Dispatches

Most political games use randomness as a design crutch. An event card fires. A roll goes against you. The outcome feels dramatic, but it is arbitrary, and you cannot learn from it, because nothing you did changed the odds.

Campaign Margins works differently. Where chance enters, it is weighted by decisions you already made. Your candidate's strengths, the platform you built, and the coalition you assembled set the odds before the moment arrives, so a well-run campaign is favored and a sloppy one is exposed. Which means when your poll numbers drop, there's a reason. When a ward flips against you, something caused it. The volatility in the game comes from competing strategies interacting. From the other players making their own decisions, from the resource tradeoffs you made in week two catching up with you in week four. Real variance, not a reliance on luck.

That's also what makes a post-game conversation interesting. Not "the dice were against us." Instead: "why did you spend your war chest on ads in week three instead of saving it for closing arguments?" The decisions are all visible. The reasoning can be interrogated.

"Your decisions set the odds. Every result traces back to a choice you made."

Campaign Margins tries to model the part of a real campaign that is actually in your hands: which neighborhoods to contest, when to go negative, where to concentrate. The feeling of being accountable for your choices, from the war room, with imperfect information and a clock running down.

The most common mistake new players make is treating endorsements like a bidding war. Whoever commits the most capital wins, right? Not quite. Across the Campaign Margins games, from the endorsement sprint in Who's Your Mayor? to the convention floor in The Primary, fit gates the transaction. An endorser who's misaligned with your candidate's platform won't endorse them at any price. Capital only decides among candidates who are already acceptable.

This is how it works in practice. The neighborhood council endorses candidates who prioritize community engagement. The business association endorses candidates who favor growth. A labor union endorses candidates with labor-aligned platforms. Each has a threshold of alignment below which no bid wins. If your candidate doesn't clear it, your capital stays in your pocket and the endorsement goes to whoever does, even if they bid less.

The strategic implication: coalition-building is about alignment first, resources second. A candidate with strong platform fit and modest capital can beat a candidate with weak fit and a massive war chest, every time. The money is a tiebreaker, not the primary variable.

This reflects real endorsement politics accurately. A union will not endorse a candidate with an anti-labor record for any sum. A newspaper editorial board that values institutional credibility won't endorse a candidate who's spent the campaign attacking institutions, no matter how much the campaign lobbies them. The relationship and the alignment have to come first. The resources are what you deploy once you've earned the right to be considered.

The corollary: when you're building your candidate's platform, think about which endorsers you're trying to qualify for, not just which ones you want. The question isn't "can I outbid the other managers for the labor council?" It's "does my candidate's platform give them a reason to consider me at all?"

This is not just a game mechanic. The 2025 New York mayoral primary is a useful illustration: the best-funded candidate in the field lost, while the winner cross-endorsed with two rivals and consolidated a majority once the ranked rounds ran. Coalition work did not substitute for resources, but resources plainly did not substitute for coalition work either.

Every stance you take in The Primary is tagged two ways: what it does for you with the primary electorate, and what it costs you in November. Move left to lock down the base and you open a lane for the general-election opponent to call you extreme. Tack to the center to look electable and you give a rival room to out-organize you with the activists who actually turn out in a low-turnout primary. There is no free position. You are choosing which problem you would rather have.

This is Anthony Downs' spatial model turned into a decision you have to live with. Downs predicted candidates converge on the median voter. The primary breaks that logic, because the median primary voter is not the median general-election voter. The people who nominate you sit well off center. The people who elect you in November sit in the middle. Win the first electorate too hard and you have moved somewhere the second one will not follow.

Richard Fenno called this the gap between the constituency that nominates you and the constituency that elects you, and the two-track act candidates use to survive both. The Primary makes you run that problem in real time. The coalition you build to win the room is not always the coalition that wins the district.

Then there is the convention, where the math turns unforgiving. Endorsement takes a sixty percent supermajority of the whole hall, and the uncommitted delegates count against that threshold, not toward it. A plurality is worthless. You have to actually assemble the supermajority, or the convention ends in no endorsement and everyone limps into the primary bloodied. Uncommitted is not neutral. It is the wall you have to climb.

