Correspondence chess is chess played at long distance and long duration: each move travels to the opponent — historically by postcard, today over a server — and the reply may take days. A single game routinely lasts months, sometimes years, and players are expected to analyze deeply between moves, consulting books, databases, and in the modern ICCF era even engines. It is the laboratory branch of chess: the version of the game where the question is not "what can you find in three minutes?" but "what is actually true about this position?"
What Is Correspondence Chess?
Strip away the clock pressure that defines over-the-board play and a different game emerges. In correspondence chess the standard allocation is on the order of ten moves per fifty days — an average of days per move, with vacation allowances on top. Within that time you may move the pieces around a real board while you analyze, keep written analysis files, consult every opening book and database ever published, and in officially sanctioned modern play, run engines. Nothing is hidden and nothing is rushed; the only resource that remains scarce is judgment.
This changes what a "mistake" means. Over the board, most games are decided by oversights — a tactic missed under time pressure, a miscalculated line. In correspondence chess, elementary oversights vanish almost completely. What remains are errors of evaluation and planning: choosing the wrong pawn structure to head for, misjudging a fortress, underestimating long-term compensation. In other words, correspondence chess isolates precisely the part of chess that is hardest to train any other way — deep, honest analysis with the tactical noise filtered out.
The culture is also distinct. Because games stretch across seasons, correspondence chess has always run on patience and good faith between opponents who may never meet. The world federation's motto — Amici Sumus, "we are friends" — is not decoration; it describes a community where a single event can accompany you through two years of your life.
From Postcards to Servers: A Brief History
Chess by letter is old — merchants and diplomats played moves inside their regular correspondence centuries ago — but organized correspondence chess is a 19th-century invention. The famous early landmarks are club matches between cities, conducted a move at a time by post, with committees debating each reply: London against Edinburgh in the 1820s, and a string of intercity matches across Europe that treated a single game as a civic project. The arrival of cheap postal service and, above all, the postcard turned this from an aristocratic curiosity into a mass hobby: by the early 20th century thousands of ordinary players who lacked access to a strong local club were testing themselves against distant opposition, one card at a time.
The modern structure dates to 1951, when the International Correspondence Chess Federation (ICCF) was founded and recognized as the worldwide governing body for the postal game, running its own title system — correspondence Grandmaster and International Master — and its own world championship. The first world correspondence champion was crowned in 1953, and the championship cycle has run continuously since, producing its own lineage of champions largely unknown to the over-the-board public but revered inside the discipline. A special mention belongs to Hans Berliner, the American who won the 5th World Correspondence Championship in the 1960s with a dominance never repeated, and whose famous deep opening analyses — prepared over months for a single game — anticipated the preparation culture that engines would later bring to all of chess.
The 1990s and 2000s replaced the postcard with the email message and then the webserver, collapsing transmission time to zero and moving the entire game onto platforms where the position, the clock, and the move history live online. The board changed; the essential activity — days of analysis per move — did not.
The Rules: Time, Sources, and Yes, Engines
Three features of the modern rulebook surprise newcomers, so let me state them plainly.
Time is counted in days. A typical ICCF control gives each player fifty days per ten moves, with unused time carrying forward and formal leave periods on top. Reflection time is a real strategic resource: strong correspondence players spend it very unevenly, banking days on forced sequences to spend weeks on a single critical decision. It is time management at a geological pace, but the skill is recognizably the same one you exercise on any clock: identify the moments that deserve the budget.
Books and databases were always legal. From the postal era onward, consulting opening literature and game collections has been part of the game — correspondence chess never pretended to be a memory test. This is why correspondence play has historically been an engine of opening theory: with unlimited access to sources and days to dig, correspondence players systematically pushed known book moves deeper into every important variation, and many a novelty that later appeared on the over-the-board stage was first tested by post.
Engines are permitted in ICCF play. This is the rule that startles people. Once engines became strong and undetectable in the early 2000s, prohibition was unenforceable, and the ICCF chose honesty over pretense: officially sanctioned correspondence chess is now openly centaur chess — human plus machine against human plus machine. Note that this legality is specific to that world: on ordinary online platforms, daily and correspondence-style games permit opening databases but strictly ban engine assistance, and using one there is cheating. Know which rulebook governs the event you are in.
The engine era has had one dramatic statistical consequence: at the highest correspondence level, the draw rate has climbed toward totality. When both sides can verify tactics to near-perfection, decisive games become rare treasures, typically the fruit of one side steering — many moves earlier and by human judgment — toward positions the opponent's analysis pipeline handled poorly. Endgame tablebases add a final layer: certain positions can be claimed as won or drawn by database, closing the game the way a proof closes a theorem.
What Correspondence Chess Teaches About Deep Analysis
Here is why I, an over-the-board grandmaster, send students toward correspondence-style work: it is the only format that forces complete analytical honesty. Over the board you can play a move on intuition and let the clock excuse the gaps. In a correspondence game there is no excuse — you had days — so every move becomes a small research project, and the discipline this instills transfers directly back to practical play.
