Are any books worth reading if they weren't vetted by an engine?

So I'm a duffer (ca. 900 in blitz here) but my son is pretty good (over 1400 USCF and rising fast).  I'm trying to get him to study books, because everyone says to do so.  I get a Lev Alburt puzzle book and the third puzzle, my son vehemently disagrees with the second move in the line.  I put it in Stockfish and he's right according to the engine eval--and further down the suggested line is a substantial mistake.  

Next I pull out Manual of Positional Chess by Sakaev and Landa (which was randomly available to me, a 2006 publication).  The first game presented is Tal-Uhlmann Moscow 1971.  I'm not going to get stung again so I run it through the engine to vet it before showing my son, lest books lose all credibility with him.  Here, the losing move identified by the authors is actually fine.  The actual losing moves (11 ...Qxc3 and 12 ...exf5) get no comment.  And the variant explored blunders by missing a hanging piece.

And though this is an opening chapter, the conclusion of the game is offered without comment, but Tal makes an unsound Queen sac and his opponent mistakenly declines it so the game has the W pattern in the eval that is common in games at my level.  IMO, opening chapter or not, this is too much of a blunder (plus six to minus 25) to pass without comment.  In fact four straight moves are gigantic blunders.

It doesn't really matter what sage advice is in this book--if I give it to my son, he'll call BS and perhaps rightfully so.  He's strong-willed, skeptical, independent minded etc.  So any book he trusts is going to have to be right all the time.

So is my sample of two representative of chess books written without the aid of engines?

Is this why so many authors say don't analyze with the aid of an engine?

Nobody got stung by an old classic.

Most good books offer ideas and understanding, not memorized lines. Grandmasters writing books after years of experience offered both. Occasional mistakes do not take away from the overall value.

The fact that your son is bright enough to question what he reads is a good sign.

Many books printed nowadays are quickly written with computer help by lesser players. They may offer reams of variations but offer little in the way of ideas and understanding.

Which puzzle in Alburts book and what does your son think is the correct line?

What engine are you using? 

The Tal game is a steady advantage for white after 11... Qxc3 and 12... exf5 with Stockfish.

I have found the book by two experienced Grandmasters you lambasted. The Complete Manual of Positional Understanding was published in 2017 by New in Chess. I highly doubt it hasn't been checked over thoroughly.

It may be too advanced for your son. I saw comments that it was aimed at players rated 2000 and above.

This is the position you claim to have checked out.

There is no doubt that some Tal sacrifices have been refuted by computers but refuting them over the board is not so easy.

According to a note on chessgames, Uhlmann took an hour and five minutes to make his choice, leaving him almost no time for the rest of the game.

There is a reason this game was chosen for the first game in the first chapter of this book.

It is precisely the kind of position which will catch your son's interest, if he really has any understanding of the game.  

There's plenty of inaccuracies in the analysis of "Chess Master v Chess Amateur" too. I was only about 1700 when I was playing competitively, so I'm not clever enough to see the inaccuracies whilst I'm going through the book. I'm still getting a lot out of it - the fact that there is some computer-style move leading to a win 15 moves down the line that the author didn't spot in some variations does not detract from what the author is trying to teach. In fact, knowing that there may be mistakes helps the reader to question everything.

After going through each game using the book and a board and pieces, I then go through each game with Stockfish - and learn even more

No, they say this because this will interfere with your own analysis. Once you've done your own analysis, switch on the engine to see how accurate your analysis was

You must have entered the Tal-Uhlmann game wrong.  Either that, or the moves are given incorrectly in the book.  Yes, of course 17 Qf8+ is preposterous in the position you give, but, assuming this version of the game is correct, that's not a queen sac because the game actually went 15 . . . Qxc2 16 Bb4 axb5 17 Qf8.  It's not impossible, of course, that something on chessgames.com is wrong, but I'm betting it's the right version in this case.  It seems vanishingly unlikely that Uhlmann blundered his queen for nothing on move 15 and that Tal missed it and instead blundered his own queen for nothing two moves later, which Uhlmann missed.

