Collective action / Teaching guide
The Square
Created by Yoav Duman · edition 2026-10-10.1. The Square is a fictional collective-action game. One student controls Citizen B1; the other nineteen citizens are simulated. No classmates, sign-in, network, audio, or AI service are required during play.
Student instructions
- Read the setting and BASE payoffs. Complete the rules check before voting.
- In each of four rehearsals, forecast how many of the other nineteen citizens will JOIN, then choose your own ballot. Either input order works. Use a whole number from 0 to 19.
- Lock one ballot per round. The political outcome resets between rehearsals; your points accumulate. Recorded conversations, proposals, and organizational choices stay fixed. Back lets you review them and revise your written explanations.
- After every reveal, compare your alternative ballot while holding the other ballots fixed. In rounds 2–4, also inspect the comparison with the new intervention removed. Open the model details to see how simulated citizens were generated.
- Complete both fixed festival comparisons, match the concepts, and explain free riding, coordination, and one model limit in your own words. Character minimums are only a way to record a response; they do not grade the explanation.
- At the end, print/save a PDF or download the text report. Submit it in the way your instructor requests. Nothing is submitted automatically.
Saving and moving your work
Check the status below the header. Progress saves in this browser when storage is available. Website play and a downloaded HTML file have separate browser storage. Before moving device, browser, file, or website, choose Download backup. The JSON checkpoint restores decisions, forecasts, writing, seed, and progress through Restore backup. A PDF or text report is readable work for submission; it is not restorable.
Older v3/v3.1 browser drafts use the same save key. If you are moving an older draft to this website, open the old game, click Save Progress, and keep that original file/browser until you have recovered the record. Old editions have no download-backup button. A new game cannot read storage from another website or local file. Existing records whose earlier settings cannot be verified are retained and labeled; their intervention comparisons are withheld.
If browser saving is unavailable, keep the page open and download a backup. If another tab changes the draft, saving pauses to protect both copies. Instructions offers a choice of drafts. Restart replaces this run only after you choose; download a backup first. Restart with the same seed to compare strategies under the same round variation, or use a new seed.
For instructors
Purpose and classroom setup
Students should distinguish wanting a shared benefit from supplying it, uncertainty about complementary contributions from participant-only incentives, and a pivotal vote from a contribution that cannot alter the outcome. Assign solo play, or ask a pair to discuss one shared B1 decision. This is a single-player simulation; classmates do not provide the other ballots.
Plan about 45–55 minutes: 5 for rules, 15–25 for four rehearsals and responses, 10 for comparison labs and synthesis, 10 for discussion, and 5 for export. These are estimates; an extended or take-home assignment permits closer model inspection. Give the canonical game link or the standalone HTML and specify whether students submit the PDF or text report. Do not ask them to submit only their score.
What each stage changes
| Stage | What changes | What stays fixed |
|---|---|---|
| 1 · BASE | No intervention. Establish a forecast and private ballot. | Twenty citizens; threshold 12; BASE payoffs. |
| 2 · local talk | Your selected conversation changes scripted tendencies in B. Other districts receive preset local communication adjustments. | National threshold and numeric payoffs. Local statements do not bind ballots. |
| 3 · coordination | Accepted targets produce a common coordination signal. Strong requires at least 12 accepted and no over-ask; 10–11 accepted, or 12+ with an over-ask, is partial; below 10 is weak. | BASE payoffs, earlier conversation, district caps. Targets are not turnout quotas. |
| 4 · organization | B gets BASE, SUPPORT (failed JOIN −1), BENEFIT (successful JOIN +5), or PURPOSE (no numeric change). Different types have different scripted responses. | Threshold, plan, conversation; A, C, D keep BASE. |
How participation is generated
The nineteen simulated citizens have fixed base indices and types: committed, conditional, cost-sensitive, or skeptical. They JOIN when base + local communication adjustment + coordination adjustment + organization adjustment + seed/round variation ≥ 50. These are scripted participation indices, not measured probabilities, utility values, or predictions of real people. Citizens are not solving for an equilibrium or automatically choosing the highest numerical payoff.
The variation is deterministic for each citizen, seed, and round: a value from −3.50 to +3.49. It can change between rounds, including with BASE in round 4. Prior totals therefore mix the intervention with round variation. The added same-round comparison holds the citizen, seed, round variation, and player ballot fixed while removing only that stage’s intervention. It isolates the programmed effect, not a real-world causal effect.
The model-details table after each reveal supplies all nineteen base indices and every adjustment. Round 2 in B: ASK changes conditional/cost by +5/+1; PLEDGE changes conditional/cost/committed/skeptical by +7/+2/+1/−1; PURPOSE changes conditional/committed/skeptical by +2/+3/+3; LISTEN changes conditional/cost by +2/+1. Elsewhere A and D use conditional/cost +3/+1 and C uses +4/+1. Strong coordination adds conditional/cost/committed +10/+3/+2; partial adds conditional/cost +4/+1; weak subtracts 2 from conditional. In B, SUPPORT adds cost/conditional +8/+2; BENEFIT adds cost/conditional/skeptical +13/+4/+2; PURPOSE adds committed/conditional/skeptical +3/+2/+3. BASE adds zero.
Debrief with evidence from the record
- Why can a person want reform and prefer HOME? Under BASE, if reform happens either way, HOME gives +4 and JOIN +3. If reform fails either way, HOME gives 0 and JOIN −3.
- When is your own ballot pivotal? With eleven other joiners, HOME leaves the total at eleven; JOIN makes twelve. At twelve or more others, your own ballot no longer decides success. Ask students to use the alternative-ballot comparison from their own run.
- What did local talk tell you, and what did it leave unknown nationally? Distinguish a spoken intention, an accepted target, and an actual ballot.
- Did the national plan change your numeric payoffs or the common expectation? Explain why an accepted target can still differ from turnout.
- What does a participant-only benefit change with the crowd held fixed? In the festival lab, HOME +4 beats BASE volunteer +3, but BENEFIT volunteer +5 beats HOME +4. SUPPORT narrows −3 to −1 in the failed case, yet HOME 0 still pays more.
- Would round 4 with BASE be a perfect control for round 3? No: round variation can change. Use the same-round intervention comparison. Ask which assumptions would need evidence before applying this story to a real case.
Assessment
Use four criteria, each 0–2: (1) accurate threshold/payoff comparison, including pivotality; (2) distinction between free riding and coordination; (3) explanation of the intervention with conditions held fixed; (4) a specific model limit and evidence needed to examine it. For each, 0 = missing or mistaken, 1 = partly accurate without a supported explanation, 2 = accurate reasoning linked to a recorded comparison or example. A high score, reform success, correct matching, or minimum-length text alone does not demonstrate understanding. HOME can be a well-supported choice.
The report retains current explanations and locked forecasts/ballots, rather than a first-draft writing history. Label incomplete records as drafts. Ask students to defend revisions orally or in a short addendum if revision is part of the assignment. These checks do not establish learning effects or replace assessment by an instructor.