Tactics Trainer Problem Difficulty

Oops,the Problem came back and now I'm only getting 900 - 1100 Rated ones only ... juz got 2 Wrong one after another after all correct ones and lost 70 plus points ; all correct ones get 1 or 2 points only .

I think this problem only occurs if your Problem Rating falls to below 1500 . .

Earlier ,when mine went back up to 1700 plus ,it was fine coz I was only getting 1800-1900 plus problems mostly .. only once in a long while there's a 950-1100 plus problem.

I think you are onto something. I agree with the variety factor. The system should emulate real game situations. After all, isn't that purpose of it, or is it just to get you better at solving puzzles ?

In fact, I think the puzzles should be tagged ahead of time in the current format. The reason is, if I have watched a game unfold, I am already aware of my objectives. If I am suddenly staring at a puzzle, that element of the game unfolding disappears. That means I might waste a minute or two looking for a mating combination, when it is a matter of winning an exchange.

I would concede this if they did away with the timer altogether or in an untimed mode. I think in an untimed mode, you should be able to figure it out untagged. I also think that since the average time per move in classical games is roughly 3 minutes per, that is how long you should have. I have noticed in infinite mode at another site, my average solve time is 3 minutes or less. My solve percentage is quite high there also.

I also think that a factor that gets lost in the wash somewhat,  is the way a puzzle gets it's rating. If an engine can evaluate play and determine the strength of play, I think an engine should be used to properly place the rating for a puzzle, so that it can be placed in the right difficulty range ahead of time, instead of new puzzles wrecking or padding ratings until it finally settles where it belongs.

I think the timer factor wasn't considered by you enough. In a confusing position that people have tried multiple times, with many failing before they pass and then realizing the puzzle is simply a hanging piece in the opposite corner of where it looks like there are mulitple possibilities for either a checkmate, mating net or winning an exchange, even perhaps using a sacrifice, I would have saw all of that coming as the game unfolds. I shouldn't fail on time, due to the 9 second solve time, after taking 1 minute to rule out everything else, even if I got it right.  In an untimed mode, I agree, no tags...

Currently,we don't get to see the Rating of the Problem when/before/whilst we r solving it .. so I think it does not have any effect on how we solve it .. it only has an effect on our resulting TT Score ie if the Problem Rating was very much lower than our TT Ratng at the time ,we wld lose a lot of points if we fail it and gain only very few points if we get it correct ,even if we got 100% ( ie managed to solve it very fast ) .. and then our Resulting TT Score will determine the Rating/Difficulty of the next Problem we get ( currently ,I think this is the way it works as it doesn't seem to be Random )

.. So if our Rating has fallen very low ,we will most likely not be able to get much Higher Rated Problems to solve and hence our Rating cannot go up anymore ,unless we do not make any mistakes at all for a very large number of Problems ,consecutively .

Now,my Rating has fallen to around 1300 and I'm Only getting 900-1000 Rated TT Problems to solve ( has been like this for days already ),so I can't get any Higher Rated Problems to solve anymore and my Score also cannot go up anymore ,as the cycle keeps repeating.

U mean do like 50 or more 950-1000 Rated Problems in a row, without making a single mistake ?

( I 'm only getting 950-1000 Rated Problems now,not even a single Higher Rated Problem than my current Rating ! )

I can't ,tho they r very easy coz I will get distracted or something n' make a mistake somewhere along the line n' 1 Mistake is enuf to make set my Rating back where I started .. :(

Kikyo_Sushi. I think that is the problem Erik is trying to solve.

I agree with nameno1had that time is an issue. You lose points for being careful because many of the people who solve the problem correctly make the first decent move they see and if they are right they bring the average time right down, but if they are wrong the timer is not affected. I would like to see an option to play TT in Blitz mode or standard mode, with a different scoring system for each. 

Blitz is an entirely different game. It has a different strategy and different tactics due to the timer. We shouldn't be training for every type of chess with the same trainer. The results my still be helpful to the skill level you play with, but the rating would never be truly indicative of that skill. If the ratings aren't going to be or you would tell me they don't really matter to just blow off my assertion, then let's just remove the ratings altoghether from the tactics trainer ? I bet that would cause an uproar...

... I want them to not only stay but be accurate. The more accurate data for each type of game, for each user, the easier it is to assess skill level, growth, training options and likelyhood of cheating.

nameno1had. I'm confused. Are you agreeing with me or disagreeing? What you say suggests you agree but the tone suggests you disagree.

I was agreeing with you about the time factor. I was merely elaborating a bit more for others who would read our opinions side by side. There are many facets to this issue and I was simply tying a few of them together to share an over all mind set with everyone. I apologize for what appears to be a mixed message. In a since it is, but not in the sense of being contrary toward your particular opinion.

