Take BAR Fight into your games and your Discord, and meet the people who support it.

• After the answer is revealed, use thumbs up or thumbs down to rate the question.
• Your vote is saved privately for question improvements.
• You can change it while that answer screen is open.
• Ratings do not change scores and are separate from Report question.
• After reveal, applicable comparison choices also show their costs or measured values so you can compare all answers.

• Use Report question in the Activity, choose a reason and optionally explain the problem.
• The exact question and choices are attached automatically.
• New Activity reports share the question, reason and details in #feedback on the BAR Fight Discord server.
• The form links to that channel before submission.
• Your Discord user, server and channel IDs remain in the private editorial record.
• Reports submitted under the previous private-only notice remain private.
• Delivery retries automatically if Discord is unavailable.
• One report per player per question, up to 20 per hour.
• Reports do not change answers or scores automatically.

• Start /barfight quiz to open a group lobby.
• Choose an optional map, game_mode, or both.
• Without a map, replay rounds can use different maps; each names its own map and format.
• Game modes include 1v1, 2v2 through 8v8, FFA and Team FFA.
• Unit and sound questions remain general knowledge.
• Combinations without enough measured replay data ask you to change the selection.
• Daily challenges still require a map.
• Quiz and duel commands accept questions:3 through questions:50; the default is 5.
• The requested number must fit the available distinct question pool.
• Daily challenges stay at 5 questions.
• The host is joined automatically.
• Questions can take a few seconds to load.
• The 2-minute joining timer begins when the lobby opens.
• Click Join quiz in the channel within 2 minutes to join and open the Activity in one step.
• Clicking again reopens it without joining twice.
• At least two players are needed; otherwise the lobby closes without points.
• Everyone can press Start in the Activity to begin early.
• Anyone viewing the open Activity lobby can use Invite players, including other participants and viewers who have not joined yet.
• Use it to open Discord’s invite dialog: send an invite to friends or another channel, or copy its link.
• Invites require permission to create invites in the host server channel; they do not bypass channel access.
• Invitees must join before the lobby closes.
• This control is unavailable in DMs or a full lobby.
• Each new joiner must also press Start; no one can join after the quiz starts.
• Each round reveals after everyone submits or the answer timer ends.
• Players who answered click Next to continue early, or the review timer advances the quiz automatically.
• Player cards show Answer locked in and Ready for next question, plus a group progress count.
• The host can end the group quiz in the Activity.
• After two minutes, unanswered players receive zero for that round and submitted answers are scored normally.
• Players who missed the answer do not block Next.
• The next round also starts when the review timer ends; the quiz continues.
• Resource and unit-stat choices use tens when the correct answer is already a multiple of ten and at least 100; exact uneven answers keep their precision.
• A numeric round awards one point to the closest player(s); an option round awards one to each correct player.
• Final standings appear in the original channel; quizzes of up to five rounds also include round details.
• Longer quizzes keep the posted summary compact.
• Only players tied for first receive a draw point.

• Sound rounds play a real weapon recording: use the audio controls in the Activity, then choose which of the four units fires it.
• Sounds start at 50% volume; you can adjust the player volume.
• Nothing plays automatically.
• Sounds can be shared by other units outside the choices.
• Starting-position questions show an exact-version map where a verified reference is available, with the named spots highlighted, including positions listed in answer choices.
• Collapse Starting positions to save space.
• Quiz chat opens beside the question on desktop and in a closable drawer on smaller screens.
• Its message input stays visible.
• New messages follow automatically only when you are at the bottom; otherwise a new-message button lets you catch up.
• Collapsed chat shows an unread count.
• Drafts and scroll position stay in place between rounds.
• Quiz chat is visible only to participants in that duel or group quiz.
• It is separate from Discord channel chat.
• Messages are temporary: the latest 50 are kept in bot memory, expire after two hours without a new message, and disappear on restart.
• Daily challenges are solo and have no participant chat.

