Are tactics fundamentally flawed?

Tactic trainers work on the basis that someone made a mistake, then... a computer of ridiculous strength plays out the rest of the moves.

It's basically like playing Carlsen who accidentally made one stupid mistake, then continued to play at his incredible level for the rest of the moves.

How much instructional value do these puzzles really have?

Knowing the tactic is what helps you prevent making the mistake in the first place. 

-Most games on ~sub1800 level are decided by tactics, hence the tactical "knowledge" is very important to the initial improvement;
-Tactics do not appear from nothing. Generally (on high level) sufficient positional superiority is converted into material advantage.
-There are defensive tactical puzzles too;
-Tactic puzzles helps one to improve his intuition and calculation.
 

Many of the tactics here were found by real players during real games.  Moreover, you don't have to be a computer of ridiculous strength to accurately solve the tactics presented to you.  In fact, someone 400-600 points above your own rating could probably solve almost all of them at your level with high accuracy.

It'd be nice though if Tactics Trainer didn't punish you for finding an alternate winning line that is less than optimal. 

That's a really good point, though I feel I'm more shocked when they make strange moves that aren't necessarially the best and then I overthink them

I've started to binge on tactics trainer just for fun. But I do disagree with the solution to the odd one here and there. I just play by gut instinct.

I've got proof it is flawed by the one I just did. It was a mate in four tactic, I did it in two and got it wrong!?

Please just trust me. I don't know how to post a tactic so get your board out.

Starting position:

Black

Kg8, Rf8, Ra8

h7, g7, f7, b7, a7

e6

d5

Ne4, Qd4

White:

Qg4

f3

h2, g2, c2, b2, a2

Rf1, Bc1, Ra1

Now black moves king from g1 to h1.

The solution is:

1) Nf2, Kg1

2)Nh3, Kh1

3)Qg1, Rg1

4)Nf2,

My solution was straight away:

1)Qg1, Rg1

2)Nf2

It's exactly the same checkmate! It's flawed!

I'll take your word for it Milton, the only problem I have with the TT here is if you take to long you lose points even for solving it. I just solved one with 15 seconds left on the timer and I got minus 1.... that's bs man, I want my point back lol.

I solved this tactic instantaneously - bang, bang. Four moves, I did it in two and it's exactly the same outcome. Just set the tactic up as I've mentioned and see for yourself.

There's wrong solutions to some of the tactics out there.

I think having computer playing defence has been of good intructional value to me. Not only did I learn to look longer for possible defences for my opponent (and hence avoiding hasty unsound sacs), but also how to mount a tougher defence (or even refute) against my opponents tactics. The latter is also a tactical way to win games!

They are going to change that. Slow should not lose points.

Personally I like chesstempo standard tactics. The total solving time is irrelevant, take a few hours if you need. What matter there is the time between the first and last move -- so you better calculate your sac all the way before you move! (also they accept all winning lines)

Being good at tactics is very important.

I set up the position and hayabusahayate16 is right.  Kxg1 prevents mate.

You are probably correct though in the general observation that some of the tactics here have errors in their solutions.  There have been a number of times that my son has complained about the solutions here and on chesskid.com, which the chess.com folks run.  He was so adamant about one of them that I ran it through stockfish to prove him wrong.  I insisted that they must have run all of their tactics through an engine so he should stop whining.  As it turned out, stockfish favored my son's move by almost a full queen!

Here is the position in which my son found an improvement over chesskid.com.  It's the last move in what was a 6-8 move combination. 2800-level puzzle on chesskid.com, so probably a 2400-level puzzle here.

White to move.  Can you guess the move the site liked?  How about the move that my son and Stockfish preferred?

6 to 8 moves proceeded the position that is shown.  This is just the final move in a complicated combination.  Unfortunately, I don't have a record of the initial position.  My son and Stockfish liked Bd6 immediately here.  The site wanted Be5 as you guessed.  

It was probably wrong to say a 2800-level puzzle on chesskid.com is akin to a 2400-level puzzle here.  A 2800 *rating* on chesskid.com is akin to a 2400 *rating* here.  However, a typical 2800-rated chesskid.com puzzle seems to be a lot TOUGHER than a typical 2400-rated chess.com puzzle .  Because there are no time limits on chesskid.com, they apparently made sure the tactics there will take additional time to solve.  The toughest puzzles here have time limits of 5-10 minutes.  On chesskid.com, my son often struggles for 10-20 minutes on an individual tactic.

I just found the original puzzle by digging through my son's solving history on chesskid.com.  It was actually a 2923-rated puzzle.  The move my son disputed was not actually the very last move of the solution, but it was a number of moves in.  

White to move.  I'll post the full chesskid.com solution later.

You already know an intermediate position you're trying to get to.  If this were a blank slate, perhaps it would be harder for you.

By the way, 5 moves proceed the position.  Though one pair of moves is not that germane.

Incidentally, Bd6 is not chesskid.com's follow-up to Be5 f6, so you would have been scored incorrect.

Perhaps this tactic is rated so high is because the solution is wrong and, hence, been flubbed by everyone?  Ratings depend on the time taken, the percentage of people who get it correct, and the ratings of those who get it correct and wrong.

Ok, what's the solution?  I wouldn't even know where to begin

Yes, tactics trainer is for people who don't know or understand much about chess. 

The fact that experienced and decent players can routinely lose to someone doing de la Maza style calculating is the huge downfall and great flaw of chess. What's the point of appreciating the finer points when someone can just do that?!