What do they say about what the losing move is?  Is it 9 . . . Bb4?  Depending on what they say about that move, it could still be educationally helpful to identify it as the wrong idea, even if the engine can technically find some saving lines after it.  The point about learning from books isn't necessarily to get every nuance of every position right, but to identify how, practically, speaking, to think about a position and a strategy.  Here, the point is that the pawn Uhlmann can win with 9 . . . Bb4 isn't worth it, so even if there's technically a save after that move, move 9 is still where the crucial bad decision happened.  Uhlmann should have tried an altogether different plan at that point.

Engines are great at analyzing concrete sequences of moves, but they can't offer any commentary on ideas about a position -- which is a huge part of how humans play chess and utterly central to how we learn to play better.  Even if the book's techincally wrong here, it's still giving a better practical education on this game than an engine could. 

Of course, I can't help you in persuading a stubborn teenager of that.  You're on your own there.

He copied it wrong. The book has 15..Qxc2.

Sorry for entering that move wrong.  However, that was just an aside to the main point of the analysis, which was about the opening.  The authors highlight that 9 ...Bb4 is the losing move but it isn't.  The problem for black, as Rsava notes, is that 11 ...bxc3 and 12 ...exf5 are both bad, but the book presents 12...exf5 as if it is a consequence of 9...Bb4.  It doesn't comment at all on 11...bxc3, which is where things went south, and doesn't note that the losing exf5 was not an inevitability either.  The engine finds three moves that are a pawn better, and they aren't so hard to find for a human either, IMO.  Just hold the tension and attack either of the other bishops.

Now I'm an adult, and I understand in a sophisticated way that mistakes can happen, that the ideas are more important than the specific examples, that some moves or lines are pretty hard to find over the board, etc.

The issue I face in giving something to my 9 year old son is that he will call BS and ignore a book if he finds errors in it that are backed by an engine.  Once again, I'm not sure he's wrong.  

Of course I understand the advice to not consult an engine all the time, because over the board you are on your own.  But if you want to play best moves, it looks like an engine, or at least engine vetted moves, are essential.  

Also, I don't really understand how this book might have been vetted thoroughly by an engine if the linchpin of the very first example is so flawed. 

I was able to find the page of the book online.  It doesn't say that 9...Bb4 was the losing move.  What it says is:

"Even when so far behind in development, Black goes after a pawn, and in the process gives up a key piece, the main defender of his dark squares. Now, 9...a6 was simply essential."

And you know what? Stockfish agrees that Black should have played a7-a6 on move 9.

More to the point though, the analysis is exactly right from the standpoint of practical play.  The opening Tal finds on d6 is precisely because White weakened his dark squares with the ill-advised 9...Bb4 plan.  (That's not a move you play unless you're prepard to trade off that bishop.)

In other words, it's a good teaching moment.  

How many times is your kid going to need to know that in this one out-of-fashion line of the Tarrasch, Black is still okay after 11 . . . h6?  On the other hand, how many times is he going to benefit from looking at the board and saying, "My opponent's dark squares are weak.  What can I do to exploit that?"

Writing off all books because there was a perceived error in two is excessive, in my opinion.

"... Authors and publishers alike should work hard to reduce the number of errors to a minimum, but some will always slip through. This is why reviewers who gleefully point out a single analytical error are being unfair. It is not the fact that there is a single error which is relevant, it is how many there are in the book as a whole. … books containing one wrong position after another are at best tedious and at worst a disgrace. With some authors, finding a mistake is no challenge at all and is actually fairly pointless. It is a different matter with a good author, since such an author will usually choose instructive and interesting positions. If he has made a mistake in the analysis of such a position, then the mistake is likely to be subtle, and understanding it is likely to lead to fresh and useful insights. …" - GM John Nunn (2006)

Fundamentals of Chess, by JR Capablanca 

Virtually all books published after ~2000 are vetted by whatever the strongest engine is at the time.  Note that as engines get stronger, some of the recommendations (usually in the opening) change slightly.

