AP Practice

Browser-based practice quizzes with four subjects, playable solo or as a multiplayer match. No build step and no dependencies — open practice.html.

Modes

Subject	Content	Filter by
APUSH	232 questions	Period 1–9
AP Gov	257 questions	The nine required foundational documents
AP Gov: 2.3-2.8	98 questions over 56 key terms	Institution and difficulty
Prefix & Suffix Practice	115 questions over 62 affixes	Difficulty and prefix/suffix focus

Single player runs at your own pace: answer, read the feedback, continue.

Multiplayer is peer-to-peer over WebRTC — no server, no accounts. One player hosts and shares a five-character room code or an invite link; everyone answers the same question each round, sees who has locked in, and gets a standings board and a powerup shop between rounds. Players can join a match already in progress, and a dropped connection reconnects into the same seat with its score intact.

Scoring is shared across all four subjects. With the timer on, a correct answer is worth 60–100 points depending on speed; with it off, every correct answer is worth 100. Streaks count consecutive correct answers, and your best streak per subject is remembered locally.

Files

practice.html          markup and screens
styles.css             all styling
app.js                 screens, solo play, and the multiplayer client
match.js               the authoritative match rules: phases, scoring, powerups
net.js                 transport: room codes, heartbeats, clock sync, reconnects
questions.js           APUSH + AP Gov banks
gov-branches-data.js   AP Gov 2.3-2.8 key terms and question bank
affix-data.js          prefix/suffix vocabulary and question bank
prefix_suffix_vocabulary.csv   source of truth for the affix vocabulary
tools/validate-content.js      content checker
tools/test-match.js            match rule tests

Content and engine stay separate: no question text lives in app.js, and no game logic lives in the banks. match.js and net.js keep that going — rules with no DOM, transport with no rules.

The <script> and <link> tags carry a ?v= stamp. There is no build step, so bump it whenever markup, styling and scripts change together, or a browser holding one of them from cache will mix old and new.

How multiplayer works

The host is the only authority. It runs a Match from match.js and is the single place scores, phase changes and powerups are decided. Everyone — including the host — then draws from the snapshots it publishes:

client --{answer | buy}--> host --> Match --> {sync | event} --> every client

That shape is what keeps four browsers agreeing. A few consequences worth knowing:

• Nobody scores themselves. A client reports which option it picked and how much time its own clock showed; the host caps that against what it actually observed and works out the points. Two players cannot drift apart.
• The answer key is withheld. A question snapshot carries the options but not the answer or the explanation; both appear only once the round is revealed. Powerups that grey out wrong options are resolved host-side for the same reason.
• Deadlines are absolute and clock-corrected. The host stamps a deadline in its own time; clients estimate the offset from ping round trips (keeping the lowest-latency sample) and re-sync on every snapshot, so a slow link loses latency rather than accumulating drift.
• Nothing waits forever. A round ends when every connected player has answered or the deadline passes. A player who disconnects is dropped from the count within a few seconds, and one who joins mid-question sits that round out instead of holding it open.
• Identity survives a reconnect. Players are keyed by a per-tab id, not by the WebRTC peer id, which changes on every reconnect. Coming back restores the same seat and score. Two tabs of one browser count as two players, which is the easiest way to try a match on your own.

net.js handles the wire: namespaced room codes on the public PeerJS broker, heartbeats, reconnection with backoff, and a TURN relay fallback — without one, a fair share of networks (school Wi-Fi especially) cannot establish a peer connection at all. To use your own relay instead of the public one, replace the credentials in ICE_SERVERS at the top of that file.

The rules are testable because they have no DOM and no network, and take the clock as an argument:

node tools/test-match.js

That covers scoring and speed bonuses, timeouts, mid-match joins, disconnects and reconnects, every powerup, the purchase cap, and the guarantee that a question snapshot never contains its own answer.

Adding questions

APUSH / AP Gov — append a row to the matching array in questions.js. The correct answer is always written first; the engine shuffles the options at runtime.

[3, "Question text?", "Correct answer", "Wrong 1", "Wrong 2", "Wrong 3"]

AP Gov: 2.3-2.8 — append an object to govBranchesBank in gov-branches-data.js.

{
  id: "gb143", term: "gridlock", branch: "congress", topic: "2.3",
  questionType: "definition", difficulty: "medium",
  question: "As the AMSCO text uses the term, gridlock refers to ...",
  options: ["opposing forces congesting the process ...", "...", "...", "..."],
  answer: "opposing forces congesting the process ...",
  explanation: "Gridlock describes the jam itself rather than any single rule ..."
}

• branch is congress, presidency, judiciary, or mixed for items that turn on two or three institutions at once. It drives the Institution filter, and a session is drawn round-robin across branches so no one institution dominates.
• difficulty is easy, medium, or hard; a session ramps easy → hard.
• questionType is one of definition, application, scenario, compare, document, case, process. The engine spreads types out inside each difficulty band so consecutive questions do not feel identical.
• term must match an entry in govBranchesVocab (use mixed for items that compare several terms), and topic is the AMSCO topic number.

govBranchesVocab is the key-term list, taken from AMSCO United States Government & Politics, AP Edition (2022), pages 124–187 — Topic 2.3 (congressional behavior) through Topic 2.8 (the judiciary). Edit that list first if the source changes.

Question ids are not contiguous. They are identity keys, and the gaps are where Topics 2.1 and 2.2 used to sit; reusing a number would collide with the recently-seen list a returning player already has stored, so keep counting up.

Prefix & Suffix — append an object to affixBank in affix-data.js.

{
  id: "af116", affix: "ante", type: "prefix",
  questionType: "meaning", difficulty: "easy",
  question: "What does `ante-` most closely mean?",
  options: ["before", "after", "against", "around"],
  answer: "before",
  explanation: "`ante-` means \"before\" ..."
}

• type is prefix, suffix, or mixed, and drives the Focus filter.
• difficulty is easy, medium, or hard; a session ramps easy → hard.
• affix must match an entry in affixVocab (use mixed for review items).
• Text in backticks renders as inline code. Question data holds no HTML.

The vocabulary in affixVocab is generated from prefix_suffix_vocabulary.csv; edit the CSV first if the word list changes.

After editing any bank:

node tools/validate-content.js

(and after touching match.js, node tools/test-match.js.)

It checks that every question has exactly four distinct options, that the answer is among them, that ids and question stems are unique, and that every affix or key term referenced actually exists — then prints coverage by difficulty, question type, affix and institution. It also warns if any institution and difficulty combination has no questions, since that filter pairing would hand the setup screen an empty pool.