Grievances with Tactics Trainer

I've been doing a lot of tactics trainer recently, and my rating has been fluctuating from 1800-2000, so don't think that these complaints are just sour grapes because I'm terrible at TT.

Tactics trainer is amusing, you can spend a lot of time doing the puzzles there. However, it seems to me that there are a lot of problems with the puzzles and the program.

I don't think that TT actually trains tactics, I think that it trains you to become better at tactics trainer. Here is why I think that.

The time limit: The time limit in TT often forces you to make your move in under 30 seconds, this leaves you with minimal time for actually analyzing the board and the continuation of your move. Too many times I have simply seen a check, made the move and been correct. The fact is that most of the time if there is a check available then it will be the correct move (at least for problems under a 2100 rating). This is great if you want to improve your tactics trainer rating and become better at TT, but it doesn't really help you learn to spot real tactics in real games. I find the time limit to be totally unnecessary.

The tactics themselves: Too many forced mates following a similar pattern. There are not enough real game tactics. I would like to see more puzzles where the end result is the gain of a pawn, or something similar.

Inconsistent problems: The problems in tactics trainer are usually based on a combination to either win material, or to checkmate your opponent. However sometimes there is a problem where no combination is necessary (you even see them above 2000), you generally have about 15-30 seconds to complete these problems. These problems usually involve a minor piece or rook being simply hung to a pawn. It's as if tactics trainer is trying to throw you off. Since you don't have time to analyze the board you might simply throw a piece at a  tactical theme (potential fork, check, etc) and of course you get the problem wrong. Well that's alright, getting problems wrong is how we learn right? But what exactly was learned? Nothing, this is not a tactic, and you would not have missed it in a real game, with time to analyze the position. 

Multiple solutions to a problem: Sometimes it happens in tactics trainer that after making your correct move there appear to be two valid continuations. Do I fork the queen or king, or do I go for a maybe forced mate down the line? Either way I have completed a tactic and seemingly won the game. Sometimes the alternative solution is simply bait, and these problems are usually easy to spot (a queen hanging to a bishop on the first move, but a mate in two nearby), but sometimes it's a bit different. In general multiple tactical themes in a single problem should be avoided if the play doesn't have time to fully analyze the position (again it comes back to the time limit).

Endgame problems: I find it hard to believe that anyone is going to invent the Luceana position while doing tactics trainer, the player either knows the position and gets the problem right, or he doesn't and gets it wrong, not really a tactic. Some endgame problems are great, but ones which rely on systems that have to have been learned earlier are not so great.

Tactics trainer is fun, but it would be nice if it was really a good resource for training tactics, as it is the computer workout and chess mentor are much, much better.

I completely agree on the time limit. I wish you could turn it off. I often take the check or take the hanging piece just because of the time limit, whereas I would train myself to look for the "best" move if there was no time limit.

You can turn it off for unrated mode, but that's no fun.

1. Nobody, I mean nobody pushes you to make a move you are not sure about just because some time elapses. You complain about the solving time in TT? Check out the solving time on CTS, there you must find a solution within several seconds. The justification for CTS is to prevent abusing with engine usage.

2. The tactics are extracted from real games. There are forced mates, but also lot of other tactics, like winning a piece. 

3. About problems with a free piece. What is their purpose: to develop your board vision. Again, nobody forces you to move if you don't know what the problem is about.

4. TT is limited in scope due to its nature. It cannot be compared with Chess Mentor, they are simply different tools.

5. I agree that in the endgame you must know things like opposition, Philidor and others. Your objection is however not valid: if you don't know these motives beforehand, it is likely you will not be able to invent them during an OTB game. Moreover, knowing them enables you to develop a correct plan even early in the middlegame. You are a pawn down in a rook ending, you must try to set up Philidor, for instance.

1. What does CTS have to do with anything?

2. Just because they are extracted from real games does not mean that they are going to commonly occur in your games. I don't generally have a problem with the puzzles where the object is to win a piece though.

3. Balderdash, you don't develop anything with those problems.

4. True enough, but it could at least be a real training tool for real tactical situations that will arise in your real games.

5. Obviously knowing the luceana or philidor is good for your chess playing, but it shouldn't be included in tactical puzzles. It's like including a math problem that requires the use of the Pythagoras therorem in an IQ test, without explaining what the pythagoras thereom is.

I share some of this frustration, I must admit, but I usually come out blaming myself for my errors. 

I've seen the 'red herring' problem phenomenon lots of times:  Hey, there's a tasty little Knight fork...  probably a trap.  Or is it?  :-)  But I usually decide that it's just TT's way of driving home the point that when you've found a good move, you should look for a better one.

And yes, time is of the essence.  I've rolled my eyes when I work out what I think is a tricky little sequence, only to lose 20 points for thinking 45 seconds.  But to be fair, a lot of problems in there give you far more than 30 seconds to figure it out.

The ones that make me say "get real" are the ones where you've got a couple of good options for a one-move problem and you've got to consult coach Rybka on the analysis page to figure out why, six moves later, the winning move is winning.  But again, maybe my standards are too low.

Ironically, I *like* the ones where it's a simple "Just take the Queen dummy!" problem, because I think they keep us honest.  They are sort of like anti-tactics or anti-puzzles; they make sure we get a comprehensive sense of reality on the board before we start looking for 'puzzle moves.'

Cheers

It's improved what I see on my real game boards significantly.  I have learned to reason and look for answers on the board when I don't see the final conclusion.  That said, my rating has never gone over 1635 on TT.  How much time have you spent on it?

Yeah, I think the "just take my queen dummy" problems are useful, as they keep you honest. If they weren't there, you would never take the Queen, and keep looking for some complicated mating pattern.

