I seem to remember that Erik once mentioned, in a similar discussion, that the time alloted for a problem is based on the average solution time for that problem. So, if a problem allows you 45 seconds - that's because most people have solved it in under 45 seconds.

Having said that, I do have the same difficulty myself. I find that with the clock ticking, I either solve the whole problem in under 30 seconds or else (more often) sit there scratching my head; and then when I see I'm almost out of time, I panic and say "oh no! I have to do something! Sacrifice my queen somewhere! Maybe this will work!" 

"when I see I'm almost out of time, I panic and say "oh no! I have to do something!"

This is simply a failure in logic. You don't lose any more points per second used at the end of the clock than at the beginning. And once the clock gets to the end, your score stays at 20%. So, in fact, the time pressure doesn't get worse at the end it goes away at the end!

I use tactics trainer this way: I do not move until i am either sure it is right [though i am still often wrong :) ] or I am sure it is beyond me - at least, i wont get it in 15 mins. (depends on how tired i am etc also)

You ppl care about your tactics trainer ratings???? Do you use it to learn, or to show off testosterone levels?????

There is one person who uses an engine to do tactics trainer. Said person (untitled of course) has one of the highest tactics trainer ratings. The way this discussion is mostly going, said person is being quite logical!

 I think this is the correct way to use the tactics trainer. Well, almost. Even if it is too tough, use up every second until the clock runs out. It will be similar to a situation where a person decides that s/he has looked at a position long enough.

 I know . It's similar to my approach to multiple choice tests back in the day ("if I can't get it in 1-2 minutes, make an educated guess and move on"). A queen sacrifice seems to be the right first move for about 30% of the tactics trainer problems - so if nothing jumps out at me, find a reasonable spot to sac the queen and a lot of the time I'll at least get the first move right. Needless to say, I prefer books for serious tactics study (personal preference here, nothing else).

>>You ppl care about your tactics trainer ratings????

I don't. For me the tactics trainer rating has a mild amusement value, nothing more.

It's not just sacing the queen that's illogical, it's making any move at the end of the timing window. Once the timer gets low, you could make a move immediately and get 21-25% or you could take as long as you want, double check everything, and get 20%.

This is especially true if the solution is going to be more than 1 move long. If you make the first move at 25%, you probably won't finish the line before it's down to 20% anyhow. So it's  logical at this point to just take your time, double check your analysis, and make the right move and get your 20% rather than rush at the end of the time control and get 0.

I've seen the "I make a move at the end just to beat the timer" idea from a number of people and it's just a misunderstanding of how the timer works.

"I seem to remember that Erik once mentioned, in a similar discussion, that the time alloted for a problem is based on the average solution time for that problem."

That's interesting.  In checking a few problems, it looks like indeed the allotted time is 2x the "Avg. Time" stated in the statistic for the puzzle.    So the midpoint of the allotted time is = statistical average time.  That would explain why the time seems short for me.  I know generally that I'm a clear thinker but slower than average on the uptake ;).

 correct :)

I use tactics trainer as well.  And I have the settings set to "rated (timed)", but I have unchecked the "show clock".

That way, I do get timed puzzles, but I don't have the stress of seeing the timer go down.  I just solve the puzzles, without worrying about the time.  It has helped me a lot.

edit: OK, maybe I don't want to "entirely understand" as I'm sure there's a bunch of statistics involved.  But maybe just in layman's terms how I get + 27 on one problem and -1 on another, both problems solved with just a few seconds left on the clock.
--

I don't think I entirely understand the relationship between time left on the clock and adjustment to my rating. In the image capture below, I had just a few seconds left on the clock, indicated by the short red bar, but it says "score: 58%" and my rating says 875 (+27).  Does that mean my rating gained 27 from solving this problem and is now 875?  On some other problems, if I'm not mistaken, I've solved them with just a few seconds left on the clock and had (-1) next to my rating.  So in this case I gained 27 and in other cases I've lost 1.  I'm probably misunderstanding something.  It's hard being slow on the uptake :).

Also, what's the point of the Problem Rating: 1000 (-27)? Does the problem rating adjust based on how people score on it?  It's starting to look to me like the program continuously adjusts rating of both the player _and_ the problem, based on rating of the player at the time they try the problem and their performance on the problem.  If true then your rating _and_ the rating of the problem is relative to performance of the other players on this site.  

This leads me to the following emotion:

ROOT OUT AND DESTROY ALL TACTICS TRAINER CHEATERS !!!  OK, sorry, I'm calm now :).

This tactics trainer rocks, IMO.  I'm just a little intrigued with how the rating works at this point.  I mostly use it to learn.  But I can feel the competitive itch kicking in, already.

By the way, thanks Loomis, and others, for pointing out that it's good to let the clock run out if you haven't solved the problem yet, rather than throw a guess at it.  Makes sense.

Then if the first who solved the problem thought 10 minutes then most of next players will use this time without hurry - average time will be big (and vice versa). So the knowladge about time limit is a positive feedback that directly influences on objective measurement of "average time".

 Thank you for your thoughts. We will definitely consider adding an un-timed rating for players. 

wormrose, your complains seem to be contradictory. Tactics trainer now has the untimed mode. There you can select the range of puzzle's rating to solve and to do "training". As you mentioned that it is not "Contest" - and there are no logical conflicts that your rating does not change in this mode. But to have adequate rating of problems these problems have to be passed through objective competition against players. Are you interested in precise rating of problems? If yes, then you have to understand that rating of problem that was calculated by untimed mode would be useless exactly for _you_.

I'm also a slow player. That is why I play "turn-based chess" instead of "live". My rating in tacktics trainer jumps from 1500 (even 1200) to 1900 while in untimed mode I sometimes solved even 2400 rated problems. I found that it is more interesting for me to spend more time for more difficult problems and currently I switched to unrated mode. Maybe it is not efficient enough and further I plan to switch back to timed mode. I do not see any problems with missing rating in untimed mode. Even more I think the situation has to stay unchanged to force more people to use timed mode to maintain precision of problems' rating.