How to use engines in an intelligent way

Hi all. Lately i've began to wonder: sure the evaluation engines such as SF run on Arena give you the best possible line for the position. But how do you actually analyze the game in a proper way - how do you select the most interesting lines from a human standpoint, like the famous youtubers like KC do. For example, in a site that i probably cannot nominate here,  for a given position it doesnt just give you the best line but also inferior lines. How do i get my software to do precisely that? I am my self not able to come up with interesting moves and variations. Thank you in adv.

This is the problem with using an engine.  Some people will just try and memorize moves without understanding the "why" behind a move.

well anyways. I figured theres actually a "multiPV" function in Arena.  I discovered it in another thread. You can close the thread guys 

Umm, no. Sometimes not even when you let it think for a very long time. The easiest proof is that engines (like SF) lose games to other engines. It takes some pretty big mistakes to lose a game (usually you can draw if all you've done is make some minor inaccuracies).

Plus there's the practical side.

As white would you rather have:

a +0.3 position that's very dry and equal

a 0.00 position where you'll have to defend for dozens of moves.

Or a -0.3 position that's complex and you have interesting ways to play for a win

Humans are going to choose the -0.3 position even though the engine will rate it lower than many other moves. That's why you have to combine engine suggestions with studying human games.

Good question, I see you already found your answer, too bad others misunderstood your question.

copy - paste this URL

https://www.computerchess.org.uk/ccrl/4040/cgi/engine_details.cgi?print=Details&each_game=1&eng=Stockfish%209%2064-bit%204CPU#Stockfish_9_64-bit_4CPU

first, make sure you reach the proper depth. for reliability a depth of 20 is the absolute minimum. 25 or above is ideal.

1. dont just look at the first suggestion but look at secondary ,tertiary and n-th moves. evaluate how reasonable they are by their eval. if the evals drastically drops after a certain number, you should only consider the ones above.

2. judge the difficulty of the resource the engine is suggesting. maybe move one is 0.5 advantage but requires a line that would not have been capable of playing in an OTB game or would have taken you 30 minutes to see all its variations. i nthat case maybe that simpler 0.35 advantage move two looks sexier.

3. dont be afraid to imput your suggested move but let the engine punish you if it thinks your idea is wrong. Knowing why your idea was inferior is important/ sometimes, especially as a strong player or with good positional intuition, you may actually correct the engines initial evals at lower depths. Also do the same with ideas or resources you thought your opponent had but didnt play.

4. figure out when engines can begin to be misleading in endgame positions. the transition between middlegame and endgame is often fluid so knowing when your engine may not be the best guide is an acquired skill. But be especially on the lookout, when your engine gives an eval that stays suspiciously still even with higher depths. this usually means a draw, as the engine sees no way to win the positions.

5. learn how to use livebook and the cloud engines efficiently. While engines can help these days even with openings. a combination of master level games and their statistics/frequency + engine eval is ideal. i have analyzed certain problematic positions where engines give me a very uncomfortable line with unclear ideas, only for a database to show me a highly intuitive continuation that answered my concern by a master(s) practitioner . game statistics even when an engine's eval says otherwise may also hint at "practical advantages and difficulties of a position". there may be a position for example, an engine says its equal but win percentages for black are suspisciously high relative to win draws. unless the number of sample games is really small, this may hint,that white;s position is tougher to hold on in, or blacks play is very smooth etc.

The cloud can also save you immense trouble and give you second opinion of other engines on positions at high depth. imagine thinking you reached some critical novelty in your pet line at move 16, only to check your stockfish 9 and find komodo 10 at depth 34 thinks that sideline is harmless at 0.2  . you didnt even need to turn on your engine!

You dont need an engine to play this position.  

1.Qd2 is the obvious move as it completes the Opening Principles.  Now if 1...Qb2 2.Nb3 Here its ok to move a piece backwards as it covers the a5-c5 squares. 2...Qa3 3.Qd4 Centralizig the queen, and controlling the dark squares on the queenside.  

Or...1.Qd2 Be7 Normal development, preserves the pawn structure, and prepares to castle. 2.a4 Gaining space, and preventing black from playing an eventual ...b5. 2...0-0 Rfd1 Targeting the d6 pawn.

How did i come to my conclusions?

Half Open Centers can lead to positions of imbalance.

 Half Open Centers are not permanent and can change at any moment.

 Deploy your pieces as actively as possible.

 If you have a d6-e6 pawn structure, you must find a way to free yourself, or you will remain passive, and give your opponent the initiative.

 The side with the advanced central pawn will develop his pieces actively and try and start an attack.

Thanks.  That was a good take. 

can you give an example of how to draw the line where engines are no longer effective- in endgames they are replaced by tablebases for a reason- in what kind of middlegames are they not working well?

EGTBs don't kick in until late endgame.

The endgame evaluation starts to matter a lot in the late middlegame phase when the engine is still eons away from reaching tablebases. For example some piece trades are simply off limits if you want to play for a win. They might result in a rook endgame, or opposite color bishop endgame, or etc that's just draw.

But the engine is also not good in positions with other long term ideas. It will often, for example, rate a space gaining pawn move highly when the pawn moves locks the structure and now it's just equal, and you might check the database and see that indeed no human ever plays d5 as white in that position.

Long term ideas also include attacks. Of course after the attack is close to starting, the engine will realize and give decent evaluations, but if the attack requires a few moves to build it up, the engine is sometimes oblivious, and will even rate the building up moves as bad.

I could give an example of each of these if you want, but anyway, his advice was good. You want to let the engine think a long time, and you want to explore rational moves the engine doesn't like (meaning put them on the board and see how the engine punishes them). You also want to explore the engine suggested moves because sometimes it leads to nothing... and sometimes it leads to stuff, but it's impractical, so as he says, you have to judge whether or not this is a position you could play OTB for a win / draw.

And, I mean... that's all a little advanced.

If you're a new player, just use the engine to find big tactical errors. ez pz.