Three specific habits get built. First, systematic candidate search: with unlimited time, "I looked at two moves and liked one" is indefensible, so you learn to enumerate candidate moves properly before analyzing any of them — the habit whose absence caps most club players' strength. Second, evaluation at the end of lines: correspondence analysis dies or survives by the verdicts you attach to final positions, so you learn to stop calculation at quiet points and commit to an assessment in writing, where later moves in the game will mercilessly grade it. Third, planning in structures rather than moves: when a game lasts a year, you are not choosing moves so much as choosing the pawn structure and piece configuration you will live in — precisely the structural thinking I spent a book teaching, and correspondence players are among the readers who need no convincing that a pawn structure chosen on move 15 decides an endgame on move 55.
There is also a defensive education. Because nothing falls to a cheap tactic, correspondence games are won by accumulating small, permanent advantages and lost by allowing them — which makes the format a masterclass in prophylaxis: the unglamorous art of asking what the opponent's analysis is hoping for and denying it first.
The Engine Era: What Humans Still Add
The obvious question about centaur correspondence chess: if both sides run strong engines, is the human just a courier for the machine's first line? The results say no — decisively. Rating lists and championship crosstables in the engine era still show persistent gaps between correspondence players using comparable hardware. The difference lives in the human layer, and understanding where teaches you something about chess itself.
Humans choose what to analyze. An engine evaluates the position it is shown; the strong correspondence player decides which future positions are worth showing it, steering the game toward structures where engine evaluations are least trustworthy — locked chains, fortress candidates, deep sacrifices for long-term compensation whose horizon exceeds any search depth. Humans manage the analysis tree: knowing when a line is settled and when the engine's confident number deserves suspicion, allocating hundreds of hours of compute across a branching structure. And humans handle opening direction: choosing repertoire lines not for their evaluation — everything playable is near zero — but for the richness of the problems they pose to an opponent who must navigate them with their own imperfect pipeline.
I find this genuinely reassuring. The engine era did not make correspondence chess a typing contest; it promoted the human from calculator to research director. The skills that remain decisive — structural judgment, skepticism toward evaluations, knowing where complexity lives — are exactly the ones worth building.
Should You Play It — and How to Use It for Improvement
For a club player, I recommend correspondence-style chess with enthusiasm, with one condition: play in the engine-free formats — the daily games offered on ordinary platforms, where databases are allowed but engines are banned — because at the improvement stage, you need to be the analyst. Used this way, a handful of slow games running in the background of your life is a training regimen disguised as a hobby: every move is an invitation to analyze a real position deeply, with something at stake, on your own schedule.
To extract the value, keep an analysis file per game. Write down your candidate moves, your main lines, and — critically — your evaluation and your opponent's most annoying resources before you move. When the game ends, that file is a diary of your judgment, and comparing it against a post-mortem is the fastest error-diagnosis tool I know: it shows not just what you missed but what you believed. This is the same discipline behind analyzing your own games well, applied in real time rather than after the fact.
The habits transfer beautifully. Students of mine who take up slow chess for a season come back to over-the-board play with wider candidate searches, more patience in unclear positions, and far more respect for the difference between "the engine says" and "I understand why."
Correspondence Habits at ChessMind AI
You can practice the correspondence player's core disciplines at ChessMind AI without waiting months for a pen pal. Our analysis board is built for the research-project approach: load any position or game, branch your candidate lines, and test your written evaluation against the engine only after you have committed to one — the honest ordering that correspondence work enforces. The deep-preparation habit starts in our opening courses, where you can push your repertoire knowledge past the point where your practical opponents' book runs out, correspondence-style. And because slow chess ultimately funnels into technical positions, our endgame trainer drills the theoretical endings that correspondence players convert with database precision — the destinations your move-15 decisions are quietly choosing between.
Frequently Asked Questions
Are engines legal in correspondence chess?
In ICCF-sanctioned play, yes — engine use has been openly accepted since the 2000s, making official correspondence chess a human-plus-computer discipline. On mainstream online platforms, however, daily correspondence-style games ban engines while allowing opening databases. Always check the rules of the specific event; the two worlds have opposite policies.
How long does a correspondence game take?
Typical modern controls allow around fifty days per ten moves for each player, plus leave, so a serious game commonly runs six months to two years. Server play eliminated postal transmission time, but reflection time — the heart of the format — remains measured in days per move.
What is the ICCF?
The International Correspondence Chess Federation, founded in 1951, is the world governing body for correspondence chess. It runs the world championship cycle, awards its own correspondence Grandmaster and International Master titles, maintains its own rating system, and sets the rules — including the modern acceptance of engine assistance — under its motto "Amici Sumus," we are friends.
Is correspondence chess dead now that engines are so strong?
Changed, not dead. Elite ICCF play has become a high-draw research discipline where human analytical direction still separates the best from the rest, and engine-free daily chess on ordinary platforms is more popular than postal chess ever was. What genuinely ended is the era when correspondence and over-the-board chess tested the same unaided skills.
Is correspondence chess good for improvement?
Excellent — in its engine-free form. Days per move let you practice systematic candidate search, deep calculation, and written evaluation on real games without clock pressure, and the analytical honesty it enforces transfers directly to over-the-board strength. Play a few games at a time and keep an analysis file for each.