Machines, Points & Threat

Points, machines and threat level are one system with three faces, and the fastest way to break a build is to tune one of them without the other two.

The loop

A crew kills things and holds rooms, which earns points. Points buy things from machines. Most machines raise the site's threat level. Threat level decides which rung of the escalation wheel a fight opens on, which decides how much is coming, which decides how many points a crew can earn — and how quickly they need to spend them.

That is a closed loop and it is self-tuning as long as you respect two rules:

1. Some purchases must be quiet. In Attrition Protocol the wall racks and the perk machines cost nothing in threat. If everything is loud, the crew stops spending, the loop stalls, and the economy becomes a number on a sheet.
2. Nothing pays out except the field. No machine returns points, refunds a perk, or lowers threat as a purchase. The moment one does, the crew has found a pump and the ladder stops meaning anything.

Pricing anything new

Price against the perk shelf, not against realism. In the shipped rule set a perk is 25–60, so:

• A wall weapon is a fraction of a perk (15–75). It should be affordable in the first hour.
• A gamble is just above the best perk (95). The Mystery Box is priced so that every spin is a Juggernog nobody drank. That competition is the entire tension; discount the box and you delete it.
• A campaign goal is an order of magnitude up (500). Two sessions of not buying anything else.

Then check the earning table. A crew pulls 200–350 a session at the published rates. If your new price is not legible as a fraction of an evening, it is the wrong number.

Writing a machine

The machine file type covers anything that takes points or a lever — boxes, racks, upgrade cabinets, power switches, traps, teleporters. A machine file is finished when it has all five of these:

• A procedure. Numbered steps the GM can run mid-scene without reading anything else. A machine with no procedure is scenery.
• A cost and a threatChange. What it takes from the pool and what it costs the room. Both, always, even when one of them is zero — a zero says something.
• An activation. What has to be true first. The best gates are not money: Pack-a-Punch is gated on three linked teleporter pads, which ties the economy to the site's central puzzle and charges 15 exposure before it charges a point.
• A relocation or a limit. The thing that stops it being an infinite tap. The box moves after five spins; the racks refuse at threat 8; the fly trap fires once. Without this, a machine is a vending machine and the crew will stand at it.
• A secret. Who built it, what it is really doing, what it takes that it does not advertise. In this world none of the machines were installed by anybody, and that is the point of the whole set.

Use the outcomes table for anything with a spread of results, and stock for the pool the table draws from. Keep the two separate — the table decides which band, the stock decides what is in the band, and you can restock a machine for a new site without rewriting its odds.

Writing the threat ladder

Every rung needs three things: what the site is doing, a standing effect that is true for as long as it sits there, and an opening pressure naming the escalation rung a fight there starts on.

The opening pressure is the load-bearing field. It is what makes a loud site structurally different rather than merely busier, and it is the reason retreating to somewhere quiet is a real strategy instead of a delay.

Keep the trigger list exhaustive. Publishing every way threat moves — and telling the GM to say out loud when they are improvising past it — is what stops threat level feeling like the GM's mood with a number on it.

Always give the crew levers downward. Absent decay, silence, holding a room, sealing a breach. A ladder that only climbs is a countdown, and a countdown is not a system.

Adapting it

The numbers here are tuned for a horde game with a shared pool. To move the frame:

• A scarcity game — cut the earning table in half and leave the prices. The crew will never own a Thompson and the wall becomes a museum they walk past.
• A heist or a settlement game — make the currency something the crew produces rather than kills for, and let the machines be people. The loop survives; only the fiction changes.
• Personal wallets instead of a pool — this removes the argument, which is most of what the pool is for. If you do it, raise the prices, because five separate wallets buy far more than one shared one.

What you must not do is unpick the join between machines and threat. Take that out and you have a shop in a horror game, and the crew will notice within one session that spending money has stopped being frightening.