Bump.
Tactics Trainer: Partial solutions skew rating
Sort:
I have to agree with you, on this problem.
Almost everybody will find the 1st 4 moves, without having a clue about the surprise ending, and end up with a positive score. (myself included!)
I have wondered about this as well, I have seen 6 or 7 % solve rates for 1800 problems, and wondered why they were not rated higher. But you hit the nail on the head, I think.
Sorry, I'm going to bump this on more time.
In the meantime the rating has gone up to almost 2100. The pass rate is still low. Not sure if the system works after all or a change has been made to how these TT ratings work. It's been 4 years after all...
I just had this problem, and the problem rating was 1050, so not sure how it was 2100 when you took it unless they manually reset the rating and then the partial answers brought it back down.
Sadly, I got the first 4 moves, but missed the 5th... and like HiggsBoson said, I didn't even know that I was trying to go for a stalemate! So, yeah.. I guess not ready for it. (Though now I know to think about that!)
Your answer to this post triggered a notification for me. You are right. The rating is 1050 again. Not sure how it was much higher 15 months ago. I might have been mistaken!?
I don't understand what the problem is. Why would any of this skew any rating?
This puzzle does in fact seem exceptionally difficult for its ~1000 rating.
Saratonga, you aren't wrong (from comments):
-
SCORE: 100%
12 months ago · Quote · #73
revisit 02.07.14 S
Genial puzzle. Rated 1863 as of 07/02/2014 and 9.7% pass rate.
- 11 months ago · Quote · #74
wow I'm in the 9.9%
- 11 months ago · Quote · #75
revisit 14.08.14 S
Rating varies quite a bit, now 2058 as of 08/14/14 some 200 points higher than a month ago.
- 10 months ago · Quote · #76
Problem rating 2066. Pass rate 10%. Funny.
well it is easy because there are no other moves to try :P
Game_of_Pawns wrote:
I don't understand what the problem is. Why would any of this skew any rating?
The first couple of moves are intuitive, but people fail on the last move - which is the point of the puzzle. Due to Chess.com's tactics rating system, they get a score of something like ~70-80% and get rating while the tactic loses it. If one would have to deduce the solution from the beginning or had to solve it fully to get a positive score (such as on Chesstempo), its rating would most like rise for hundreds of rating points.
how do they fail on the last move? do they just do a random king move? (ke5)? i dont understand how people play that
skotheim2 wrote:
how do they fail on the last move? do they just do a random king move? (ke5)? i dont understand how people play that
"When you don't know what to play, play the move that seems best, even if you don't see the continuation" - Rule 1 of Patzer Tactics Solving Club.
TACTICS TRAINER RESET PEOPLE! UNLIMITED PUZZLES DAILY! DUH.
NewArdweaden wrote:
The first couple of moves are intuitive, but people fail on the last move - which is the point of the puzzle. Due to Chess.com's tactics rating system, they get a score of something like ~70-80% and get rating while the tactic loses it. If one would have to deduce the solution from the beginning or had to solve it fully to get a positive score (such as on Chesstempo), its rating would most like rise for hundreds of rating points.
How does that skew any ratings? The problem is rated as it should be. Also, I just looked at the problem... How the hell does anybody get that wrong??? There are two legal moves, one is clearly lost and one is clearly drawn. How is anybody getting that wrong?
Yea, I found the last move. I didn't get why it was the right move, I didn't even see the stalemate (Even after reading a comment below that said "Beautiful stalemate" I STILL didn't see how it was a stalemate) but I saw that the only possible way to make progress was by pushing the pawn forward.
It seems more likely to me that people are getting the previous move wrong.
letsgohome wrote:
No sh*t sherlock
Are you talking to yourself or...?
Take tactic #0044551 for an example. There is no way that is a 1200 rated problem. The 10% pass rate will attest to that. My theory is that partially correct solutions affect the rating of the problem while the high failure rate assures that players not ready for the problem are the ones getting it.
I'm not sure what the fix would be, but I have encountered this more than a few times.
Edit: To clarify my thesis: partially correct solutions reduce or eliminate the change in rating even though the players rarely get the problem right, so the rating never moves higher.