Why I hate Tactics Trainer
Sort:
[COMMENT DELETED]
Windows-7_ wrote:
Tactics Trainer is great, but it needs a LOT of fixing.
we have done some brainstorming about it, and some improvements will be coming; however, i mostly disagree with your points:
-Saying "Incorrect move-good try!" is very discouraging. What I find bad is the "good try!" part of it.
'good try' is friendly enough. don't be so discouraged! for one thing, from the rest of your post, i glean that you have too much of a 'competitive' take on tactics trainer. this thing is for getting better. not for improving your tactics trainer rating, or for feeling good about yourself! it's also secondarily for seeing some awesome and gratifying tactics; and can be used to test yourself. but honestly, whatever language we choose, someone could tell us it's not the best word choice. but i don't see any compelling reason to change this.
-Forcing all moves right for a pass. This is not rational at all-it should be around 60%.
you should find all moves correct in advance before you even make the first move. failing this, you don't pass. but don't take it as such bad news!
-Allow multiple RATED tries on each problem.
this would skew ratings, because obviously on a second try after seeing the solution, people would get the problem right. what's the point of this anyway? this option is just to increase your tactics trainer rating, which is an absolutely pointless goal. especially when it's done not even by getting tactics right!
-The countdown graph that shows time is distracting.
yeah, i agree. so use the unrated setting that doesn't have a countdown. this is something we have thought about, and an extra mode with respect to time/no time is a possible development in the future.
-Problem ratings are not rational.
um, yes they are. there are a few which somehow are out of whack, but bc of the dynamic system whereby problems are rated based on whether or not you solve them, over time they tend towards having accurate ratings. that is in fact one of the great strenths of tactics trainer.
-Multiple solutions to each problem are needed. They could just give part credit for the solutions that aren't best-that's the way most trainers go by.
currently if a problem has multiple solutions, we just delete it, or cut down the number of moves back to the first point where there is more than one solution. it would be onerous to reprogram tactics trainer to have multiple solutions, so that's just not happening any time in the next 1-2 years.
-Where's the Report Problem link when I need it?
-We should be able to tell Chess.com exactly what's wrong in problem reports.
a form to fill in to tell us what's wrong with a problem instead of just flagging it? maybe... although flagged problems get reviewed by someone carefully, so what kind of specific input are you going to give them that they wouldn't already figure out looking at the problem?
-Please only use truly tactical problems.
if there is a problem which is not tactical, flag it. it'll get deleted. we certainly aim to only have tactical problems, and in my experience (344 attempts, plus unrated mode), i've never seen one that was not a tactic.
-Saying 0% after getting the first move wrong is not rational because you couldn't even try some moves. Better would be to say wrong, but let you try the next move.
sure it is. you're not supposed to "try" the later moves. you are supposed to find ALL the moves right from the very beginning. and again, why do you care if it tells you 0%? so you got a problem wrong. it's ok.
I'm considering boycotting Tactics Trainer for those reasons.
The main issue with the timer is that it gives you a hint to the depth of the problem. If I see the timer ticking down super fast I know not to look beyond simple captures etc.
dpruess wrote:
Windows-7_ wrote:
Tactics Trainer is great, but it needs a LOT of fixing.
we have done some brainstorming about it, and some improvements will be coming; however, i mostly disagree with your points:
-Saying "Incorrect move-good try!" is very discouraging. What I find bad is the "good try!" part of it.
'good try' is friendly enough. don't be so discouraged! for one thing, from the rest of your post, i glean that you have too much of a 'competitive' take on tactics trainer. this thing is for getting better. not for improving your tactics trainer rating, or for feeling good about yourself! it's also secondarily for seeing some awesome and gratifying tactics; and can be used to test yourself. but honestly, whatever language we choose, someone could tell us it's not the best word choice. but i don't see any compelling reason to change this.
At least add a control for us to customize this!
-Forcing all moves right for a pass. This is not rational at all-it should be around 60%.
you should find all moves correct in advance before you even make the first move. failing this, you don't pass. but don't take it as such bad news!
Not agreed-the key in Tactics trainer is that you get the idea. A mouse slip for instance can't fail someone.
-Allow multiple RATED tries on each problem.
this would skew ratings, because obviously on a second try after seeing the solution, people would get the problem right. what's the point of this anyway? this option is just to increase your tactics trainer rating, which is an absolutely pointless goal. especially when it's done not even by getting tactics right!
