What a Chess Engine Is Actually For

What the number means

An engine evaluation is one thing: the result with best play from both sides, all the way down, assuming nobody makes a mistake.

Every part of that sentence matters. It is a statement about the position under perfect play. It says nothing about the position under your play, or your opponent’s, or about what happens when either of you has four minutes left.

Once that is clear, most of the confusion around engines disappears.

The move you should play is not always the best move

Illustrative, using the Blackburne Shilling where Nxe5 loses a piece to Qg5 and still wins games because the refutation is hard to find.

Two moves can be a tenth of a pawn apart and be completely different propositions in a real game. One leads to a quiet position you understand. The other leads to a sharp position where you must find seven accurate moves and your opponent must find none, because they are the one attacking.

The engine rates those the same. You should not.

White to move. Taking on e5 looks free and loses a piece to 4... Qg5. Nxd4 is correct and is played far less often, which is exactly the gap preparation exploits.

For repertoire building this is the central point. You are not looking for the objectively best line. You are looking for the line that gives you the best chance of a good result, which depends on your understanding, your time control and your opponent. The engine has no information about any of those.

Three questions it cannot answer

Can I play this? A position that needs precise calculation from move twelve is worth less to you than a slightly worse position with a clear plan.

Will my opponent go wrong here? The engine assumes perfect defence. Practical chess is largely about the moves people actually find, which is why openings with a poor evaluation can score very well below master level.

Will I still know this in June? A line of twenty memorised moves is not preparation if you cannot maintain it. That is a question about how big a repertoire you can hold, not about evaluation.

What it is genuinely good for

Checking a move you invented. If you find a plausible reply while building, the engine tells you in seconds whether it drops something. This alone justifies having one open.

Finding the punishment. When your opponent’s most likely move is a mistake, the engine finds the refutation. Preparing that at home is the highest-value use of an engine there is.

Settling tactics. Not plans, and not which opening suits you. Whether a specific sequence wins material.

Using it without being led by the nose

Look at the top three moves, not the top one. If they are within a third of a pawn, the choice is yours on other grounds.

Look at the depth. A shallow evaluation in a sharp position is close to worthless, and evaluations often move as depth increases.

Ask what happens if the opponent goes wrong. The engine’s line assumes they do not. Your preparation should account for the far more likely case.

Do not chase the last tenth. Below master level no game has been decided by the difference between +0.2 and +0.3. Plenty have been decided by somebody playing a line they did not understand because a computer preferred it.

The pairing that works

Engine and database answer the two halves of the question. The database tells you what will happen. The engine tells you what to do about it.

Neither is much use alone. A move that is popular and bad is an opportunity, and you cannot see that with only one of the two. That combination is the subject of the most popular move is often not the best one.

In Chess Prep Pro the engine runs on the same board you build on, beside the game statistics, so the sound check and the frequency check happen in one place.