I think TT is a valuable tool all in all, I just don't like the time limits. I am never that under the gun time wise in a real game.

I admit I play the 'ratings game' on TT to some extent, as it's good for on-again/off-again bragging rights around my office.  But I try to take it easy and lately I've been aiming more for accuracy than speed.  I'm almost up to a 60% success rate, and I try to focus on that when I'm PO'ed because my rating is momentarily in the toilet.  :)

Also, I would agree that my OTB play has gotten sharper since I started dumping hours (well... days) into TT.

If some tactics do not occur in your games this is because you or your opponent did not know to create and execute tactical shots. As for me, I can say that about 90% of my games (live or correspondence) are decided due to tactical errors. Also, the large majority of GM games are also decided with tactical shots.

thanks for your thoughts. here is my feedback:

there shouldn't be problems with multiple solutions. there should be forced clear wins in all tactics. but it is this same logic that bites the other way - winning a pawn can't be a tactic because some people might prefer to build an attack. we have to set the limit somewhere, so we use the limit that the outcome must be 2.5 points higher than the next best option (or checkmates must be 2 moves faster - like mate in 2 vs mate in 4). it isn't a PERFECT solution, but it's pretty good and keeps things clear. of course we try to manually perfect the tactics, but it isn't so easy. tactics that simply win a pawn are unfortunately too difficult to judge for the most part.

as for the time - those times are set by other players. if everyone else is solving them faster than you, then frankly you are doing tactics above your level in this system. if you haven't looked, here is how we calculate ratings: http://www.chess.com/tactics/help.html#rating - you would have to be like 1.5 standard deviations slower than average to lose points on a tactic.

and rated tactics with no timer doesn't make any sense. i know we've debated this already in other forums, but it's impossible to have rated tactics without the timer (both for cheating reasons and the fact that people's ratings would be more indicative of their level of patience than their ability to solve tactics).

multiple solutions is for chess mentor. honestly, it would be incredibly complex to do that for tactics trainer.

personally i find tactics trainer to be incredibly powerful at tuning MY ability to spot tactics. if it isn't working for you, then i am sorry.

A bit more than 11 hours.

Looks like I am about 70 rating lower than you in live chess, am I to understand that the ability to create and execute tactical shots is developed exclusively from 1630-1700? (disregard my embarrassing CC rating please :P). Tactics do decide my games, but generally not 6 move forced mates starting with an obvious queen sacrifice.

No, as I stated, I think TT is a great tool, and one of the reasons I paid to join this site (and yes, I know there are free tactics elsewhere, I wanted everyhting in one place).

And who cares if someone cheats on tactics trainer? They are only hurting themselves and wasting their time.

I didn't realize that the time limits were set by the users. I guess my main gripe with the time limits is that generally, my problems come from playing too fast, so I generally am trying to slow myself down, so being on a short clock doesn't help that.

That's a good point about pawn capture problems, but I was really only using that as an example, I would just like to see some more realistic tactics problems (yeh I know they come from real games, but I think that you know what I mean).

I did not realise that was how the time was set. I generally don't have a problem with the timer myself, but the fact that it is there is often enough to make me (and I believe others) make a move in haste upon noticing a tactical theme.

Who cares about cheating in tactics trainer? What does it matter if someone cheats to get all the puzzles right? Unless you are offering prize money to the top 5# tactics trainer players, I don't see how this could be a problem.

rookandladder.  you have the option of hiding the timer, but your time will still affect your rating like everyone else.    

I've noticed a lot of people having issues with the timer.  For me, it makes me sit upright and focus on the board more.

 I also find that there are plenty of problems to throw you off if you're going to just move the first eye-catching piece conflict on the board.  I've really learned to stare at a board and go "no no...it's not that...hmm..." and that's when you're learning to think.  When you solve AND understand a 4 move problem at 94% - that's when you see the fruits of your learning.  

I wonder if a simple option to either hide the timer, or to have it count up instead of down might alleviate any psychological impact that you feel it has.  This way Tactics Trainer would still be measuring the time, but you wouldn't see how you were performing against the norm until after you'd completed the tactic.

That being said, in an OTB game your clock is very visible and counts down towards an eventual flag.  Maybe learning to play with the psychological pressures that come from having a clock in front of you is actually an ancillary benefit that can be obtained from the Tactics Trainer in its current state....

It raises what are, for me anyway, some interesting 'philosophical' questions about how to use the tool.  If I play for ratings (i.e. trying to 'beat the clock') then I go with my instinct and 'guess' at the moves.  Playing that way, I can sometimes push my rating to the extremes (highs AND lows).

But I sometimes wonder if this isn't a waste of time.  Am I, by this practice, training my instincts at some subtle level?

When all that gets too frustrating, I'll decide to ignore the clock and go for pure accuracy.  Is the (less volatile) ratings range that results from getting more problems correct (more slowly) my "real" rating?  One thing I've noticed is that a lot of the folks in the upper echelons of TT ratings do not have enviable success percentages, which I find interesting.

Sidebar:  Is using the visible timer "cheating" - in that it provides an index of the complexity of the problem?  (Or is that offset by the stress it causes? lol.)

i find most of the tactics to be realistic. i dunno what to say about that...

yep. times come from average solve times.

you'd know if you had to deal with support issues :) the other reason is that we are not only rating the players, but the problems! and if the players are not rated consistently (based on cheating or patience), then the problems will not be either - ratings will be out of whack and you will see really hard problems with 1200 ratings.

Kupov: the large majority of TT problems are not 6 moves forced mate starting with a queen sacrifice. In fact, such problems are quite rare, it is very difficult to come up with such a sequence of moves without any other good alternative. 

On the other hand, more "realistic" problems can not be included because of the requirement of having an unique solution. This explains why the number of endgame problems is so small in TT.