Chess engines will have a Principal Variation (PV) that may be as many as 30 (or more) ply deep. Of course, the PV and the other facets of the engine help to determine the best move at the root position (the first ply).

However, how accurate is the rest of the PV? Some of the PV is most certainly wrong; for example, the very last ply of a depth 30 search will almost certainly change in evaluation if it were instead treated as the root position. Is there an easy way determine how much of the PV is trustworthy, and how trustworthy it is?

  • 1
    How long is a piece of string? – Brian Towers May 10 '19 at 10:10
  • The real question is whether you should use them at all! – David May 10 '19 at 11:13
  • your question is basically how much can you trust an engine. – Matthew Liu May 10 '19 at 19:10

The answer is, of course, it depends:

Sometimes the engine will accurately calculate a very long sequence of forced moves and all of it will be correct (the more powerful the engine, the most likely this is to happen)

Other times, not even the first move will be correct. It may also occur that the computer plays "correctly" but still gives a wrong evaluation (this is often the case in positions with fortresses)

In summary, your question is too broad, as the answer will vary too much depending on what software you are using, on which machine it is running, and most importantly, what position they are evaluating.

Not the answer you're looking for? Browse other questions tagged or ask your own question.