between the firefighters and the young man who has turned up to drive the truck away as part of a gig he got ten minutes earlier on Fiverr. The situation is a tense stalemate until the Welland Corner police arrive.

After cuffing, tasing and pepper spraying everyone (in that order), the Welland police check their phones for follow-up instructions. Somewhat confused, they turn the truck over to the half-blind Fiverr guy, who (once he recovers) simply drives it back to the fire station and leaves the keys on the seat, as per his original instructions.

While this is happening, the stock price of Блакитна енергія (a Ukrainian fracking company) briefly dives as sell orders flood the market, but rebounds as they are snapped up by a second numbered corporation. This only takes a few hundred milliseconds, so nobody really notices.

* * *

This kind of thing happens constantly, the result of the collaboration between several forces: Liquidr (and similar asset management-as-a-service operations), freelance schismatics, and the National Commercial Interest Gateway (NCIG).


Liquidr is the category leader in the AMaaS space (Asset Management as a Service, or just "amass"). For a low monthly fee plus a percentage of capital gains, a Liquidr AI will manage individual, corporate or public entity assets. Initially this is simple investing, but it soon expands into automated mergers & acquisitions, initial public offerings and other ownership restructuring—to take full advantage of the rapidly changing regulatory landscape.

Liquidr: add liquidity to anything!

Higher tier Liquidr accounts are advertised as getting more AI time and better market intelligence, but for the most part it's just data mining privileged information from lower-tier Liquidr accounts.

("Are you liquid?" asks the flawless hottie in the ad)

Freelance Schismatics

Also in the mix are former intelligence services propagandists, now freelancing to create dissociant conspiracy theory groups with actionable false belief systems. These come complete with a custom target designation symbology, so they can be weaponized as deniable lone wolves for sale.

QMove is just one of many violent splinter conspiracy groups whose ideology and identity is the deliberate product of manufactured information. (Having no influence over its own destiny whatsoever, QMove doesn't get its own heading.)

Eventually, schismatics realize that control of dissociant groups count as assets, and can be managed as part of AMaaS portfolios, which dramatically shortens the time between radicalization and action.

National Commercial Interest Gateway (NCIG)

The National Commercial Interest Gateway started out as a joint venture between an AMaaS consortium and the US Department of Justice to create an auction system to designate private commercial interests as being of significance to the national interest.

The time to make back room deals with key government contacts was identified as a significant source of friction in the AI age. Why not add liquidity™ and make it a marketplace?

Now, multinationals no longer had to play all that tedious golf to convince their friends in the CIA that there are billions at stake if such-and-such a country gets the wrong election result, or to persuade their DOJ contacts that this or that movement should be classified as a domestic terrorist organization. Just put up a bid on NCIG, and the matter is resolved immediately.

During the kleptocrat-led weakening of federal government institutions, NCIG rapidly became a welcome source of clarity in a time of confusion, despite early criticism that it was simply "Fiverr for spies."

Things Get Weird

After the complete shut-off of federal subsidies in 2024, state, provincial and municipal governments turn to Liquidr out of desperation in an attempt to find creative ways of covering climate reconstruction and refugee management costs.

In 2029, Liquidr acquires Legal Dynamics, a machine learning-powered contract negotiation company. Liquidr starts using this to better exploit municipal asset portfolios, and it starts writing laws solely to create just-in-time opportunities for Liquidr gambits. Things that were once illegal become legal, temporarily legal, or mandatory, all as part of creating pathways to new ownership structures and financial instruments. These changes could be permanent, or last less than a second—just long enough for a complex trade or restructuring to take place.

By 2035, laws are completely indecipherable to people, a stew of machine-generated legalese that grows by terabytes per minute.

The Quick Stop Bench Bombing, Redux

In October 2035, the owner of the Quick Stop Gas Station places ownership of the gas station into her Liquidr portfolio (a Hustle! tier account). Moments later, the bench out front is carved off into a numbered corporation subsidiary, and has an automated initial public offering on the basis of its expertise at natural gas management. The Hustle! AI transfers the entire value of the Quick Stop gas station to the park bench, reverse mortgages it, and uses the proceeds to buy Блакитна енергія stock. This is a bit risky, maybe, but it's part of an A/B test. It could be great!

(The Hustle! tier AI has no idea what a park bench is, only that it technically counts as real estate and—unlike the gas station business, has no employees to complicate its tax situation.)

Meanwhile, a tier 9 Liquidr account checking its notifications sees the gas station's precarious situation and takes immediate action, doing three things simultaneously:

One, it injects a rider into an anti-competition settlement it's negotiating with the Michigan state legislature, making it temporarily illegal for Michigan-based corporations to own financial instruments in the Ukraine if they hold no US real estate. The Michigan state legislature account AI grants this rider cheaply. Michigan doesn't own any such instruments at the moment, and the proposed rider expires in three hours anyways, so no big deal. The rider goes into effect at 8:03pm, becoming law.

Secondly, the tier 9 account places an NCIG bid to have the Welland Corner Fire Department designated as a domestic terrorist organization. WCFD doesn't appear in the NCIG database, but the briefing attached to the bid has solid intel that WCFD has only twelve members (barely a cell) and the required action section is blank, so again, this is inexpensive.

Thirdly, it buys the activation key for a small QMove cell for $820. The control software sends a dynamically generated video to the three members in Welland Corner, alerting them that the park bench outside the Quick Stop Gas Station is the target they've been waiting for. (You can't see it at night in November, the video explains, but it's where Antifa has used graffiti to spread their plan to bring socialism to the gaming consoles in electric vehicles.)

* * *

QMove springs into action at 8:22pm. (The pipe bomb has been sitting ready by the front door for weeks, and they only live around the corner from the gas station, but Alex is so excited that it takes him six minutes to find the keys for the van.)

By 8:44, 720 people have watched QMove's vertical video of the bench fire, enough for a YouTube algorithm to recognize it and tag it as footage of a terrorist incident. The tier 9 updates the NCIG bid with the video, recommending a 50% commission for itself on any resulting asset forfeiture arising from law enforcement action.

Armed with video of destruction of its only US real estate, the tier 9 triggers de-legalization of the bench's holding company. Benchco's assets in the Ukraine are sold in less than a second, with the tier 9's preregistered lowball bids scooping up the shares.

The schismatic control software reports a near certainty that the QMove cell can be reactivated again, later (even for non-bench targets), so the tier 9 sells the activation key back to the schismastic's Liquidr account for 78 cents on the dollar.

By 9:15, it has also sold the fire truck back to Michigan state, exactly offsetting the cost of the legislative rider before it even expires!

Fucking tidy, it thinks.

* * *

By 10:28pm, the gas station's Hustle! tier Liquidr AI has realized that something really bad is up. With no assets, and a fresh invoice from the fire department for putting out a fire of some kind, the account's net worth is suddenly deeply negative. For Hustle! tier customers, this triggers auto-bankruptcy mode, instantly terminating the whole account to prevent the customer from becoming even less profitable by letting them complain to customer support.

Its final action is to list the gas station personnel in an eviction authorization for the property's new owner, the Welland Corner Fire Department, in consideration for 33% of the value of any assets seized from its deadbeat former owner.

Bunkers & Billionaires

As I mentioned, Bunkers & Billionaires doesn't exist, but even nonexistent games must answer the question: who are the PCs and what do they do?

They're the people who are trying to destroy this system so that something better can take its place.