Not sure if Help & Support is the correct category, but...

I was thinking that the one problem with the tactics trainer is that (as other people have mentioned) it encourages you to just start calculating without fully assessing the position (because of the time component). This is not a problem in terms of purely recognising attack / mating themes - you know who's better, you just have to work out what the idea to win is.

I was thinking that it would take a couple of small changes to the tactics trainer to make it useful as a generation "calculation trainer". Simply

a) Start the position with a random colour. Not hard to do, a lot of the positions already start with a computer move. This way you dont know who's got the winning continuation (or at least have to do a basic assessment to work it out).

b) Make the person trying to solve the position play both sides. This makes sure that the best defense has been seen when playing the initial move. I know that I am sometimes surprised by the defensive moves when solving problems in tactics trainer (and I'll still get the solution, because I didnt have to find the defensive resource).

I think that this should be a seperate thing to the tactics trainer, and could use the same problem base, but with seperate ratings for each way of solving. I think it would be interesting to see how different ratings were for different positions.

 

On a completely side note, I navigated away from this page before posting, pressed the back button and my text was still here :D Ajax is painful, but has its uses!

wow. amazing idea! but that would be really complex to build. i'll stick that on the pie-in-the-sky to-do list for now. i really like it - but we are focused on explorer and teams for now!

and yes, our site does do some cool things ;)

Shouldnt it be quite similar to the tactics trainer? The code should be pretty similar. I guess the complexity is in setting up the problems with different variations. Surely this is something that you could get a chess engine to do automatically though?

:) if people knew how complex this stuff is they would cry. i cry frequently.

It would bring some other problems though. If you say the user should also find the best defense, then what is the best defense if there are two variations like below:

• One variation loses the queen in one obvious move with a hopelessly lost position for the defender
  
• One variation leads to checkmate in five hard-to-find moves, while if the attacker doesn't see the mate, the defender is left with a good position

A computer would say the first variation is the best, but a human being would pick the second one, since the chances of winning are higher. How would you handle this?

Phobetor wrote:

It would bring some other problems though. If you say the user should also find the best defense, then what is the best defense if there are two variations like below:

One variation loses the queen in one obvious move with a hopelessly lost position for the defender
One variation leads to checkmate in five hard-to-find moves, while if the attacker doesn't see the mate, the defender is left with a good position

A computer would say the first variation is the best, but a human being would pick the second one, since the chances of winning are higher. How would you handle this?

---

That's subjective to some degree (I would argue that something which is teaching you to get better should not be encouraging you to hope that your opponent will not make the best move!).  Either way I'm assuming that the problems (as they currently are with the tactics trainer) lead to some forced, significant, advantage regardless of the defence.