• Mixed quizzes rotate across combat, economy, unit knowledge, unit and weapon sounds, build shortcuts, build timings, map knowledge and replay patterns.
• Multiplayer map rounds use a marker in the Activity: closest to the fixed target earns one point, and equally close players each score.
• Use Zoom in, Zoom out and Reset view to inspect the map; drag or swipe to pan.
• Reset view preserves your marker.
• Markers and the target appear together after everyone locks in.
• Starting-position rounds use exact-version authored centres and name the side.
• Destruction rounds separately ask for the highest enemy-kill count or highest metal value destroyed between 10:00 and 15:00.
• They need a clear, repeatable hotspot in a prepared published cohort.
• Metal means full recorded unit costs, not HP damage.
• Their reveal shows mean enemy-destroyed units and metal value per game in the grid square containing each pin, including games with zero losses there; these are not HP-damage measurements.
• Starting-position rounds cover unambiguous named 8v8 roles on Supreme Isthmus v2.1, All That Glitters v2.2.3, Rosetta 1.4.4 and Hades Ponds 1.1, including mode-only quizzes.
• Destruction coverage is prepared across supported exact-version map images and game formats, using up to 1,000 recent published games per group.
• A 16 by 16 grid is tried first, then 8 by 8; the question states which grid is used.
• Sparse, tied or single-game-dominated hotspots are skipped.
• OS-filtered rounds count losses belonging to players in the stated starting-OS range; opponents can have any OS.
• Only exactly prepared ranges are used, never an unfiltered substitute.
• If eligible, destruction-map rounds get priority after 15 recent rounds without one, with at most one per quiz.
• These competitive rounds are excluded from solo daily challenges.
• Build-shortcut questions specify default Grid keys on QWERTY, the selected constructor or factory, and the menu state.
• Constructor shortcut alternatives stay within the same build category; the reveal names what every offered shortcut builds.
• Categories without enough distinct alternatives are skipped.
• Building sequences are pressed in order, not simultaneously; custom, Legacy and Grid60 bindings are outside their scope.
• A quiet clock ticks during the final 10 seconds while your answer is still unlocked.
• It stops on submission or when the round ends.
• Use Mute clock sound in the Activity header to silence it; browser audio requires a prior click or key press.
• Fresh questions take priority over category coverage.
• Recently used question types cool down when alternatives exist, and each sound recording is tracked separately.
• Sound rounds include weapon firing and impacts, selection, order acknowledgements and supported unit-action cues across factions and tech levels.
• The question names the sound context.
• They prefer unheard clips and then the least recently played recordings; shared recordings get one place in the pool.
• Sounds remain one category, so a larger library does not increase their category weight.
• Units sharing the played sound are excluded from the other answer choices.
• If the available pool is exhausted, older questions can return.
• Picture-recognition choices use the same broad unit class.
• Interactive group rounds let you reorder unit cards by cost, assemble a force within a metal budget, drag a base-range circle on a map, or estimate the metal in intact wrecks.
• Ordering requires the entire order to be correct.
• The cheapest submitted force that can destroy the target in one simultaneous volley scores; over-budget or insufficient forces score zero.
• Range and wreck estimates use closest wins; ties share the point.
• Reveals show sorted costs, an optimal force and damage assumptions, the correct range, or each wreck value.
• Use arrow buttons instead of dragging on mobile, and sliders or numeric entry for estimates.
• These competitive interactions are excluded from solo daily sets.
• Silhouette rounds show a large, high-contrast outline with text-only choices; the original colour artwork appears after the answer is revealed.
• Constructor build-power comparisons separate T1 and T2, and rotation tracks the units in each comparison.
• Generic and conditional T1 factory-fate questions are excluded pending review.
• Obvious factory-to-unit, wreck-resurrection capability and prerequisite-like milestone-order trivia are excluded.
• Metal-sharing position trivia is excluded because transfers do not reveal player intent.
• Position comparisons must offer distinct named roles, not several slots with the same role label.
• First-ready comparisons exclude prerequisite pairs and widely separated game stages.
• Construction-change rounds compare each selected player's own starts in the two minutes before and after an enemy completion.
• They count full recorded metal costs at start, not spending or enemy builds; reveals show before, after and change for each choice.
• Combat questions hide supplied HP and damage figures; the answer review explains the calculation.
• For a continuous quiz window, choose Join quiz on an open lobby, or Open quiz window on a duel or daily challenge.
• Each participant opens the Activity in the original channel.
• Accept the duel there; answer, wait, review and click Next without closing the window.
• Both duel players must click Next.
• Daily challenges are solo.
• Closing and reopening resumes an active quiz.
• The BAR FIGHT header and a persistent Explore BAR Fight link beside the connection status link to bar-fight.com, and completed quizzes show an Explore the site banner for player comparisons, rankings and replay statistics.
• Links open through Discord in your browser without closing the quiz.
• The quiz window matches the website: cyan highlights show selected answers and players who are ready; amber marks waiting messages.
• Revealed answers use green for correct and red for incorrect, with text labels.
• Numeric duel results label the closest guess instead of calling an inexact answer correct.
• On wide screens, animated red and blue robots cheer you on.
• Use Pause animations to stop them; reduced-motion preferences are respected.
• After the final round, Close quiz returns you to Discord.
• Duel results and daily score summaries appear in the original channel; daily summaries do not reveal the answers.