Now, the larger issue you have:  You are sub-1000 strength, and your son, while strong for his age, is still a class C player.  If he studied nothing but books written before 1990, never looked at an engine, and mastered the concepts covered in those books (the key aspect of this statement), he would still end up as a strong GM - at which point he would have already realized the small mistakes in those books and adjusted accordingly.

Annotated game collections from older masters are excellent, even if engines find improvements that GMs did not see at the time.  Endgame books written in the last 50 years are all excellent.  Strategic books written in the last 50 years are largely excellent.  The only books I would recommend sticking to more recently published works are opening books.  The reason is simple:  some lines played now, at all levels, were not common (or in some cases, even considered) 30+ years ago.  That said, opening books are also not going to help your son improve at the moment anyway, so sticking to the other books, it doesn't matter.  For whatever subject he is studying, there are many books you can find.  Check out the reviews for a book you are considering and see what stronger players (preferably master level or higher) say about the book.

Also, keep in mind that tactics books may have multiple winning lines, and sometimes the author will prefer a line that is not the (currently strongest) engine's best pick for practical reasons.  For example, if you can win a queen in 2 moves or checkmate in 15 moves (but you must play 15 very accurate moves), it is better to take the queen unless you can see all the way to mate.  The engine will say "Hey dummy, you missed a mate with these ultra-sharp moves ...", but if you are up a queen for a piece, you are still easily winning!  Who cares if you are not writing chess poetry as long as you win the game?

Don't go after the OP too hard; "does occasionally incorrect mean not useful?" is a very reasonable question. I think the answer is "No," for several reasons.

First, a chess player must think for her or himself at the board; refusing to take books as gospel truth helps you practice those same thinking skills. Asking "why?" helps you grow.

Secondly, some examples the OP is seeing may well be good examples of practical play. It looks like someone made a mistake in the recording of that particular game, but Tal was well known for making practical choices like this; so was Lasker. 

On this subject of practical decision-making, Dan Heisman says that "[the goal in a chess game is] to find the best move possible in a reasonable amount of time. The 'best move possible' is often not 'the best move.' If the goal were to find the best move, then human limitations would often prevent this from happening in any reasonable amount of time (if ever)." [The Improving Chess Thinker, 2nd Ed.]

Chess is a practical game. If the objective is winning the game in front of you, engine perfection is not necessary. What is necessary is playing better than your opponent. To learn to do that, well-written, insightful books can be very useful, even if they have their flaws.

The game you posted is not Tal's game. Your game is different starting on move 15.

I think you're missing the point. The book doesn't say Bb4 lost the game, it says grabbing material during lack of development and weak dark squares is the incorrect idea that lost the game.

Which seems to be vindicated by the engine i.e. even though the engine doesn't mind Bxc3 it doesn't like Qxc3 which is the move that wins the pawn. And in the final sequence black loses due to white coming in on the dark squares, and of course lack of development played a role.

I'm not sure what you mean about 12...exf5 being "not inevitable." I guess you mean black had other options? But the engine says black is lost anyway, so it's nice to show the win after the piece is captured. Grabbing "free" material is typically the line "weak" players are interested in seeing anyway.

Sure it could show how white wins in a dozen different lines, but that's not very instructive. It's better to have a main idea that ties the lesson together.

Well, in the end you can't (or at least shouldn't) force him to use the book, so I don't know what to tell you. Get a different book I guess.

That's not how chess works, even at the highest level. First of all engines make mistakes. You have to have good hardware and know how to use them to minimize this. Secondly engines sometimes play impractical moves. When analyzing with an engine you'll need to explore practical alternatives to make sure you're not going down a line that requires an unreasonable amount of accuracy, or a large disparity in what's required by each player (if your opponent's position is much easier to play, then even if you're objectively ahead your results will be worse in practical play).

Unfortunately new players aren't going to be able to understand this, shrug*

"... I'm not sure what you mean about 12...exf5 being 'not inevitable.' I guess you mean black had other options? But the engine says black is lost anyway, …" - mariners234

Does eks74071 agree with mariners234 that Black was lost after 12 Nf5 ? If so, doesn't that make it incorrect to refer to 12...exf5 as "losing"?