Guess the Output

A live, host-led multiplayer game where players read a short Python snippet and type what they think it prints. Think Kahoot / Jackbox, but for Python. Built for a school STEM council, grades

• Host Runs the game
• Players join from their phones or laptops on the same Wi-Fi.
• Up to ~50 players per room.

---

The two modes

You choose the mode on the Host screen before the game starts.

Mode	Best for	Rounds	Default timer
Quick opener	A 5-minute warm-up at the start of class	5 (fixed)	45 seconds
Full session	A longer activity; you pick the length	You choose (1 up to the number of snippets available)	60 seconds

Other host options:

• Answer mode — Type it (players type the exact output) or Multiple choice (players tap one of four colour-coded options: the real output plus three plausible wrong answers, shuffled). Both are graded and speed-scored the same way.
• Team mode — players join as teams instead of individuals (they enter a team name instead of a nickname). Off by default.
• Content (tier) — All, Core only (pure Python, no imports; friendlier for younger or newer students), or Library only (snippets that use an import, for an older crew).
• Difficulty — Any, or restrict to Easy, Medium, or Tricky.
• Topic — target a single concept (e.g. strings, dicts, functions, itertools) or leave it on All topics. The dropdown is built live from the bank, and the setup screen shows how many snippets match your current filters.
• Timer (Full session) — set the per-question countdown. Allowed range is 10–300 seconds.

The snippet bank holds 175 snippets across two tiers — core (no imports, any grade) and library (uses an import like random, numpy, collections, itertools, …, aimed at an older crew) — tagged across 22 topics and three difficulty levels. Every snippet's output was produced by actually running it in Python (see Verifying content below), not typed from memory. Your filter choices combine, the deck is shuffled, and rounds are capped to however many snippets match — so no two games play the same.

---

How scoring works

• A correct answer is worth between 500 and 1000 points. The faster you lock it in, the closer to 1000 you get — points scale with how much time was left on the clock when you answered.
• A wrong answer (or no answer) is worth 0 for that round.
• The leaderboard between rounds shows everyone's running total and animates rank changes so you can see who's climbing.

How answers are graded

Grading is forgiving about formatting but strict about the actual result:

• Whitespace doesn't matter. Leading, trailing, and internal spaces/newlines are ignored, so [1, 2, 3] and [1,2,3] both count.
• Quote style doesn't matter. Single quotes, double quotes, and smart quotes are all treated the same — type 'a', "a", or ‘a’ and they're equivalent.
• Error snippets are lenient. Some snippets intentionally crash. For those you just need to name the exception — your answer counts as long as it contains the exception name (case-insensitive). So IndexError, index error... IndexError, and it throws an IndexError all pass.

There is no partial credit and no fuzzy matching beyond the rules above. An answer is either right or wrong.

---

Reconnecting

If a player's phone drops Wi-Fi or they accidentally close the tab, they can rejoin the same room code with the same nickname and their seat and score come back — they won't be duplicated or reset. The host screen can likewise recover its session if the host machine's browser is reloaded.

---

How to add a snippet

All game content lives in one data file: c:\Projects\Untitled\src\snippets.js.

It's a plain list. To add a question, append one object to the SNIPPETS array — nothing else needs to change:

{
  id: 176,                         // any unique number
  tier: 'core',                    // 'core' (no imports) | 'library' (uses an import)
  difficulty: 'easy',              // 'easy' | 'medium' | 'tricky'
  topic: 'numbers',                // one concept tag (lists, strings, numbers, dicts, functions, ...)
  is_error: false,                 // true if the snippet is meant to crash
  code: `print(2 ** 10)`,          // the Python the players see
  output: `1024`,                  // the EXACT printed text (the ground truth)
  explanation: `** is exponent...` // the "aha" shown on the reveal screen
}

A few rules so things stay correct:

• output must be the exact text the code prints. For multi-line output, join the lines with \n inside the string. Keep outputs deterministic — avoid anything that changes run-to-run or machine-to-machine (printing sets of strings, object addresses, unseeded random, numpy .dtype).
• For an error snippet, set is_error: true and put the exception name (e.g. IndexError) in output — that's what lenient grading matches against.
• Pick tier: 'core' if the snippet uses no imports; use 'library' if it does. topic shows up in the host's topic filter automatically — reuse an existing tag where you can.

After editing, verify your output is real (requires Python on PATH):

npm run verify

This runs every snippet through a fresh Python interpreter under several hash seeds and fails loudly if any stored output doesn't match real execution or isn't deterministic. Then restart the server (Ctrl+C, then npm start) to pick up the change.

---

Verifying & testing

Outputs are ground truth, so the project ships its own checks:

• npm run verify — re-runs all snippets in real Python and confirms every stored output matches and is deterministic (tools/verify_snippets.py + tools/verify_all.js).
• npm run smoke — boots the server and drives a full game over real sockets (create → join → answer → score → reconnect → podium).
• npm run filters — checks the tier/difficulty/topic filters, /api/meta, and filtered room creation.
• npm test — runs the smoke + filter suites together.