Quoridor FAQ
These are the questions that actually get asked, answered as directly as we can manage. Where a full explanation lives on another page, the answer here is short and points you at it rather than repeating it badly. If you have never played Quoridor at all, start with The Complete Rules and then How to Play; nothing below assumes you have read either, but everything below makes more sense if you have.
Is TheQuoridor.com free?
Yes. Open the site, pick an opponent, and play — against any bot on the roster, or against a friend sharing one screen — with nothing to pay and nothing to install. Ranked play against other people is free too, once you have an account. Optional paid plans sit on top of that free game rather than in front of it, and they have their own question further down this page.
Do I need an account?
Not to play. As a guest you get the whole bot roster, local pass-and-play, and open access to the leaderboard and the position designer. The leaderboard is a thing you read: it shows you who is playing and where they rank, and the sign-up prompt appears one click further in, when you follow a row through to a player's profile.
The designer is the exception, and it is the one worth knowing about, because it is fully interactive for a guest. Drag the pawns where you want them, drop walls into the slots between the squares, undo and redo until the position is the one you had in mind — and then play that position out against a bot, without an account and without losing the board you just built.
What a guest cannot do is sit down opposite another person over the internet: matchmaking, private rooms and friend challenges all need an identity on both sides of the board, so those are account-only, and so is the rating and the reviewable game history that comes with one. Signing up later is not a fresh start, either — a recent guest session's games can be carried into the new account rather than thrown away, so there is no penalty for trying the site out first.
Does it work on a phone?
Yes. It is a website, so there is no app to download and no store to go through — the same URL works on a phone, a tablet and a desktop, and the layout reflows below a set breakpoint so the sidebars collapse and the board gets the screen. Worth knowing before you commit to a fast game on a small handset: the slots a wall goes into sit between the squares and are inherently smaller targets than the squares are. Nothing stops you playing bullet on a phone, but the bigger screen is a real advantage there.
Can I play against a friend?
Two ways, depending on whether your friend is next to you. On one device, start a pass-and-play game and hand the screen back and forth — no account needed for either of you. Online, create a private room and send the invite: the page gives you a shareable link, a short join code, and a QR code for the person sitting across the café table. Private rooms need accounts on both sides, and the games in them are casual rather than ranked.
How strong is the AI?
The roster is seven opponents: six difficulty tiers built on one Monte Carlo tree search engine, plus a separate, clearly labelled experimental neural engine. Across those six tiers it is one algorithm throughout, and strength is almost entirely a question of how much searching it is allowed before it has to commit (how the engine picks a move). The gentlest tier is built to be beatable by someone who has just finished reading the rules; the strongest is built to be out of reach for typical players.
Bots shows the current roster with the rating each opponent carries. Those numbers are fixed labels for where a tier sits on the ladder, not a live measurement: a bot's rating does not drift up when it beats you or down when it loses, the way yours does. That is deliberate: a ladder that shifted while you climbed it would be useless for measuring the climb.
Which bot should I start with?
Start one rung below where you think you belong and climb. Scout, the gentlest tier on the ladder, is the right first opponent if you are still counting path lengths on your fingers — see Scout, the beginner bot. Once you are beating it regularly, Mason is the next rung up, and Warden is the first tier with the deliberate weakening switched off, which is the point where the ladder stops going easy on you. Above that the tiers stop being about learning and start being about survival. The Bots page lists all of them with their current ratings.
Is the Neural bot the hardest one?
No — it is the weakest on the roster, and it is meant to be. The neural engine is an experiment shipped in public while it is still learning, and its own roster description says plainly that it is no match for the top bots. The hardest opponent is Nemesis at the top of the six-tier ladder, not the one with the interesting name. The neural bot explains what it is and why it is on the roster at all.
Do the easy bots play badly on purpose?
Yes, at the bottom of the ladder, and only there. A wrapper that runs outside the search sometimes discards the move the search chose and substitutes a weaker one. That interference is at its maximum at the gentlest tier, well down by the second, and its move-substitution settings are zero from the third tier upward — Warden and everything above it plays what its search actually produced, every move. Scout, the beginner bot walks through what gets substituted and what that does to the games you will play against it. The practical consequence is worth holding on to when you are judging your own progress: if you lose to Scout, you did not lose to a bot that was trying its hardest — and if you beat Warden, you beat one that was.
Why does a bot think longer on some moves than others?
Only the higher ones do. From the third tier upward a bot resolves its think time per position, sliding between a floor and a ceiling according to how tangled the position looks; below that there is no range at all — the bottom two tiers get one flat allowance and spend the same amount on every move of every game. Magnus, the master bot names what feeds that score and turns it into a triage rule for your own clock. If you are reading the pause itself, read it as a gauge of how undecided the position is rather than of how worried the bot is; and note that a fast game clock can cut any tier's thinking short before it reaches its own limit (why that matters most at the top).
