Wednesday, August 09, 2006

Chess Tactics Server oddity

I recently came across something a little odd at CTS. There are two problems there that are nearly identical, number 59505 and 06868 have the same starting position:

In one problem white plays Qc8 and in the other white plays Qc2, in each case defending the knight on f5. The solution to both problems is the same! You can take a minute to look at it, but here comes the answer: Qxg3+ Nxg3 Rg5#.

The problem with Qc2 is rated 1605 while the problem with Qc8 is rated 1679. Why is there a 70 point rating difference between essentially the same problem? One explanation is that on Qc8, the solver may immediately be attracted to Rxc8 and take time to realize that is incorrect before finding the correct solution; while on Qc2 the solver will have nothing to look for but checks on the king.

The less satisfying answer is that the rating system is flawed. While I do believe there are some problems at CTS where the rating system is not accurate, I hope it's not flawed here. The potential flaw is that the rating of a problem depends sensitively on which tacticians have attempted it and when. I have my fingers crossed and hope this isn't the case.

Labels: , ,

1 Comments:

At 11:38 AM, Blogger Temposchlucker said...

That kind of oddity you will find quite frequent. CTS claims to have generated the problemset by computer. This shows that that claim might be true to at least a certain degree.

The ratingsystem isn't flawed. The difference of 70 points is purely caused by the automatic taking on c8 by alot of people.

 

Post a Comment

<< Home