Tactics Trainer Problem Difficulty

@TomHaegin Agreed something is wrong with Unrated mode when you enter an explicit range. Looking into it now.

And also agreed that something might be fishy with no defined range too, although that case is more interesting. Right now, it seems to be serving puzzles that hug the user's current rating. Technically, if there's no min or max it could serve puzzles rated 500 or 2200, etc., but would that really be a good experience? I think we need to randomize within a "sane" range based on user rating... like anything +/- 250 (i.e. a 500-point range).

Hi DeepGreene, sorry I'm reading your post only today.

Yeah thank you for looking into this. I'm glad you acknowledge some weirdness.

Just for clarification, now we are talking only about any features / behavior in "unrated" mode because "rated" looks definitely very good with the proposed changes for v3 as outlined earlier in the thread (I was basically happy with the old rated as well, was not upset to lose some points for slowness)

Anyways, about "unrated":

There are two scenarios: User has defined a range, or not. If defined, well, just let TT serve puzzles at random from the available pool of puzzles. Easy.

If no range is defined, yes, it would make sense to somewhat limit the range automatically as you say. However, even total random is in order. Because if the user wanted a range, he can just define it.

What I like about unrated and "total random" is simply that it kinda simulates a live game situation where you may not really know if you are looking at a difficult or easy position. Just work on the position and see what you can come up with.

I agree that "rated" training is probably better to actually train the eye and the various motives gradually at about your ability.

"rated","unrated + random within range" and "unrated - total random (or very wide range)" are just different types of training. It's great that the TT allows for all of them.

So again, thank you for looking into it. It would be cool if you could post some kind of follow-up once the team has looked into it a bit .Thanks once again!

*thumbs-up*

I am a little off topic, but TT may be useful for premium members. For "free" members like me it's not that useful, 3 problems per day is nothing. Therefore I stopped using it and I practice tactics on other websites.

About the main subject here: sure there is a problem, and sure it needs to be fixed.

I notice that negative time penalties are still there for a correct solution.

So no renewal of membership for me.

I wonder too if any version 3 or whatever will ever arrive...

At the same time as Santa Claus

Tactics Trainer is becoming a real pain.

It is almost impossible to progress, I have been up over 2000 now I struggle to get past 1700.

a plus score of +1 is not uncommon for a correct solution 10 seconds or so over standard time, time penalties of -6 and 7 add to the difficulty of progressing.

Problems of great complexity are now being rated  1600 or so.

The result is that you are now endlessly seeing the same problems and overall the training experience is very demotivating.

I would also suggest that you turn off the show timer function.

Of course you can improve your rating by trying to solve every problem carefully! In the process, you'll become more proficient at solving problems and your rating will go up. (This is what a rating should be, a measure of how good you are!) Solving problems for 15 minutes a day, I saw steady progress during the course of two years — and only when I actually took time to make sure I saw the entire line did I improve. Sure, going from 1200 to mid-1900s in two years is not that spectacular, but tactics is not my forte exactly. Which is kind of my point =) 

(But that lower rated problem bug is really frustrating, though. I've had problems after switching between TT here and on my phone. Resetting TT seems to fix the problems though, at least for a while)

TT's behaviour is in some ways just strange and illogical.

E.g. in a 4 move puzzle, you may get the first 3 moves right but miss the last one, you may still get points even though you did not solve the full problem. I think this is not useful.

Likewise it makes me laugh when I lose more points on a problem I have solved correctly but too slow, than i gain points when I solve a problem still in time (say 65%).

For slowness (more than double the avg. time) you should just receive zero points, and for not solving the full problem you should always lose some points.

But it's not a score, it's a rating! Wouldn't you agree that it's likely that a problem that you solve quickly is easier to you than one that takes you longer?

Now, let's say you can solve 100% of the problems rated 1200 within the average time, 100% of the problems rated 1300 within twice the average time, 100% of the problems rated 2000 in 256 times the average time &c. (Of course, the numbers are arbitrary). Now, if you began with a rating of 2000, you would stay on 2000 but you might spend a day or two on every problem. Doesn't it make more sense to lower your rating so that TT can give you problems that are suited to your level? And would you say that a rating of 2000 accurately describes your problem-solving capability?

