Quoridor Notes — News and Engine Write-Ups from TheQuoridor.com
Most of what happens to a website happens silently. A page appears, a setting changes, an opponent starts choosing differently, and nobody is told — the site is simply not quite the same one morning as it was the night before. This page is where that stops being true for TheQuoridor.com. When something here changes, or when something gets measured and the result turns out to be worth a player's attention, it gets written down, with the reasoning attached rather than summarised into a version number.
There is one post so far. It is linked near the bottom, and the honest thing to say about a blog with one post is that it is a blog with one post. What comes first is not a table of contents for an archive that does not exist — it is a statement of what belongs here, and of what the written side of this site is trying to build, so you can tell in a minute whether any of it is for you.
The gap this is trying to close
Chess players inherit centuries of annotated games, a name for every pawn structure, and an argument about each one that has been running long enough to have sides. Go players inherit a vocabulary so well developed that parts of it have escaped into ordinary speech. Quoridor players inherit a rulebook.
The game is genuinely deep, and very little of it has been written down for players. Finish your twentieth match, decide you want to stop losing races you were sure you were ahead in, and the material waiting for you is thin: some forum threads, a handful of scattered posts, and a few strong programs whose source you can read but whose reasoning nobody has translated into something you can use at the board. Worse, there is no agreed vocabulary. Five players describe the same structure five different ways, which means the good ideas that do exist cannot accumulate — each one has to be rebuilt from nothing every time somebody wants to use it.
Closing that gap is what the written half of this site is for, and it is why the writing is treated as part of the project rather than as marketing bolted to a game client. The target is that somebody can arrive knowing nothing and leave understanding why a wall placed on move four is usually wasted, without leaving the site and without watching anybody's video. So the pages are built as a ladder, each assuming exactly what the one below it established: the rules as the letter of the law, How to Play as a first game that teaches the shape of a decision rather than its legality, Strategy and the six articles under it as the framework that prices the move-or-wall decision, and the Glossary so that all of them can use the same word for the same thing.
What belongs on this page
What changed, and why it was worth building. Release notes are usually the most boring writing a project produces, because they list what moved without saying what it was for. A note here is meant to be readable by someone who does not care about the internals at all: here is what is new, here is the problem it was meant to solve, here is whether it worked.
What the engine turned out to do. The site plays you against an AI, and a difficulty label is a remarkably poor description of an opponent: being told a bot is "advanced" predicts how often you will lose to it and says nothing about how. An engine has habits — things it reliably sees, things it reliably does not (how this one chooses) — and those habits are worth describing in plain language rather than leaving you to infer them from thirty defeats.
What a measurement actually said. Anything on this site that claims a bot is stronger than another bot, or that a change made the engine better, is standing on a set of games that were played and recorded. When the result of one of those runs is interesting — and particularly when it is inconvenient — it belongs in a post. A note that only ever reports good news is not a note, it is marketing with a date on it.
Notes go up when there is something to report. There is no schedule, and a month with nothing worth saying is a month with no post, which is the correct outcome rather than a failure of discipline.
Why an opponent deserves a page
Underneath all of that sits the move-or-wall decision, taken over and over under changing conditions, and Strategy is where it gets priced.
Playing a computer is one of the better ways to learn that trade, and it is also one of the easier ways to learn nothing at all. Beaten by a black box, you shrug and press rematch. Knowing what the thing across the board is actually doing — where it calculates hardest, what kind of position it treats as easy, which shapes it is genuinely bad at seeing — turns the same loss into a diagnosis. You stop asking "how do I beat a strong bot" and start asking the answerable question, which is "what does this opponent underestimate, and can I build one of those against it."
That is why the opponents on this site have written pages rather than one-line blurbs, and why each page is about a different part of playing well rather than a rewrite of the one above it with bigger adjectives. The roster itself lives on Opponents; the pages are the part that does not change when a rating does.
The loop all of this is meant to feed is the ordinary one: read something, try it, find out whether it worked. Puzzles hand you a position with a single right answer, so you can check your reading without spending a whole game to find out, and Coach Mode puts candidate moves on the board with a plain-English note about what each one does to both players' path lengths. None of that replaces the writing, and the writing does not replace it. The reading is what turns a loss into information.
Dated writing and undated writing
Everything else written on this site is deliberately undated in spirit. The rules will be the rules next year. The beginner's walkthrough will still be the right first game to play. The strategy articles are about a board game rather than about a website, and board games do not ship patches. The glossary defines words that will mean the same thing when the site looks completely different. Those pages are written to be true for a long time, which means they are written to leave out anything that will expire.
A blog post gets to be the opposite. It is allowed to be about a moment — about a decision that was made on a particular Tuesday for reasons that made sense with the information available then, including the reasons that turned out to be wrong. That freedom is the entire justification for maintaining a separate section rather than folding announcements into the existing pages. It also means a post is the wrong place to look for a durable answer: if you want to know how the game works, or how to play it better, the undated pages are the ones that get maintained. If you have a question rather than a curiosity, the FAQ is faster than any of this.
The standard a note has to clear
Three rules, stated here so they can be held against what gets published.
A claim about the engine traces to the engine. Not to a memory of how it used to work, not to a plausible-sounding description of how a program like that would probably behave — to the code as it currently stands, or to a run that was performed and whose logs are kept in the repository beside it. Software that is a few years old accumulates documentation describing versions of itself that no longer exist, and describing a deleted engine as though it were the live one is the easiest mistake in this entire genre.
Numbers that go stale do not get baked into prose. Ratings move, budgets get retuned, rosters grow. A page that says a bot is rated some specific figure is a page that is quietly wrong within a season, so positions get described in words — the gentlest, the strongest, one rung above — and the current figures stay on Opponents, which reads them live. The same principle applies to what a subscription includes: that is on Pricing, which is the page that gets updated when it changes, and nothing here will attempt to summarise it.
Direction beats tabulation. Where a measurement is the point of a post, the post says which way the result went and links to the record, rather than reproducing a table that becomes a fossil the next time the run is repeated.
Posts
Launch notes, August 2026 — the written material that went up this month, including the opponent pages, the glossary and the FAQ, and an account of what happened when the difficulty ladder was measured from one end to the other.
If you got here first
If this is your entry point to the site and you have never played, the order that works is the rules, then a first game, then the strategy series once losing starts to feel like it has a pattern to it. About covers what the site is and how the AI behind it works, which is the context most of these posts assume.