Petroff, Two Positional Courses and an Engine Gap: What ChessBase Sold in August 2026

On 13 September 2026, ChessBase published its August product roundup under the headline "All new releases in August!". Inside: two courses on positional play, one Petroff Defence database update family, three named authors, and not a single engine evaluation anywhere in the announcement. For a publisher whose currency is theoretical authority, shipping a training package without one line of engine assessment is a more revealing detail than the entire course list combined.

I have read hundreds of announcements like this. After forty-four years in the industry, the lesson has not changed: what is sold and what is proven are two different stories. "My question did not sound like a woman's question, but their evasive answer did not sound like a man's answer." This time the evasion sits exactly where it always sits: in the data.

The chess training-content market runs on a monthly rhythm. Every cycle, publishers release a few courses, a few database updates, a few analysis videos, then quietly wait for the renewal figures. The 13 September 2026 announcement is one link in that chain: August brought "a range of new products", in the publisher's own phrasing.

Three names appear: Bahne Fuhrmann, Andrew Martin, Kostya Kavutskiy. Andrew Martin is a long-established ChessBase author, an International Master tied to the opening-product line for decades. Kostya Kavutskiy is a US-based International Master, familiar to audiences as a commentator and online teacher. Bahne Fuhrmann is less known to the general public but sits on the same instructor roster.

The interesting part is the catalogue structure, not the names. The publisher segments content by playing strength, from beginners to club players, and describes its products through the standard pedagogical triad: theoretical explanation, annotated games, practical exercises. That template has been industry standard for more than a decade. It works, and it is also worn out.

One clarification matters here: the announcement carries no tournament results, no active players, no ratings, no sample games. The object under analysis is a content market, not human form. Anyone looking for rankings or historical head-to-head records will find nothing here. "Silent applause can kill emotion, but it cannot kill tactics." There is no applause to measure, only inventory to weigh.

The only technical touchpoint into opening theory is the Petroff Defence database update family. The defence begins 1.e4 e5 2.Nf3 Nf6, takes its name from the Russian master Alexander Petrov (1794–1867), and has survived nearly two centuries on one very specific property: it seeks early symmetry, invites balance, and strips away much of White's initiative without demanding major theoretical risk.

For Black, it is the choice of safety. For White, it is a patience test. At elite level, Petroff games are often judged as low-drama, and precisely for that reason the defence becomes a weapon for players who want a draw without bowing.

An opening database earns its price only when it answers a concrete question: at which move does the old theory die, and which line replaces it. The announcement answers nothing. No engine evaluation, no novelty, no statistical data, no win-rate breakdown by variation. We know an update exists. We do not know what it updates.

Here I have to be blunt. An opening product without a novelty and without engine data is a promise, not evidence. Based on my experience following games over more than three decades, database buyers pay for three things: coverage, freshness, and searchability. The announcement confirms only the third, and only by inference from promotional copy.

In this industry, ACPL — average centipawn loss per move — is the standard measure of move quality. No product in the announcement carries that figure. Without it, a buyer has no way to verify any claim about theoretical accuracy.

The two positional-play courses are the easiest detail to skip and the most worth discussing. In one release cycle, a publisher ships two products on the same theme. There are two readings.

The first is depth: the courses target two different levels, one for beginners building foundations, one for club players refining judgement. The publisher explicitly states the material addresses different playing strengths, so this reading has support.

The second is internal cannibalisation: two products competing for the same buyer, splitting revenue instead of enlarging the pool. For an already crowded catalogue, adding two courses on one theme in a single month looks like gap-filling, not territory-opening.

I lean toward the second reading, and I know I may be wrong.

What matters more is the ratio between content types. The August package contains opening theory (the Petroff update family), calculation, strategic decision-making, and positional play. Three of the four categories are transferable skills, not memorised theory.

The publisher is selling thinking habits more than it is selling moves. That is a significant shift. Fifteen years ago, a chess publisher's value lived in its opening vault: whoever held the most lines won. Once engines became universal and equally strong for everyone, the vault lost its monopoly. What remains sellable is the training process.

This is the two-sided model of the modern training market: one side serves the elite theory race, the other serves the mass skills market. The first group pays not to fall behind. The second pays to improve. Both need serving, but only one generates stable recurring revenue. The monthly release cycle exists because of the second group.

A publisher living on renewals must ship steadily. Monthly cadence pushes volume, and volume pushes back against depth. When you must have something new every thirty days, choosing safe, familiar, easy-to-produce themes is a rational business decision. It is only irrational if you expect theoretical breakthroughs from every launch.

At this point I have to check myself. I have never opened either positional course. I have not run the Petroff update. I read a product announcement and analysed its shell. Critiquing a product through its promotional copy is like grading a game from a scoresheet with no clock.

Fairness also demands this: a press release without engine data does not prove a product is weak. It proves the evidence was not placed where the public can see it. In this industry, silence about numbers is usually a communications choice, not a sign of poor quality. Absence of data is not the same as guilt.

The more serious problem is the structure of the source. ChessBase is simultaneously the publisher, the presenter, and the holder of every image credit in the piece. A self-review can be honest, but it cannot be independent. That is a journalism problem, not a chess problem.

One detail bothered me from the start. The note I received carried the header "Chess historians meet in Copenhagen", while the actual content was the August product roundup. One of the two is wrong, and it reminds me why I always re-check a source before going on air. "I mispronounced a player's name, then read the position correctly when the slow-motion replay appeared." This time I read the content correctly before reading the label correctly.

And if someone genuinely has an event about chess historians in Copenhagen, that is a different story, with a different value, requiring a different source. It is not in the data I hold. "Ten years in the middle of arguments, and the fire lit by a 2026 question still burns." I still ask directly, even when the answer is silence.

On competition in this market, I have no share data to compare. Chessable with its interactive course model, and Lichess's open ecosystem with free databases, create pressure at both ends: one pushes on course value, the other on price. Neither is mentioned in the announcement, and I raise them as industry observation, not as sourced facts.

For buyers, here is the minimum checklist before paying. First, does the product state how many games or lines were updated. Second, is there a last-updated date. Third, do the authors state how lines were selected. Fourth, is there a trial or sample game. Those four questions cost far less than one bad purchase.

My prediction, for everyone to verify: within the next two release cycles, the share of skills content in this publisher's catalogue will rise, and the number of new opening products will fall. If I am right, the chess training market has shifted fully from selling theory to selling method. If I am wrong, I will sit and rewatch the whole catalogue, as I once rewatched footage of thirty-two teams in two weeks.

"Mistakes are not frightening; bending down to pick the tape back up at 2 a.m. is what makes a forecaster."

Anyone buying a positional-play course today is buying a promise about who they will become in a year. The publisher selling that promise has a duty to make it verifiable. If they will not, someone else will, because this market has never lacked people willing to take on the hard part.