Why did the bot place a wall that seems to help me?
Usually it did not, and the impression comes from judging the wall by where it looks like it is standing rather than by what it does to the two path lengths. A wall that sits nowhere near your pawn can still be the strongest move on the board, and a wall that visibly blocks you can be close to worthless. Count both routes before and after, and the question answers itself: Tempo and the Race covers how to produce the two numbers, and Wall Economy prices what the trade between them is worth.
There is a real exception, though, and it is confined to the two gentlest tiers, where the weakening wrapper can hand you a move the search did not prefer. At the very bottom of the ladder that is not a rare event — a substantial share of Scout's moves are substituted rather than searched — so a wall from Scout that makes no sense may genuinely make no sense. From the third tier up there is no such excuse available, and if the wall looks strange the count is where you find out why.
Can I play a four-player game against bots?
Yes, on one device. Set up a custom game, choose four players, and switch any of the other three seats to a bot — one, two or all three. The first seat stays human on the setup screen, so there is always somebody at the board; the seats you do not give to a bot can be filled by people passing the same screen around. A four-player online room is a different feature and does not take bots at all: those three seats are held open for other people to join, and opening one requires an account.
Be aware that four-player bots run on a different engine from the two-player ones: a heuristic built around the dynamic that dominates a four-way game, where every opponent piles walls onto whoever is currently leading. The two-player search's assumptions about how opponents behave under-predict that, which is why it is not reused there. The experimental neural bot does not support four players at all — a four-player request for it fails with an honest error rather than quietly getting a move from some other engine, so if you want to see that bot play, keep it to two. Two other things differ in a four-player game: there is no clock on it, and each player starts with half the usual wall supply, because four full hands of walls on one board would bury it. The second change is the one that catches people out, and Wall Economy is the page for what a scarcer supply does to the price of a wall.
How is my rating calculated?
It is an Elo-style rating, updated after each rated game from your rating, your opponent's, and the result. Everyone starts from the same default. The rating moves in larger steps while it is still provisional over your first run of games, then settles down, and settles further still once it is both established and high, so a single upset does not swing a rating that took a long climb to earn. Each time-control category carries its own separate rating. The leaderboards are a separate measure: within a category they rank players by points from finished games over the day, the week or all time, and show each player's rating beside the score.
Are guest games and bot games rated?
No, and neither are most online games. Only ranked matchmade two-player games move the ladder — casual matchmaking, friend challenges, private rooms, four-player games and every game against a bot all sit outside it by design. Guest games are outside it too: a guest seat carries no rating for a result to move, even though the session itself is remembered well enough that those games can follow you into an account later. If you want your results to count, use the ranked queue while signed in. Everything outside it is somewhere you can try an idea without paying for it, which is exactly what a bot game is for.
What are bullet, blitz and rapid?
They are the three time-control categories, and they are the only three — there is no fourth, slower bucket. By default bullet gives each player about a minute with no increment, blitz about five minutes with a small increment, and rapid about ten with a slightly larger one; custom games let you set your own clock, and the category a custom game is filed under follows from the clock you picked rather than being chosen separately — shave enough minutes off and the same game stops being rapid and starts being blitz. If you are new, start in rapid. Quoridor punishes a miscount more sharply than most games punish a weak move, and counting fast is a skill you acquire after counting slowly, not instead of it.
What does a paid plan actually add?
Nothing that stops you playing Quoridor. Paid plans add cosmetics and remove advertising — the game itself, the whole bot roster and ranked play are all in the free tier, and nothing a plan adds makes a pawn move further or a wall block harder. Pricing is the authority on what each plan currently includes; we deliberately do not duplicate that list here, because a stale copy of a price list is worse than no copy.
How is my data handled?
The short version: we collect what is needed to run the game and keep the community usable, and not more. Account data is kept while your account is active; game records and chat are kept while they are still useful for the leaderboard, rematches and moderation review. Third-party services are involved — analytics, and Google AdSense, which funds the free game — and personalised advertising runs only if you consent to it, with non-personalised ads shown otherwise. You can change that choice at any time from the "Manage cookie preferences" control on the policy page. Account deletion is requested from your account settings, and depending on where you live you may have further rights — access, correction, export, restriction — which the policy explains how to exercise. Read the full Privacy Policy for the specifics.
How do I get better at Quoridor?
Faster than you would expect, because the game has one central skill and almost everything else follows from it: knowing your shortest path length and your opponent's, and updating both every time a wall goes down. Players who count reliably beat players who eyeball, at every level, and it is a habit rather than a talent. How to Play builds that habit from a first game; Strategy is where it turns into judgement — what a wall is really worth, when to race instead of block, how a pure race resolves, and the shapes that punish a purely reactive game. If a term anywhere on the site was unfamiliar, the Glossary defines it in a line and points at the page that argues it properly.