different rating for different failures

[uri blass]

The current rating system encourage users to guess a solution even if they do not solve the problem instead of playing a quiet move that they consider as best based on their calculations.

I think that better rating system should reduce more points to users that choose a losing move relative to users that choose a move that leads to equal position.

[andreacoda]

I think points have to be taken in context. Number one, they are merely an indication of your improvement (or lack of). Number two, if the objective is “to find the best move”, then there is no grey area there – either we get the best move, hence we solved the puzzle and got full credit for it, or we don’t get it, therefore no credit.
My 2 cents, of course!
Cheers,
Andrea

[drahacikfm]

Uri,

Sorry, I don't understand any of your points.  I don't see why the rating system encourages guessing.  I never guess on Chess Tempo.  I think until I find a winning move, even if it takes me 20 minutes.  I use Standard, not Blitz.  There's no need to be under time pressure in my opinion, when practicing calculation.

I don't understand your comment about reducing points for users that choose a losing move.  That's what is done now already.  And your comment about equal positions, I don't understand.  You pick the best move, or another winning move which gives you another chance to find the best move, or you lose rating points.  That's the way it should be when solving problems.  Maybe you are looking for something different than what Chess Tempo is offering, and that is fine, but I don't see anything wrong with Chess Tempo.  Your points didn't make much sense to me.

[uri blass]

In a practical game there is clearly difference between choosing losing move and choosing drawing move even if there is a winning move.

This is clearly not the case in chess tempo.
Even in standard mode there may be cases when the player may find no winning move even after thinking for 20 minutes.

He can choose between playing the best move that he found that is obvious not a forced win and gambling about some sacrifice that may work inspite of the fact that he did not see a forced win.

gambling is clearly better for rating because the sacrifice may work for reason that he does not see when it is obvious that some quiet move that is the choice in practical game is not going to help.

I like the site to train my tactics for practical OTB chess.
In practical chess(90+30 time control) I have time limit and I am also punished for mistakes and even if there is a winning move then there is still a difference between missing the win by a drawing move and losing the game.

Uri

[drahacikfm]

What other tactics site takes extra points away for a losing move, compared to a drawing move, when there is a winning move in the position?  I don't think any.  Not CTS.  On CTS if there is a winning move, and you play a drawing move, you lose points.  And you lose the same points as for a losing move.

[uri blass]

What other tactics site takes extra points away for a losing move, compared to a drawing move, when there is a winning move in the position?  I don't think any.  Not CTS.  On CTS if there is a winning move, and you play a drawing move, you lose points.  And you lose the same points as for a losing move.

I did not say that there is another site that does it
but only that it is better to do it.

This is a disadvantage of all sites that I know.
 
Uri

[richard]

I think there is some merit to punishing for "bad" moves more than "even" moves, however there are some implementation level details that make this tricky, the biggest one being that I need to evaluate not just the 4 best moves, but potentially ALL the moves in a position (well actually this depends on how many moves are close to even, I could stop processing when I got to the first "losing" move).  So given the significant changes required, I have to weigh up how much extra training value would be provided by the changes, so this would likely be a longer term task.

I'd imagine Revenant would be pretty happy with this change as it would also make it fairly easy to implement his "rating per move" idea as I'd be having to make rating change decisions on a per move basis :-)

Regards,
Richard.

[tomohawk]

Sadly, the problem with any chess tactics book or website is that you know something is going to happen, and analyse accordingly. It is why when CTS puts in the occasional red herring (say a simple capture or recapture) people get mad ("What a waste of time!", "Not a tactic!" etc.). However, I personally think that if you added another 10% or so of the problems with NO solution (i.e. made it so that every reasonable move scored points, but that all of the tricky sacrifices didn't work), that would tend to catch up those folks who do indeed do as Uri is suggesting and play the most active move (e.g. piece sac) even though they didn't really analyse it. It would duplicate game conditions a bit more closely.

Two observations about tactics:

1) Part of the skill of a tactician is being able to recognize when there is no tactic (yet) and you need to build your position up more.

2) Being guided that "hey, there is a tactic here!" is a HUGE help to the player as it allows them to forgo much of the preliminary "soft" analysis that goes into finding moves. You also could spend as much time as you like finding the solution because you know that there is a solution. In a game, obviously, that would be a foolish (though understandable in some positions) road to take since you cannot know for certain that your very promising position had a knockout blow.

Not a complaint about the site, because hey, as far as I know every site and book does the same.

[uri blass]

I agree with the last post and it seems that the simplest way to improve the site is to add positions when there is no tactics when the user needs to play some quiet move or maybe even better simply to say that there is no tactics and get points for it.