The ballots narrow the field. The lowest-place candidate drops on a rising threshold, the uncommitted delegates break, and you either build to the number or you deadlock. There is a hard cap of three ballots, so you cannot grind forever. You court the blocs you can actually move, you spend your capital before the break, and you live with where the room lands.

"The coalition that wins the room is not always the coalition that wins the district."

The Primary is not about being the most popular candidate. It is about understanding that popularity in one electorate is a liability in another, and managing that gap on purpose. Win the base, hold November. Both count, and they pull in opposite directions. That is the job.

Start with what makes the game different at all. The solo Campaign Margins games put you against a system: the opponent is simulated, the fog of war is real but the other side is not actually reading you. Who's Your Mayor? puts three to six real people in the same room, each running a rival campaign on a phone, all resolving on one shared screen. The fog of war stops being a mechanic and becomes social. You don't know which wards your rivals are contesting, who has been leaking to the feed, or who just filed opposition research on you.

What the three cities change is where the votes actually live. Each map has its own wards, its own coalitions, and its own local voices in the feed, and those decide which move is correct. A ground game that dominates a dense, contested ward in one city is wasted in a city where the same votes sit somewhere else. You cannot bring a memorized opening from one map to another. You have to read the board in front of you.

Apple Valley is a suburb, where the race turns on a handful of contested wards like Riverside and Oakwood and the anti-establishment streak in Southgate. Juniper is a mountain town with a developer's shadow over it, where growth and who benefits from it is the fault line. Fremont is a river city with its own institutions and its own grievances. Same rules, different terrain, different winning coalition.

The endorsement sprint is where any of the three cracks open. Sealed capital bids, more managers than endorsers, and fit gates the transaction before money does. Over-commit and you lose capital you cannot recover. Under-commit and you lose the endorser to someone who read the room better. It is a few minutes of silent, simultaneous negotiation, and then the results post and half the room recalculates everything.

"You cannot bring a memorized opening from one map to another. You read the board in front of you."

This is why the game holds up to repetition. For a group that plays often, three maps is three distinct strategic problems, not one game with new paint. For a classroom or a club, it means you can run it again next week on a different board and the winning strategy is genuinely different, so nobody coasts on last time. The rules stay honest. The city decides what winning looks like.

Political science has spent decades on the questions campaign managers ask in real time. Where do you compete? How do you allocate what you have? What does contacting a voter actually accomplish? Campaign Margins is built on a handful of those findings. Here is where they show up.

The median voter, and where it breaks

Anthony Downs' spatial model predicts that candidates converge toward the middle. The Primary is built on the place that logic breaks: the electorate that nominates you is not the electorate that elects you. Richard Fenno described that gap as the difference between the constituency that nominates and the constituency that elects, and the two-track act candidates use to survive both. Every platform stance in The Primary is tagged twice, what it does for your base and what it costs you in November, so the squeeze stops being a diagram and becomes a decision you have to live with.

Who's Your Mayor? complicates it further. With three to six candidates competing at once, moving to the center means splitting the same voters with four rivals. The managers who win are the ones who find an underserved coalition, commit to it early, and defend it when someone tries to poach it late.

Minimum winning coalitions

William Riker's size principle holds that rational actors assemble exactly enough support to win and no more, because a larger coalition splits the same spoils among more members. Campaign Margins applies that logic to resources rather than spoils: once a ward is yours, the next dollar is worth more somewhere still contested. The school board game makes it explicit. You have five precincts, you need three, and the opening triage of where to contest, hold, and concede is the whole problem. The Primary's convention is the same problem under a clock, where a plurality is worthless and uncommitted delegates count against the threshold rather than toward it.

What voter contact actually does

The field-experiment literature on voter contact, most associated with Alan Gerber and Donald Green, finds face-to-face canvassing to be the most reliable way to raise turnout, and impersonal contact to do far less. The harder finding, from Joshua Kalla and David Broockman, is that campaign contact does very little to change whose side a voter is on, at least in a general election where partisanship already supplies the answer.

