Rating Problems

[tmr]

This was definitely true before the last update since so many easy problems were given out.  For me now my percentage correct (since the change) has dropped to about 50% which is where I think Richard intends it to be.

[richard]

tama: The problems may get easier in that you will be marked wrong less often, but you still have to find the "best" line.  You will not get rewarded for finding alternatives, you will just be given a chance to keep looking for the "best" line.

As others have pointed out, if you continue to improve your rating then you will get exposed to harder problems, now for example your average problem served up is around 1900. If the new problem set means you get less problems wrong then your rating will rise and the problems you receive will get harder.

Regards,
Richard.

[chipschap]

I like how it is now because its tough

I have to agree.  I am not here for an easy workout.  I am here to be challenged and to work hard and learn to improve my chess thinking.   I find the site very challenging for my relatively low skill level, and that is a good thing --- no pain no gain as the expression goes.

[tacto]

There will always be hard problems, your about 1600 standard or so ... I've seen 2400 problems which noone has yet to get right.

Some really tough problems are outthere. There will always be a hard part to chess lol.

[tmr]

What the new problem set has exposed for me is my weakness on problems involving gaining material.  With the last problem set's higher percentage of mate problems I did much better.  I think this is because I've worked mate problems more historically and the fact that there is less chance for ambiguity in mate problems.

I agree the goal here should be to challange the user vs winning a ratings battle.  With this I like to see a wide range of problems and I think Richard has the selection algorithm about right.  I expect though that some users will get discouraged though with only getting 50% of problems correct.  It might be nice to see a poll of users to see the percentages who are more driven by their chance for rating increase vs tactics training.

[richard]

tmr: the new premium features do provide a lot more flexibility with custom unrated problem sets, however

a) they are unrated so the thrill of rating increase is not there (although success rates on the default set of problem sets that will be available to all premium users does provide a point of competition [although currently there is no easy way to compare your success rate on unrated sets to other users (without manually going to their stats page)]).
b) users that are discouraged by 50% success rates are unlikely to enjoy the site enough to sign up.

I think the selection width narrowing for lower rated users discussed in the other thread is definitely worth some further thought.

Regards,
Richard.