Of course, GMs are very strong at tactics. The only piece of benchmark I have is a tactical test published in a French magazine a long time ago, when the author wrote : if you manage to solve all problems (there were one hundred) in less than 20 seconds (each), then you have GM tactical ability.
On a side note, Tactics Trainer gives more points for a correct solution after a long thinking time than an immediate mistake, which seems fair.
GMs are absurdly strong at tactics.
ANY tactical puzzle you see from a GM game where your side has less material? You're only solving half way through the real tactic... the GM had this sequence in mind many moves ago and had been trying to set it up... you're just trying to finish it off. Kind of frustrating when it's a very hard one that you end up giving up on 
At my last tourney the TD knew this IM and set up a tricky mate in 3 position before the round started and bet him 20$ to solve it in 5 minutes... (he gave him way too much time IMO) the IM looked at the board and within 10 seconds he reached out and tapped the square where the first move would be (he'd already seen it) then spent another 20 seconds or so making sure, then played it and took the money :) For reference it had taken an 1800 player almost 10 minutes to find the correct sequence the previous day.
Yes, they're very good.
Also, one beef I have with the Chess.com tactics trainer is that it's timed and you get more points the quicker you answer, which encourages you not to actually figure out the tactic, but instead choose a move that LOOKS like it could be the tactic at first glance without actually calculating it through.
Yes, this is a big complaint of mine too. Other tactics places do the same thing.
Check out Chess Tempo where they have these blitz type problems but also have a "standard" puzzle set/ratings where the only thing that counts is the correct answer, and ignores the time you use... also they have endgame puzzles, all free -- worth checking out.
yeah, try chesstempo. its a great site... on a good day I might be 1750 rated, a bad day 1600s. to constantly solve 1800 rated problems I would have to put in a lot of time on each one, so that under 20 second thing hicetnunc said is amazing.
Tactical puzzles where the GM is down a knight? Look up the game, he didn't find himself down a knight and then luckily 20 moves later he's winning now -- he saced the knight before and you're trying to find the continuation.
You're right about finding tactics they miss -- they're human too.
Of course GM's are extremely strong in tactics. How quickly they can solve difficult (requiring accurate calculation for several moves) problems is what we all want to know. As someone mentioned, the fact that you know there is a combination (if you're solving a puzzle) helps a lot, as one can't afford to take a big look at a position for a long combination every move.
Yes, this is a big complaint of mine too. Other tactics places do the same thing.
Check out Chess Tempo where they have these blitz type problems but also have a "standard" puzzle set/ratings where the only thing that counts is the correct answer, and ignores the time you use... also they have endgame puzzles, all free -- worth checking out.
Sure, so that everybody can take the time, set up the board in Fritz and then "solve" the puzzle (aka press the space bar). Not that this would not have been possible here :) The true tactical ability comes *not* from being able to calculate long lines, but from being able to spot the right move. Yes, looks strange for those who believe things like " if a 1800 ELO player is given enough time, as in CC, then he is able to produce games of the same quality as Kasparov".
Yes, this is a big complaint of mine too. Other tactics places do the same thing.
Check out Chess Tempo where they have these blitz type problems but also have a "standard" puzzle set/ratings where the only thing that counts is the correct answer, and ignores the time you use... also they have endgame puzzles, all free -- worth checking out.
Sure, so that everybody can take the time, set up the board in Fritz and then "solve" the puzzle (aka press the space bar). Not that this would not have been possible here :) The true tactical ability comes *not* from being able to calculate long lines, but from being able to spot the right move. Yes, looks strange for those who believe things like " if a 1800 ELO player is given enough time, as in CC, then he is able to produce games of the same quality as Kasparov".
In my opinion to be a good tactical player you can't just have one or the other, you have to have both. We could argue which one is more important, but it's kind of pointless since if you're missing either one it's a problem.
That's very exaggerated, but in fact I've heard regular masters are capable of holding their own against GM's in correspondence not even with computers. They seem to make good use of being able to move the pieces and are good enough that they can often figure out a nice plan given enough time. I bet the regular master for the tactics would be more reliant on the analysis board and take more time of course, but anyway the extra benefits of cc seem to favor the master, covering up some of his tactical weaknesses.
Tactical puzzles where the GM is down a knight? Look up the game, he didn't find himself down a knight and then luckily 20 moves later he's winning now -- he saced the knight before and you're trying to find the continuation.
You're right about finding tactics they miss -- they're human too.
20 moves later? Ehh... doesn't happen and that's not a good thing to exaggerate on IMO. It's extremely rare for a GM to just sacrifice a knight with no checkmate or winning material in a few moves, it can happen and make spectacular combinations but it's just very rare. Sometimes they might sacrifice a knight for two pawns or something and only a few moves later it becomes apparent they're winning.
If even in material fairly early in the game, a GM that sacrifices a knight and has no clear advantages from it in 2 or 3 moves at MOST has absolutely lost the game against another GM, no question.
Other people should agree with this don't you...?
I'm also saying it doesn't happen, you misread what I posted.
It isn't rare for a GM to sac without checkmate or material gain, it's called a positional sacrifice (or a "real sacrifice") and it happens all the time.
Fischer's game against R Byrne comes to mind when a sacrifice wasn't apparently winning even to other GMs who were watching -- you saying if there's no advantage after 2-3 moves that he's lost is silly, the compensation is either there or it isn't and that's my point, it doesn't appear 20 moves later.
