Game Systems/Campaigning

The campaign week

Electioneer is played in weeks. Each turn is one week of campaigning, and the week is planned as a whole before it resolves.

Planning a week#

A week has 7 days, and you assign actions to specific days. The engine resolves the week in day order, so sequencing within a week matters.

Two limits apply:

Candidate-days. Your candidate has a pool of actions per week, refilled at the start of each turn. This is the real scarcity — the candidate can only be in one place at a time.
Three actions per day.

Surrogates and your running mate exist precisely to work around the first limit: they campaign where the candidate cannot.

What decays, and what does not#

At the end of every week the transient quantities decay. This table is most of the game's strategy:

Quantity	End-of-week treatment
State momentum	×0.7 (zeroed below 0.3)
National momentum	×0.75
Media narrative	×0.8
Bloc enthusiasm	relaxes 40% toward neutral
Debate/policy prep	relaxes 35% back toward the candidate's real traits
Ground game	no decay

Ground game is the only thing that sticks. Everything else you buy is rented. That single asymmetry is why early field investment pays off and why a late advertising blitz cannot substitute for organisation you never built.

Prep decaying at only 35% a week is deliberate too: prepping a week or two ahead of a debate still pays off at showtime.

The recap#

After a week resolves you get a recap. It groups every cause logged that turn by its human-readable label, sums the margin each contributed, and sorts by absolute impact — led by the change in projected electoral votes.

This is not a summary written after the fact. The causes are the model: your campaign margin in a bloc is literally the sum of the causes logged against it (see The vote model). The recap is the model showing its working.

The timeline#

Once at game start and once at the end of every week, the game records a trend sample: national poll, electoral votes for each side, tossup EV, momentum, cash, and the Democratic share in every state. That is what drives the trend charts.

Difficulty#

Difficulty configures the opponent, not the arithmetic:

	Efficiency	Mistake rate	Foresight
Easy	0.55	30%	3
Normal	0.80	15%	5
Hard	1.00	5%	8

Foresight is how many weeks ahead the AI plans. On hard it plans 8 weeks out and makes a mistake one time in twenty.

The AI also flexes its competitive band — how far from a 50/50 contest it will reach — against the scoreboard. When it is behind it reaches further; when it is comfortably ahead it defends only genuine tossups.

What this means in play#

Spend candidate-days on what only the candidate can do. Rallies need them; surrogates and ads do not.
Build ground game early. It is the only permanent asset. Momentum bought in week two is gone by week four.
Sequence within the week. Days resolve in order.
Read the recap, not your instinct. It tells you exactly which of your actions moved margin and by how much.
A comfortable lead makes the AI narrow its focus. Extending a lead gets easier in the tossups and harder everywhere else.