The # of tries could be based on the rating of the problem
-The countdown graph that shows time is distracting.
yeah, i agree. so use the unrated setting that doesn't have a countdown. this is something we have thought about, and an extra mode with respect to time/no time is a possible development in the future.
I then won't be able to gauge my progress.
-Problem ratings are not rational.
um, yes they are. there are a few which somehow are out of whack, but bc of the dynamic system whereby problems are rated based on whether or not you solve them, over time they tend towards having accurate ratings. that is in fact one of the great strenths of tactics trainer.
I can't agree with you, because the low pass rate of many problems tends to skew lots of problem ratings. I would say most of the problems can be gotten by at least an 1800-rated player.
-Multiple solutions to each problem are needed. They could just give part credit for the solutions that aren't best-that's the way most trainers go by.
currently if a problem has multiple solutions, we just delete it, or cut down the number of moves back to the first point where there is more than one solution. it would be onerous to reprogram tactics trainer to have multiple solutions, so that's just not happening any time in the next 1-2 years.
Sad. Psuedo-code:
import whatever libraries you need
Set up the variables
loop this until window closed:
-randomly pick problem
-check to see the problem's rating is appropriate
-Display problem. loop the timer. Wait for move
-If move is equal to #1 choice, set Correct=1
-If move is equal to #2 choice, set Correct=2
-If move is equal to #3 choice, set Correct=3
-If move isn't equal to any of those, set Correct=0
-change move # by 1
-If Correct=1, do not change percentage correct, say Correct Move!
-If Correct=2, change percentage by x (depends on problem), say Alternative, Partial Credit
-If Correct=3, change percentage by y (depends on problem), say Alternative, Partial Credit.
-If Correct=0, say Incorrect, decrement percentage correct.
-If Move=Last Move #, load comments page, wait for Start Training.
-Change position (make opponent's move)
-Where's the Report Problem link when I need it?
-We should be able to tell Chess.com exactly what's wrong in problem reports.
a form to fill in to tell us what's wrong with a problem instead of just flagging it? maybe... although flagged problems get reviewed by someone carefully, so what kind of specific input are you going to give them that they wouldn't already figure out looking at the problem?
A form, yes that's what I want, similar to reporting ads.
-Please only use truly tactical problems.
if there is a problem which is not tactical, flag it. it'll get deleted. we certainly aim to only have tactical problems, and in my experience (344 attempts, plus unrated mode), i've never seen one that was not a tactic.
What about those long endgame problems?
-Saying 0% after getting the first move wrong is not rational because you couldn't even try some moves. Better would be to say wrong, but let you try the next move.
sure it is. you're not supposed to "try" the later moves. you are supposed to find ALL the moves right from the very beginning. and again, why do you care if it tells you 0%? so you got a problem wrong. it's ok.
What I'm suggesting is just so that you lose a move, then try another move, and then perhaps 1/3 moves wrong=no pass.
I'm considering boycotting Tactics Trainer for those reasons.
i think the long endgame problems are tactical; the ones i have encountered have been some of the most delightful problems for me.
i think you are right, there may be an issue where the pass rate on the first move skews problem ratings somewhat. but i don't know how many problems suffer from this. i mean, what percent of problems you encounter are mis-rated?
and maybe we can have a rated mode where the time is ticking down but you don't have to see the bar. that might be mildly useful, but i don't know-- how many people want that?
Just because a tactic is long and in the endgame doesn't mean it's not a tactic. EDIT: dpruess beat me to it
Also allowing multiple tries, answer seen or not, would give assistance to the user and mess up problem ratings.
I didn't realize you could get rid of the timer - cool! Also, this may just be in chess mentor, but when it says "Alternate correct move" I wish it would let me go ahead and do what I was going to do and then say, "You could have also accomplished this with _____ ." Just an idea.
dd- chess mentor is also not able to go down multiple branches right now; it would require some extra programming (whose extent i'm not qualified to say). i personally like for authors to make a comment explaining why the alternate move is correct (so that you can compare that to what you had in mind) and tell you what the mainline move is, so you can get on with the mainline without trying to figure out a second correct move. but that depends on the position a bit. sometimes i think it's useful for them to let you try to find the second correct move.
windows-7_ sorry for getting off your topic there.
dpruess wrote:
i think the long endgame problems are tactical; the ones i have encountered have been some of the most delightful problems for me.
i think you are right, there may be an issue where the pass rate on the first move skews problem ratings somewhat. but i don't know how many problems suffer from this. i mean, what percent of problems you encounter are mis-rated?
and maybe we can have a rated mode where the time is ticking down but you don't have to see the bar. that might be mildly useful, but i don't know-- how many people want that?