Tactical puzzles where the GM is down a knight? Look up the game, he didn't find himself down a knight and then luckily 20 moves later he's winning now -- he saced the knight before and you're trying to find the continuation.
You're right about finding tactics they miss -- they're human too.
20 moves later? Ehh... doesn't happen and that's not a good thing to exaggerate on IMO. It's extremely rare for a GM to just sacrifice a knight with no checkmate or winning material in a few moves, it can happen and make spectacular combinations but it's just very rare. Sometimes they might sacrifice a knight for two pawns or something and only a few moves later it becomes apparent they're winning.
If even in material fairly early in the game, a GM that sacrifices a knight and has no clear advantages from it in 2 or 3 moves at MOST has absolutely lost the game against another GM, no question.
Other people should agree with this don't you...?
Might I recommend a few Tal or Bronstein games?
With that said, orangehonda did exaggerate.
Yes, this is a big complaint of mine too. Other tactics places do the same thing.
Check out Chess Tempo where they have these blitz type problems but also have a "standard" puzzle set/ratings where the only thing that counts is the correct answer, and ignores the time you use... also they have endgame puzzles, all free -- worth checking out.
Sure, so that everybody can take the time, set up the board in Fritz and then "solve" the puzzle (aka press the space bar). Not that this would not have been possible here :) The true tactical ability comes *not* from being able to calculate long lines, but from being able to spot the right move. Yes, looks strange for those who believe things like " if a 1800 ELO player is given enough time, as in CC, then he is able to produce games of the same quality as Kasparov".
Tactics puzzles are all about learning, and learning happens via slow repetition. Spending a long time trying to solve a tricky problem does wonders for your board vision (and chess understanding in general).
The problem, IMO, with chess.com's Tactics Trainer is that you have no choice but to move fast, which is an obstacle to the learning process.
Gms are going to have to be good at tactics. For me, I am learning a lot of tactics and am trying to get to 1300 by 2011
Gms are going to have to be good at tactics. For me, I am learning a lot of tactics and am trying to get to 1300 by 2011
Nothing personal, but I have always been puzzled by such statements. In Russia if you come to a chess coach, he will typically expect to help a novice become a 1-st class player (FIDE 1900) in a year (if he studies seriously enough). Or in 2-3 years, if he is studying so-so. At the same time, at Chess.com I have seen a lot of NY resolutions saying "I'll become 1200, 1300, 1400, 1500, etc." in a year...
Gms are going to have to be good at tactics. For me, I am learning a lot of tactics and am trying to get to 1300 by 2011
Nothing personal, but I have always been puzzled by such statements. In Russia if you come to a chess coach, he will typically expect to help a novice become a 1-st class player (FIDE 1900) in a year (if he studies seriously enough). Or in 2-3 years, if he is studying so-so. At the same time, at Chess.com I have seen a lot of NY resolutions saying "I'll become 1200, 1300, 1400, 1500, etc." in a year...
We approached it differently here. No chess coach, no trainer, few if any tournaments. Play online blitz a few hours a day 3 days a week -- then a few times a month flip through a puzzle book or look at an endgame. That's why it takes much longer 
With that said, orangehonda did exaggerate.
Exaggerate and said things that are entirely not true. Such as that if a GM is down in a game and manages a win out of it that he must always have gone down on purpose earlier in the game all leading up to his magnificent win (even "usually" wouldn't be the right word there IMO). A clearly planned sequence of moves like in the Fischer Byrne game is different, and that is a game picked out of millions.
It's nothing personal of course orangehonda and you have a significantly better rating than me, but it's just that I know for a fact that GMs go down material and manage a win sometimes and then it ends up in the newspaper puzzles all the time. Maybe your chess puzzles are different to the ones in the newspaper I get.
Now I see what you guys are saying... ok, yes it was an exaggeration. I also see puzzles where a GM is in the middle of his tactic and plays a wrong move order and looses (meanwhile the puzzle lets you know what should have happened) so I'm not sure what I was thinking...
I saw the Fritz argument again, and I just need to throw in my 2 cents here:
As good as computers are at calculating positions, the score that the computer comes up with is based on values that grandmasters have assigned to certain positions, and is based on how desirable the position or material is. No Chess engine really knows how to play Chess (yet), they're just programmed that way. Someone had to tell it what to do, and what was good. So, although Fritz is lightning-fast when it comes to burning through mountains of positional data, there is rarely something Earth-shattering in the results, at least from the viewpoint of the grandmaster. i.e. Fritz tells you two bishops are good, but every grandmaster already knows that. When you get above 2500 (for real, not with computer help), you'll start to look at positions and just know that certain moves are better than others. You might even be able to say that a certain position is worth a rook-and-a-half (+7.5) just like Fritz would.
Bottom line: grandmasters are usually very good at tactics, sometimes perfect, if they are well-rested. Just read Susan Polgar's Chess Tactics for Champions to see what I mean.
As I go through a tactics book I have and I spend a few seconds to a few minutes on each problem, I often wonder how good grandmasters are. Would they take one glance at a puzzle and instantly know exactly what to do? Or would they still have to think about it for a little bit, and just never make a mistake? Or do they make mistakes sometimes?
Also, one beef I have with the Chess.com tactics trainer is that it's timed and you get more points the quicker you answer, which encourages you not to actually figure out the tactic, but instead choose a move that LOOKS like it could be the tactic at first glance without actually calculating it through. If you told a grandmaster that he HAD to solve a tactics problem in less than 10 seconds, would he do it perfectly, or would he screw up sometimes as well?