My question, is how can I improve the accuracy/speed of my tactics?  Especially in 'branching' variations situations, i.e. where there is not just one, clear-cut forcing line?  I'm making a ton of mistakes in live chess because I have so little time per move (doesn't everyone at my level...) and I can't seem to snuff out these blunders/oversights except for when I play OTB (at much longer time controls, say Game 60).

Should I keep practing tactics slowly, and try to gradually work up my speed?  This approach is terrible in Tactics Trainer because of that stupid time limit, where I have a 1500-1600 rating just by guessing on about half the puzzles.  I feel like tactics/accuracy are what are holding me back right now; I have had a chess.com 1500's rating in standard live chess for awhile, but I'm making far too many errors so I can't improve.

I've got the same problem - I take a long time to figure things out.  I think it's too bad that the tactics trainer actually deducts from your score if you run out of time but still get the answer correct.  Surely you should get some credit for making the right move?

Forget about the score for now.  Just concentrate on getting the correct answer.  Eventually your speed will improve and so will your score. 

Thanks, Okolo.  That sounds like good advice.

I'd suggest to run the tactics trainer in the untimed+unrated mode. The problem if you run it in standard mode (timed+rated) is that as your rating goes down, even if you don't care about it, you will only be presented with the corresponding low rated tactics problems which will soon be much too easy for you since you're rated 1700. In the unrated mode TT will bring up problems of all difficulty levels.

There is also at least one non-chess.com tactics trainer on the internet, where you can choose between timed and untimed mode, and they have separate ratings. But I won't mention it here...

I have the opposite advice of the rest of the thread :)

Tactics trainers that force you to move quickly are training your pattern recognition. If you have to calculate out every possible tactic, you will always have tactical errors in your games. You need to train your mind to recognize the simple patterns without the need for calculation.

My advice which has worked quite well for me. Buy 303 tricky chess tactics and another puzzle book like chess tactics for advanced players or 1001 winning chess sacrifices and combination (I own 1001 personally) and go over them repeatedly. If you go over them enough times you will be very strong at tactics and improve significantly in blitz.

Tactics trainer or other internet tactic programs force you to move quickly and almost guess tactics when you should really be finding them and calculating them out enough times that it becomes easy and you have lots of pattern recognition to go on.

Tactics trainer is a much better way to gauge your strength then help you to improve. Maybe you're thinking Phelon just hasn't tried tactics trainer, or that I don't know what Im talking about. I do know what I'm talking about, and I'm 2610 on tactics trainer.

I think both sides of training are valuable ... Personally I usually do quick chess (blitz games, timed tactics trainer) on work days, and slow chess (longer games, untimed tactics trainer, chess mentor, and the good old chess books) on free days :)

On the one side, the timed TT is great for pattern recognition (like the smothered mate, or simply mates-in-one or hanging pieces - that's one thing I like about TT, that it doesn't only give you high-sophisticated combination, but a simple hanging piece that calls for being overlooked as well ...). On the other side, I like it when it gives me a problem far above my rating, somewhere in the 2000s, and I have time to ponder it at least a few minutes, which is good for concentration in long games ...

It isn't at all improving my play to have to guess tactics incredibly quickly.  I am under the distinct impression that most people, on the short time limit problems, just blitz out whatever tactic they see without also working out the consequences.  Hence why many problems with, say, a move that looks good but isn't, have such tremendously low pass rates.  Accurate tactical play isn't just about recognition (although of course that's a huge part of it) but it's also about calculating ability, and I need to improve at BOTH.  So I like the idea of putting tactics trainer on unrated/no time limit.

It sounds to me like you are justifying what you enjoy the most as being what will help you improve the fastest. Maybe it will, but I don't think that's automatically the case.

But any practice is better than no practice, so doing what you enjoy is probably good in the long run :)