It will cause users who really find nothing not to guess some move to get points because they may think that maybe there is really nothing in the position and they can get points for mentioning that there is nothing.

If you add problems that there is no tactics then
I think that it may be also good not to give rating points to problems when there is nothing but only to reduce rating points from people who "solve" them
and give rating points to users who say that there is no solution.

I think that human-human games may be a good source for problems when there is nothing and you may take positions when humans blundered by making some sacrifice that does not work when practically there is no decisive tactics.

Uri

[revenant]

Tactics books and sites are like those automatic tennis ball or baseball throwing machines that the serious competitive players use for training.  You can tune them to hurl the ball at different speeds or in slightly different directions toward the batter who is practicing at home plate or the tennis player who wants to learn how to return a serve.  They don't duplicate actual game conditions but then again they're not meant to.  In any sport there is a need for an intensive, systematic way of developing one's reflexes, focus, strength, stamina and coordination.  Machines provide this functionality.

Chess Tempo can be likened to a sort of advanced, $3,000 pitching machine that knows just how to vary the pitches to test your limits and hopefully, help you push those limits continually back.  And we get it for next to nothing!  I didn't just make up that dollar figure, by the way... Google "baseball machine" and see
http://www.sportstutorinc.com/baseball-softball/triple_play_bb/
for an example.  Richard, is that you peeking out from behind the device in the photo?  :-)

I think having CT insert tacticsless "red herrings" into the mix would be like having a pitching machine that sometimes hurls a rotten peach from its bin instead of a baseball.  You hit it as hard as you can, thinking it's solid, but it just goes SPLAT and then you and your uniform are covered in pulpy bits of fruit.  (At least, that's what happened to poor kid they were playing a joke on in Little League once in "America's Funniest Home Videos".)  Some players might find this useful once they get beyond a certain skill level, but even the proponents would probably agree that it's best left as a checkable Preferences dialog option rather than the default behavior.  By the same token, some players might want the board display occasionally turned upside-down so they can enter the mindset of "What tactics does my *opponent* have?"  (Hey... a new creeping feature!)  But that too should be a selectable option and not the norm.

A steady stream of problems with tactics we know are there beforehand is not only fun but may be the best way to develop the specific mental "muscle" that solves them.  It's certainly true that you don't want to depend on that one muscle for an entire game.  Nevertheless, doesn't combinational expertise usually trump positional expertise over the board?  Like many readers, I admire the works of Reti, Nimzovich, Capablanca and Hans Kmoch for their intuitiveness and simplicity.  But personally I feel I have gained much more from "Chess Combination as a Fine Art" by Werner Golz, or "Blunders and Brilliancies" by Ian Mullen and Moe Moss.  (Out of print, but worth their weight in gold if you can find a used copy!)

[boromir]

I think the rotten peach analogy is faulty. In my opinion it would be more correct saying that the pitching machine occasionally throws a ball (as in not a strike). Part of being a good hitter is knowing when not to swing, just as part of being a good tactician is recognizing when the potential tactics don't work.
Ultimately the goal of this site is to be the best training tool possible, and to this end I think that the more variety there is in the problems the better. I'm talking defensive problems, endgame problems, pawn-winning tactics, drawing problem, and, yes, even the (I admit) slightly controversial 'no solution problems'.
Anything that gets the thought process closer to real-game situations is in my opinion good, and having a certain percentage of problems (wouldn't even have to be much. 5-10% would be more than enough to put that little uncertainty in the back of your mind) where the solution is to click that little 'no tactics' button to the right of the board would certainly get closer to real life where you can never be sure that there is a winning tactics, even in the most overwhelming situations.
Spotting your opponents brilliant defence is just as valuable as finding a brilliant sacrifice.

As the site is now, I think it's great, but quite limited by the fact that you: a. know there is a game winning tactics, and b. know the threshold for the winning move is +2.75 (or something like that), i.e. roughly a minor piece.
There has been many problems where I would never have spotted the solution if the position had come up in a real game, in fact many of those problems I wouldn't even have considered that there could have been a game winning move if I hadn't been told.
There is a certain pattern to the problems (of course not all the problem, but many of them have common traits), and once you've 'cracked' that pattern many of the problems, which in a real game would've been very difficult, actually become relatively easy. I'm not a particularly good tactician, but I am very good at analysing data and seeing common patterns, which I think is why my rating is where it is. And, while being in the CT top-10 gives me some weird sort of personal satisfaction, I don't think recognising a typical 'CT solution' does a whole lot of good for my actual playing strength (although it probably doesn't hurt either).