I can't say a %, but a double bishop sac definitely is not as high rated as a once-in-a-lifetime combo.
Does it matter what, if anything, you sac? Isn't rating intended to show a problem's difficulty?
dpruess - Yeah, I can imagine the programming challenges associated. Anyway, I paid my money and I want it now! 
(just kidding)
P.S. - I've learned a ton from your videos, and I appreciate your time very much.
Edited my large quote to include pseudo-code for tactics trainer and more replies to dpruess' answers.
oinquarki wrote:
Does it matter what, if anything, you sac? Isn't rating intended to show a problem's difficulty?
Yes. But lots of similarly rated problems are very different in actual difficulty-age of problems also is a factor.
i can't read the green writing in the quote -any chance of changing it?
*edit, it's ok, I just highlighted it.
windows i follow your pseudo code, and i can certainly ask our programmers how feasible it is to make such a change at some point, i have just been under the impression from previous conversations that it was not on the horizon yet. you have to keep in mind that we have tons of projects we are working on, and limited manpower.
dd1123-- you had me there for a sec :-P i've gotten so many comments like that hehe. thanks for appreciating!!!
as to difficulty-- note that what is easy to you might be harder to someone else, and what is hard to you is easy to someone else; the system used to rate problems tends, over time, to give them an accurate rating based on a huge amount of data.
dpruess wrote:
windows i follow your pseudo code, and i can certainly ask our programmers how feasible it is to make such a change at some point, i have just been under the impression from previous conversations that it was not on the horizon yet. you have to keep in mind that we have tons of projects we are working on, and limited manpower.
dd1123-- you had me there for a sec :-P i've gotten so many comments like that hehe. thanks for appreciating!!!
as to difficulty-- note that what is easy to you might be harder to someone else, and what is hard to you is easy to someone else; the system used to rate problems tends, over time, to give them an accurate rating based on a huge amount of data.
But still, many of the problems aren't 2000+ rated and neither are any of the users (except for cheaters).
I'm rated 2000+ in Tactics Trainer.
oinquarki wrote:
I'm rated 2000+ in Tactics Trainer.
2400+ once. It's absurd-the problems aren't really rated that high.
In order to truly solve a tactical puzzle, you have to calculate all the moves without moving the pieces. After all, this is the way you have to do it in an OTB game where you don't have an Analysis Board and you're opponent won't jump in and say, "Great Rook sac! Now you have a Mate-in-2!"
Tactics Trainer babies you by allowing you to make the first move (of say, a Mate-in-4 puzzle) making it much easier to determine the rest of the moves by letting you know that your move is correct and it gives you a picture of what the board looks like after your move in Analysis Board like fashion.
I've solved many puzzles on Tactics Trainer where I had no clue whether the position leads to a mate or a gain in material, only to see one forcing line, like a check or a piece sac, and I gamble on it, which of course turns out to be right, and then the rest of the moves flow easy based on how the computer reacts, i.e. where it moves the King or how it chooses to retreat the piece.
If you cannot even get the first move correct, you absolutely deserve a 0% Fail for the problem, without question.
Tactics Trainer is great, but it needs a LOT of fixing.
-Saying "Incorrect move-good try!" is very discouraging. What I find bad is the "good try!" part of it.
-Forcing all moves right for a pass. This is not rational at all-it should be around 60%.
-Allow multiple RATED tries on each problem.
-The countdown graph that shows time is distracting.
-Problem ratings are not rational.
-Multiple solutions to each problem are needed. They could just give part credit for the solutions that aren't best-that's the way most trainers go by.
-Where's the Report Problem link when I need it?
-We should be able to tell Chess.com exactly what's wrong in problem reports.
-Please only use truly tactical problems.
-Saying 0% after getting the first move wrong is not rational because you couldn't even try some moves. Better would be to say wrong, but let you try the next move.
I'm considering boycotting Tactics Trainer for those reasons.