OK. Good. The main issues I can see with TT are: 1) Problems with average time, due to people having different goals (already mentioned). 2) Tactics with "wrong" ratings. Some may say they can't be wrong because they are based on the scores achieved by the people who have taken them, but that's a bit disingenuous because: 2a) Partial scores for getting some moves right but not the whole thing. This makes no sense for standard chess. I've seen some where you get the last move wrong which leads to checkmate against you but get a plus score for getting some moves correct. 2b) Ambiguous tactics. I believe the latest rules for approving tactics are designed to eliminate tactics where e.g. you fail for finding a mate in three because there's a mate in two. But there are still lots of tactics approved under different rules and you can easily lose lots of points this way. There should be an easy way to report these tactics so they can be got rid of. In a way it would be a pity but TT would be the better for it, probably. 2c) The problem Erik is trying to sort out at the moment, which is to do with the exaggerated effect on your rating of failing problems rated a lot below your own TT rating. It's a bit unfair at the moment and the effect is exagerated if you have exhausted all the problems at around about your rating.

But all of these problems go away if you worry less about your rating and more about your pass rate, as benkku52 rightly asserts (IMHO).   

@benkku52. I believe that the current guidelines would stop a tactic early if there was a mate in 3 and a mate in 2. Very often you can tell that a problem is "ambiguous" from the pass rate. If a problem has a pass rate of <20% it isually means that there is a good alternative. e.g. You can win the queen in 2 moves or mate in 4. Both are clear wins but only one is correct. And 80% of the candidates are picking the wrong option. Yes, one line is clearly better but in the context of TT it's not really fair (IMO). I have also done over 12,000 tactics, so I've seen it all, I think. Like you if I get one wrong or even if I get it right but I find it interesting, I will look at the analysis and the previous comments and, if I'm still not clear, run it through Rybka. The previous comments are particularly instructive from the point of view of this discussion because they tell you where people are going wrong. Sometimes (usually) there's lots of people claiming they've been robbed because their solution is better, and they are just plain wrong. But other times the low pass rate and the consensus amongst the comments alerts you that something is really wrong.

But in the end TT, warts and all, has improved my game hugely.

.. but yr Current TT Rating determines the kind of Problems u'll be getting to solve,Next ...

A Few Days ago,I was getting lots of interesting 1800 - 1980 plus Problems to do,one after another and now I'm only getting 900-1000 Problems,one after another .. very boring .

Why not juz make all the Problems we get ,juz Randomly Rated ( Higher of Lower than our current Rating ) so we get to solve all kinds of Problems all the time ? ( so our Rating got chance to go up too )

Btw,imo,Percentage Pass isn't very important either coz u can increase it by getting 100% or almost 100% correct on lots of lower Rated/Easy Problems but it's not much use.

I am still having problems with TT this has been going on for some time and it is infuriating, that nobody will sort a fix out.

Quite simply it jumps back a move after a correct move, it freezes and will not show the solution the clock also stops and NO, it is not my browser I have tried

Google Chrome

Firefox

Internet Explorer

No other application that I use gives me problems just this site.

Yesterday I got 3/3 correct and my score went down on all 3. I was only 15-30 seconds behind the average score on each one...  go figure.

Sceadungen - wrong forum for that problem, really. There must be somewhere where you can report technical issues. My advice (for what it's worth): Try taking screenshots etc. Record exact dates and times. Make a note of the problem numbers where it occurs. Then send it in. 

I hv also experienced this a few times recently.

Came back this morning, still no fix, using firefox froze straight off the bat, useless,.

Resigned my membership

Issue is still going strong. I'd wager this is something hidden in the selection process that was a good idea but takes it too far.

It started for me some weeks ago, when I decided to do some drunk puzzle solving. Usually I do ok, but on this night I tanked 200 points from 2800, and then started getting puzzles rated 2200 at most. I continued tanking down to 2200. I tried recovering from this rating, but was getting problems in the 1800-2000 range where I would win +1 and lose -36. I dipped below 2000, until slowly climbing to 2150, where it took 20 straight correct to reach 2250, and suddenly I had broken free of the algorithm to start getting 2400 problems, then as I got to 2500 I started getting 2900 rated problems.

2900 are a bit too hard for me, so my percentage was quite low, but I had enough success and partial success that I got back to 2700. At that point I was still getting most problems over 2900, and had a stretch where I failed 18 of 20, which knocked me down to 2500, and suddenly I was getting problems rated 2000 again.

Now I am breaking even at 80% success rate, with a personal average under half the time the problems use for average solve time.

Here is 23/25 correct to gain 50 points, averaging under half the average solve time, not counting the 7 second vs 5 minutes (I recognized the puzzle, took a minute the first time):

All that progress could be lost in one or two puzzles! Now, these easier puzzles aren't useless for me, but usually the ones I miss have only two possible options, and I don't want to take the time to figure out which is best because I'll lose rating for getting it correct, so I guess. At least for the puzzles at the rating I'm used to I often use the analysis to fully understand why the best moves are really the best.

Update, I took some extra time on a few and came up with 48/50, with my fails as 3/4 moves and 4/8 moves. The 3/4 I still gained a point, but at half time my 4/8 still managed to lose me 20 points (As many as the 16 questions were worth around it). I am now at several hundred puzzles at 300 and even 400 points below. I can only imagine how frustrating this would be to someone who could only play 3 per day. I'll admit the challenge of working problems through to conclusion or risk losing the last half hour of effort is somewhat rewarding, but also exhausting. I usually use TT for light fun entertainment.

EDIT: Just got one fully wrong. Lost 41 points, and the progress of the previous 47 minutes