If you're a paying member you can always have your 'slugfest' by making up your own problem set to your own exact specifications (i'm sure Richard could put in options to search for 'only defensive problems' or so).

[drahacikfm]

Anything that gets the thought process closer to real-game situations is in my opinion good...

On the other hand, if the site gets too perfect and becomes exactly like a real game, you might as well just go play a real game instead   As revenant points out, it's sometimes good to train certain things, rather than always be in a real-game situation.

As the site is now, I think it's great, but quite limited by the fact that you: a. know there is a game winning tactics, and b. know the threshold for the winning move is +2.75 (or something like that)

I think it's now only +1.80, which is much better than a few months ago when it was +2.75.  Now you only need to be up two pawns, or the exchange, or a piece for a pawn, etc.  It's much more common in real games to have a tactic that gets you +1.80, than +2.75 or more.

[tama]

honestly the fact that you know there is a game winning tactic only changes your mindset when looking at the positions, in a real game if you miss a tactic its because you lack experience.

[uri blass]

I think the rotten peach analogy is faulty. In my opinion it would be more correct saying that the pitching machine occasionally throws a ball (as in not a strike). Part of being a good hitter is knowing when not to swing, just as part of being a good tactician is recognizing when the potential tactics don't work.
Ultimately the goal of this site is to be the best training tool possible, and to this end I think that the more variety there is in the problems the better. I'm talking defensive problems, endgame problems, pawn-winning tactics, drawing problem, and, yes, even the (I admit) slightly controversial 'no solution problems'.
Anything that gets the thought process closer to real-game situations is in my opinion good, and having a certain percentage of problems (wouldn't even have to be much. 5-10% would be more than enough to put that little uncertainty in the back of your mind) where the solution is to click that little 'no tactics' button to the right of the board would certainly get closer to real life where you can never be sure that there is a winning tactics, even in the most overwhelming situations.
Spotting your opponents brilliant defence is just as valuable as finding a brilliant sacrifice.

As the site is now, I think it's great, but quite limited by the fact that you: a. know there is a game winning tactics, and b. know the threshold for the winning move is +2.75 (or something like that), i.e. roughly a minor piece.
There has been many problems where I would never have spotted the solution if the position had come up in a real game, in fact many of those problems I wouldn't even have considered that there could have been a game winning move if I hadn't been told.
There is a certain pattern to the problems (of course not all the problem, but many of them have common traits), and once you've 'cracked' that pattern many of the problems, which in a real game would've been very difficult, actually become relatively easy. I'm not a particularly good tactician, but I am very good at analysing data and seeing common patterns, which I think is why my rating is where it is. And, while being in the CT top-10 gives me some weird sort of personal satisfaction, I don't think recognising a typical 'CT solution' does a whole lot of good for my actual playing strength (although it probably doesn't hurt either).

If you're a paying member you can always have your 'slugfest' by making up your own problem set to your own exact specifications (i'm sure Richard could put in options to search for 'only defensive problems' or so).

Hi boromir,
If you train only in standard mode then we know nothing about your tactical ability because you can get higher rating if you invest more time.

I solved 2 more problems at standard to get to the top 20 at the time of this post and I am going to try at least another problem but probably most of my time is going to be at blitz in the second site and I think that getting place 30 at blitz is clearly harder than getting place 10 at standard(I clearly got better place at standard inspite of the fact that I used blitz for most of the problems).

I think that the only good comparison between humans is comparison when there is time limit(like blitz mode).

I wonder if you tried the second site when you need to use even less time

http://chess.emrald.net/

I had to work hard to get to the top 300 active users there so
I guess that the level of tacticians there is higher.

Maybe it can help more to your playing strength because I think that in games you need to detect tactics fast because you cannot search all the time for the possibility that maybe there is some tactics.

I also wonder what is your playing stength in OTB games.

Uri

[tomohawk]

On the other hand, if the site gets too perfect and becomes exactly like a real game, you might as well just go play a real game instead   As revenant points out, it's sometimes good to train certain things, rather than always be in a real-game situation.

Really? Wouldn't it be better to have problems that are of the type most like those faced with in a game? The site is already limited in that it (very understandably) can only provide tactical problems. That's very useful, but  I fear that people who primarily do problems on sites like this are going to find that they rarely get positions which require that they find the only sequence of moves that leads to something tangible. Tactics are maybe the most important part of chess, but finding the absolute best continuation in a position in situations where you are already told that something is there to be found, could lead to lots of frustration when you get into the big, bad world of chess and find out that you can't tell the positions that are fruitful from ones that are just deserts, combination-wise. ;-)

Just one man's opinion ...