When looking at alternative moves or continations, how do you guestimate how the opponent plays?

The only conclusion I can reach here is that you are drunk.

And my conclusion about you is much less generous.

Okay, let's leave it at that. The whole thing is offtopic anyway.

Not only because you're playing a human, more importantly because you are a human yourself. Impractical moves will cause you to make more errors.

When I said the engine will mislead you all the time, I didn't mean every move, I meant at least once for each game you analyze (if you use it to look at every move).

It would be really, really good if you showed at least one example of these mystical moves, and explain why it is impractical or misleading or whatever. I don't think these exist.

Sure, next time I play I'll look at the game and post something.

It's strange you're so sure that nothing is better than following the engine when you must know that professionals don't do that all the time... and it's well known Caruana purposefully choses moves the engine doesn't like while preparing his openings.

What on Earth are you talking about, man? I never said anything about following the engine all the time. I honestly think you are confusing me with someone else from another topic.

Until then here's a well known thing i.e. that engines tend to overvalue the d5-e4 pawn structure here (and in similar situations, not only in this opening)

-

-

In the image below you'll see stockfish recommends d5 for white at depth 30 (bottom of pic).
You'll also see under "visits" d5 is the most often viewed (likely because the engine recommends it)
You'll also see under "Result" that in practice, white scores below 50% so this is not good.
You'll also see under "Games" that more players have chosen Re1, presumably because it scores better... in other words evidence that high rated players purposefully choose to not play d5 in favor of something more practical.

-

-

I never assume the opponent will play along with my plans; I figure they will play whatever is the best move. I look before initiating an attack to see what the best defense will be and would not count on the opponent missing that move. Also, I try to be aware of possible attacks or counterattacks they might make. Unfortunately, I often am surprised by moves I did not consider, but that is part of the learning process!

When you say the engine "recommends" d5, are you kidding? What I see is that there are a lot of reasonable candidate moves are evaluated roughly the same (Be3, Bc2, h3, d5 and others appear in the top 3 at various depths.). The engine rates these as roughly equal to each other. That is probably true, as there are many reasonably looking book moves that all score well.

Where is the misleading part?

In practice d5 scores less than 50%... so black wins more often.

I still don't understand your point. Yeah, engines, especially at low depths are not very good at evaluating openings, except in sharp tactical lines. Everyone knows that. Also, it has absolutely nothing to do with the issue discussed in this thread.

I would phrase it that engine lines are subject to misinterpretation. Saying that it is misleading is similar but does seem somewhat wrong objectively speaking.

I saw a Magnus Team Prep video on YT where all the seconds where going through a plethora of computer lines, their own ideas, recent games, etc... This was followed with a meeting and Magnus asked "Anything good?" and the team basically discussed their findings. Magnus would challenge their finds by throwing moves at them to see how they respond to different ideas. If Magnus liked it, he looked into it more. I felt like that is a great strategy in general for preparation at a high level. Everything else is up to personal styles and the strategy chosen for a particular tournament.

llama: makes a point
you: that's stupid and dumb
llama: it's obvious and well known
you: show me a position
llama: [shows position]
you: that has nothing to do with the topic

Ok man, whatever. This stuff is obvious to most people your rating. If you want me to educate you that badly maybe you should pay a coaching fee.

llama: makes a point that is extremely weird and bizarre
me: that's extremely weird and bizarre
llama: it's obvious and well known
me: show me a position
llama: [shows position that has nothing to do with the topic]
me: that has nothing to do with the topic

😅😅😅

Regarding the "engine" argument on this thread:

Engines (like Stockfish) are brute-force calculators. So the lines they often choose are based on that specific ability.

Humans don't have that ability, so we should not be expected to mimic it. As Jacob Aagaard said in Excelling at Positional Chess - we should not try to be engines, but should strive to "think like a human" - because that's what we are best at.

To use an analogy: imagine you are on a journey to reach a destination, but a rocky mountainside lies before you. You see a path over it that is manageable - a flattened path across the mountain, long and winding.

But the machine beside you suggests an alternative: a precarious, jagged overhang, jutting out over a cliff. It tells you that climbing across this overhang - and dangling from your hands, at several points - is the most efficient way across. The quickest. The "correct" way.

But you know that this way is beyond your human abilities.

Do you take the safer, slower, less "correct" route? Or do you try the machine's suggestion?

(Perhaps not the perfect analogy, but one that encapsulates how I feel about the many convoluted engine lines that I frequently come across in my analyses - and which I frequently ignore ...)

Not sure why there's all this pushback against Llama's point.  He's right -- this is basic and well known in thinking about engines.

He's not saying they're wrong.  He's saying they can be impractical.  In particular, they're poor at directing you to the sorts of answers the OP is looking for.

The question of learning what needs to be calculated in a particular position is a very difficult one.  One thing to keep in mind is that there are some positions where you have to calculate a whole lot of very concrete possibilities, and there are others that don't really reward that sort of calculation, at least to the same extent.  One of the things that comes with practice is learning when you need to spend a lot of time calculating and when you can save your time and energy and just play intuitively.  (Even there, assuming you're in a slow game, you'll want to look at a few potential moves by your opponent, but often you can keep that examination to a short depth.)

On the other hand, when you look at an engine, it will always give you the same number of candidate moves -- three, five, whatever you set.  And it won't tell you which ones are +.25 because nothing much is happening in all reasonable lines, and which ones are +.25 only because you our your opponent has to find six straight "only moves" in order to maintain equality, and even at the end, one side is defending an unpleasant endgame where they have to suffer for ages in hopes of holding a draw.

Engines are great when you want to figure out why some unexpected idea works.  But not so great when you want to understand the basic way of how to play a relatively untactical position, in a way you can replicate in another position later.

Tlay80, I agree with everything you wrote. Except that Llama did not say those things that you attribute to him.

What he actually said were a bunch of completely absurd things. I quote a few:

"The engine will mislead you all the time. "

"Would you rather play a move that works 9 out of 10 times, or 1 out of 10 times? The engine will go for the 1 out of 10 "