• The original popup controls remain available.
• Duel: your friend clicks Accept; you click Start quiz.
• Both answer privately, then review both guesses and the correct answer.
• Both click Next question; the first person then uses Open question.
• Discord closes the answer popup on submission; your private card shows waiting, results and Next question.
• Accept, answer and review each allow two minutes.

• Daily questions are shared per UTC day, map and OS range.
• You get one attempt per map/day across all servers and ranges.
• Resume in the original channel with the same range.
• Daily progress is saved; unfinished duels and open quizzes expire on a bot restart.
• The Activity verifies your Discord identity and channel access.
• It does not request your email or message history.

OS bounds are inclusive and use the rating recorded in each replay. Leaving both empty includes all ratings. Filtered quizzes exclude unknown ratings and whole-match loss questions.

For races, accept within 24 hours, then play within seven days. Each account’s first standard human PvP match on that map, lasting at least five minutes, counts even if the target is never completed. Faster completion from game start wins. Replay processing can delay results; check /barfight races.

Search players with at least two characters, or enter a BAR account ID. Select maps and units from the picker. Brackets mark optional fields. Traits, timing and dailyboard default to Supreme Isthmus v2.1.

Quizzes mix openings, first factories, build counts, unit popularity, technology order, production mix, teammate handovers, construction outcomes, timing, economy and losses. Replay questions use at least five measured games, never one selected replay or BAR player. The question names its map, position, format and optional OS range. Available topics depend on the data.

New replay rounds include opening continuations, production changes, T2 constructor supply routes, metal sharing, position energy curves, construction duration, factory outcomes, rebuilding, unit survival, recorded kill relationships and construction changes around enemy technology. These require at least 30 measured games and ten players; unavailable evidence is skipped.

• Unit knowledge rounds include icon recognition, cost, health, theoretical DPS, speed and build-power comparisons, anti-air range and metal-and-energy budget puzzles.
• Highest-stat comparison choices span at most 30% from lowest to highest; build-time and unit-stat numerical choices stay within 30% of the answer where four distinct choices are possible.
• They use pinned game definitions and do not depend on the selected map or OS.
• DPS uses the pinned unit definitions and stated target assumptions; actual battle damage can differ.

• Combat puzzles include hits to destroy a unit, combined volleys, reload timing, remaining health, overkill, cheapest demolition, damage per metal and damage per energy.
• Combat rounds rotate across T1, T2 and T3.
• Comparisons use similar tiers and competitive options; multiweapon units name the one weapon being tested.
• Targets and all answer-option units count towards repeat avoidance.
• These are theoretical calculations, not predictions of real battles.
• Unit pictures appear beside applicable Activity answer choices, except unit-recognition questions.

• Replay puzzles compare destruction counts with metal lost, T1 factory outcomes after early T2, and completed Fusion or Advanced Fusion reactors lost to enemies per player over a full game.
• Build-completion questions focus on higher-end units.
• They require at least 30 measured games and ten players.
• Rare role/unit combinations and obvious targeting trivia are skipped; new lifecycle puzzles appear as preparation finishes.

Most new rounds offer four answer buttons in the Activity: select one, then Lock in answer. Some rounds still accept a number or time. Correct option answers earn one duel point or 100 daily points.

• The bot checks the last 30 days of question topics per channel and participant, clearing older history when a quiz is created.
• It rotates question families and ways of answering, avoids recently used units and questions, and balances broad categories rather than limiting unit-knowledge rounds.
• Small pools may repeat.
• Daily sets rotate against recent daily sets on that map and remain shared for everyone in the same day and OS range.
• Published replay data requires ranked, human-only games lasting at least five minutes with equal-sized opposing teams.
• Games with unit bans, custom unit definitions, gameplay multipliers, altered starting resources, no-rush timers, disabled fog of war, evolving commanders, zombies or assist drones are excluded.
• Extra or Scavenger unit packs, disabled ally assistance or unit sharing, sharing taxes, quick-start bonuses, alternative victory conditions, lava water, base builder turrets, ruins, lootboxes and custom balance patches are also excluded.
• During backward backfill, isolated older imports are held out of all public statistics until the scan reaches their dates.
• Published results update with the daily refresh.
• Replay features refresh after the daily publication; new lifecycle evidence becomes available as preparation completes.
• Existing saved daily questions stay unchanged.