(Of course, this is example is a bit contrived, but it does illustrate the problem.)

Instead, you could focus on your accuracy (number of correctly solved problems vs. total number of problems attempted). Then, the number of problems you've solved is more of a score than a rating.

The problem of what to do with partially solved problems is more complex. Also, what to do when you see/guess the first move and then need to keep on solving the problem? (I know that another site uses a system where you get penalized for taking a long time after the first move(s). Something like that might be useful here as well.)

I agree somehow with Chrka. However the tactics trainer doesn't support an in my opinion proper thought process. In my rating area (1600) there are e.g. many short problems asking you to sacrifice the queen for a mate in two or three (to give an example) which require a total solving time below 10-20s. There is no time to verify the line and one should be allowed to consider a sacrifice even if you know the pattern because sometimes there is a piece in the way and you need to verify that. I would opt for a reasonable time slot somewhere between 2 and 3 minutes without any punishment for correct solutions. This would support a proper thought process and would be in the range of standard tournament times. Problems which are to hard for me take me much longer than the proposed time.

I think the rating system is fine/normal now. To be a good tactics also means to be able to see quickly tactics, so I think it's acceptable to lose points when you are too slow (I'm slow myself). Besides the rating should only be secondary. Tactics trainer is meant to train tactics, and that it does well. Only thing that is really annoying is this bug that I regularly get, which is after I made a correct move, suddenly I can't play any more moves with the pieces of the colour I was playing, and it also flips to the beginning set-up. So than I have to move the other colour, and it of course says incorrect, after which I get some strange error page, and if I want to go back in my browser to the tactic I was playing, I will get a different tactic, and after solving that one, I'll get flipped back to the original tactic? This particular bug happens like once every 40/50 tactics or so. Very annoying!

My chess mentor rating is at 1600+ but chess mentor is showing me how each piece moves. why not tactics anymore?

@dokter_nee: it is possible that the tactics trainer is ok for you but I think we can agree that the individual performance increases with thinking time. So, if you are interested in blitz this is ok. I'm interested in standard time otb (2 hrs each per 40 moves + 1hr) which means that I should use 3 minutes per move as an average. I don't want to get punished when I have to evaluate the position after I gave a king queen fork if there is a better move than taking the queen and often there is a mate in two.

Chesstempo has two modes for blitz and standard which means only a second table and a button to choose. Which makes sense because those are two different disciplines in chess.

I agree that Tactics Trainer does not encourage a healthy chess thought process if you are interested in standard chess. I have since switched to chesstempo, I'm particularly pleased with the mixed mode there which offers a balance of realistic positions, attack and defensive, and the timing system there offers me a realisitc time frame reflecting the time I could spend on a move OTB. This is because you have between 5-10 minutes per move before you start losing ratings. Also the quality of the games are high. Switching over is a no brainer for me.

thank you for taking the time to comment.

I did also mention cutting out the "show timer" this gives assistance to solvers in regard to the complexity of the problem they are solving.They would not have this in real life situations at the board, for myself I never use it

Also more importantly I think it is being abused. Problems of high complexity are now featuring in a very low rating window, in my opinion these are well beyond the capacity of inexperienced players.

The common denominator in them is a long solving time  shown on the timer, sufficient for an engine to be set up and used to solve the problem.

I find TT a useful tool in my chess development, my comments in this post are to improve a good tool further.

I seem to be having trouble using Google Chrome Browser

The problem resets as soon as I make a move the clock continues to run but the trainer is then uresponsive.

Oh and by the way in case I forgot to mention it before deducting points for correct solutions is demotivating and unfair.

I've also seen this puzzle several times.

Yet you've been using it for years. You know of chesstempo where they don't do the same?

I was using Tactics Trainer despite the same reason Sceadungen got upset because there was a mobile app for Chess.com. Chesstempo didn't have that.

However, I found out last week that finally Chesstempo offers a browser based mobile version (beta) which works fine for me. So I switched back to them to train tactics.

All other stuff works fine on chess.com (online games, chess mentor etc.) so I really can recommend a membership - which I have - but I never got happy with TT.