The games follow that split. A ground game raises turnout in the zone you send it into, scales with the volunteers you recruited rather than the money you spent, and pays off hardest where your support already lives. Paid media moves numbers less efficiently. The temptation to buy ads instead of knocking doors, because ads feel bigger, is the mistake the research documents in real campaigns.

Why this matters for the classroom

Each of these ideas is easier to teach after students have made the decisions themselves. Assign The Primary before a lecture on primaries and students arrive having already felt the base-versus-November squeeze. Run a forty-minute round of Who's Your Mayor? and it generates the mistakes the theory explains, live. Students who spread resources evenly across six wards and finished third understand minimum winning coalitions before you have said Riker's name.

Campaign Margins is free, runs in the browser, and needs no install or account. If you want the concept-by-game map, debrief prompts, and reading pairings, the classroom guide has them.

The games are not the lesson. The decisions students make in them are. Everything about teaching with Campaign Margins comes down to one move: let them decide first, then teach the concept that names what they just did. A student who spread resources across six wards and finished third understands minimum winning coalitions before you have said the words.

The lowest-friction way in is a pre-class assignment. Send one single-player game as homework before the relevant lecture. The Primary before you teach primaries and spatial voting. The Mayor before incumbency. The Initiative before direct democracy. Students arrive having already made the calls, where to spend, which coalition to court, when to go negative, and your lecture becomes a debrief of what they did and why it did or didn't work. No account, any browser, under half an hour.

The higher-energy option is Who's Your Mayor?, run live. One screen on the projector, students split into campaign teams on their own devices, forty minutes, an election-night reveal at the end. It generates the mistakes the theory explains, in real time, in the room. The endorsement sprint alone teaches more about coalition strategy than a slide can, because a team just lost an endorser by misreading it.

The debrief is where the learning lands, and the questions make it. Ask the team that lost: where were you strong, and where were you merely present everywhere? Ask the team that went negative: what did the attack cost you that you could have spent converting your own voters? Ask the winner to reconstruct the one decision that turned it. Because nothing in the games is arbitrary, every answer traces to a choice, and the reasoning can be interrogated. That is the difference between "the dice were against us" and "we spent week three on ads instead of the ground game in the ward we needed."

If you want a graded artifact, have students write a short strategy memo after they play: what they decided at the midpoint, what it cost, and what they would do differently. Every student has a different campaign to write about, because no two playthroughs are alike. It reads as analysis, not a game report, and it is hard to fake without having actually played.

"The games are not the lesson. The decisions students make in them are."

You need nothing but a link, and for the live game, a projector. The mechanics are built from the same research you already assign, so the strategy that wins in the game is the strategy the literature says wins. Pair a round with the reading and the two reinforce each other. The classroom guide has the concept-by-game map, debrief prompts, and reading pairings if you want them.

The election night results had just posted. Rae's campaign had won Apple Valley with 44 percent, a comfortable margin in a five-way race. Dana's machine had finished second at 29 percent. And the player who'd run Sam's tech mogul campaign wanted to know why their numbers had stalled in week three and never recovered.

The answer was in the decision log. Sam's manager had split their action budget evenly across three neighborhoods every week, a little canvassing here, a media buy there, lawn signs somewhere else. It felt like coverage. What it actually was: thin presence everywhere, dominant presence nowhere.

The concentration math is unforgiving. When two candidates contest the same ward, the converted voters don't split proportionally, the candidate who out-commits the ward wins a disproportionate share of it. Sam had been competitive in six neighborhoods and dominant in zero. Rae had written off two wards entirely and poured everything into the four that mattered. Same resources, different result.

Dana's manager had a different question: why hadn't the attack ads worked? Three weeks of negative campaigning against Rae, and the gap had barely moved. The problem wasn't the attacks, it was the timing. Going negative in week one, when Rae was still building momentum and the undecided pool was large, had consumed resources worth more as ground game. By the time the attacks landed, Rae had already locked in enough support in Riverside and Oakwood that suppressing her elsewhere couldn't close the gap.

The Forum question had mattered more than either of them realized. Rae's answer had moved numbers in Southgate, the most anti-establishment ward on the map, by almost four points in a single week. Neither rival had been watching that ward closely enough to respond.

Every swing traced to a decision. That's the whole point.