Tactics & General Principles

History
Originally chess was played without a clock.  The chess clock was
introduced in the 1880's when high-level chess organizers found
that some spectators, who had paid a fee to watch the masters,
wanted their money back because a few of the masters spent all
day trying to find the best move.  So the objective of the game
changed - from trying to find the best move to trying to find the
best move given the time available.  Remember this objective,
because we will return to it later.
By the time I started playing (not too many years later ), local
tournament were all 50 moves in 2  hours (written "50/2"; if the
number following the slash is a small integer it represents hours; if
it is a large integer it is minutes) followed by 25/1 for the second
time control, etc.  This standard was soon to be changed to 48
moves in 2 hours because then the same rate was available for "1/2
hour" 2nd and 3rd time controls, e.g. 48/2 followed by 12/30 (min). 
The problem with these time limits was that organizers had to
adjourn (or worse, adjudicate) games that either ran too late into
the evening or, in the case of multiple rounds in the same day, into
the next round.
Adjournments have to be played off at a later time; this causes
problems with swiss system pairings, which require the result of
the game.  It also allows third parties to help with the adjournment
analysis.  
A major change occurred after the 16th game in the 1990
Kasparov-Karpov match.  They adjourned in a very tricky
position, the kind that computers play very well.  Of course
Kasparov, who has used computers regularly to train, legally
sought out computer analysis to find a crucial win.  After that,
chess officials decided that "sudden death" time limits would be
allowed in order to force a game to be finished without
adjournment.
Sudden death requires a player to play all the rest of his moves in a
given time, e.g. 30 minutes.  This is written as Game in 30
minutes, or "G/30."  Sudden death had been made legal by the US
Chess Federation even before the 1990, but after international rules
changed, sudden death became much more popular at the national
level.  And, while international games were initially 40/2; 20/1;
G/30 min for their three time controls, many major events began
using sudden death for their first time control, e.g. G/90.
To combat the possibility that in a long game a player might have
to make many moves in a very short period of time, the time delay
feature was proposed.  One of its original proponents was Bobby
Fischer!  With a time delay, the clock either does not run for a
specified period (say 5 seconds) at the start of each player's move,
or else time is added each time a player moves.  Thus if there is a
five second time delay and you have 1 second left on your clock,
you can play an indefinite number of moves so long as you don't
take more than five seconds for each move.  This type of playing
rate, which requires a digital clock, has become standard not only
on the Internet, where transmission times and mouse movement
require it for fairness, but also over-the-board.
The most recent side-effect of the sudden death time limit is the
US Chess Federation's "insufficient losing chances" rule.  This
states that if you are NOT playing with a time delay clock and you
are about to lose on time in a clearly non-losing position, you can
claim a draw.  Without going into the complexities of this
controversial rule, suffice it to say that one of the options the TD
has, should the opponent not want a draw, is to institute a time
delay clock into the game at that point.
PROBLEM
What all of the above has done is make time management, which
used to be a relatively minor issue at 50/2 or 40/2, as a major
element for almost all players.  Players who don't manage their
time well during a sudden death time period are at a distinct
disadvantage.
In the "old" days, players such as Sammy Reshevsky, Donald
Byrne, and Walter Browne used to regularly run their clock down
to 1 minute or even 30 seconds to make their last dozen moves or
so. At that point, if the game was still not over, they could get
more time for their 2nd time control.  Under the new rules this
additional time often disappeared, but with a new twist.  If they
were playing with a time delay clock, they could run their clock
down as much as they wished and still get five seconds per move! 
And if they were playing without a time delay, then running their
clock down to make an indefinite number of moves (rather than a
specified number) would definitely be bad strategy!
As a result of these new time rules, I now spend a lot of time (no
pun intended) with my students teaching them time management. 
What can be done to learn to use the clock better ?
Let us illustrate the importance of time management skills with a
question: If you were playing without a time delay - and in a
tournament that did not allow "insufficient losing chances" - would
you rather be up a Queen in the middlegame with 10 seconds left
or be down a Queen with 10 minutes left?  Not too many players
are speedy enough to checkmate in 10 seconds from a middlegame
position!  
Therefore, depending upon the rules, at some point the clock
becomes more important than the position!  With a time delay
clock, that is not always true.  However, relying on a time delay
clock to save you is not always good strategy either - in many
complicated positions, most human brains need more than 5
seconds to arrive at a decent move (international play often allows
more than 5 seconds per move, but I have read articles from
international participants that the same logic still applies).
So therefore one needs strategies that will avoid or mitigate these
problems:
STRATEGIES AND METHODS
1. Before any tournament, look at the time controls and the
relevant rules (Is the insufficient losing chances rule allowed?  If
you use time delay, do you have to start with less time on your
clock?).  Figure out about what pace you should play.  While it is
illegal to have analysis notes on your scoresheet, making
"milestone" time marks is not (at least thus far).  So you can mark
down in a G/90 tournament that, ideally, you would like to have 45
minutes left at move 20.  Just put a line under move 20 and circle
"45."  This preparation is similar to readying your opening
repertoire before the tournament starts.
2. Try to find the best move given the time available!  Remember
this principle from the start of the article?  What it means is that
the theoretically best move often cannot be determined by a human
in a short time.  So therefore your goal is to find the best move you
can, given how much time you have left; taking inordinate amounts
of time to prove the best move in every position will probably get
you into time trouble even in a slower game.  Sometimes you just
have to be practical and say to yourself, "Well, after some analysis
of my candidates moves, this move is safe (or interesting, or
reasonable); let's play it and see what happens."  That is not to say
that you shouldn't analyze properly or look at other moves; it's just
that doing so to the Nth degree on every move is not always
practical. -- On the other hand, sometimes the best move can be
determined in a reasonable amount of time.  In that case,
sometimes players waste time after they have determined the best
move, trying to figure out what might happen.  That extra analysis
is unnecessary; if you have proven that a move is the best one, play
it now! -  worry about what might happen next during your
opponent's move or your next one.  As a trivial example, suppose
your opponent gives a check and you only have 4 legal replies. 
You analyze the first 3 and see that each leads to an "easy" mate-
in-one for your opponent.  You can then play fourth move
instantly, because it cannot be any worse than the other three, even
if you have no idea what might happen next!
3. Avoid playing too fast or too slow, no matter how fast your
opponent plays.  When one of my students says he played too fast
because his opponent played fast, I ask him, "Would you jump off
a cliff if your opponent jumped off a cliff?"  On the other hand, it
is helpful not to fall too far behind a reasonably-paced opponent. 
One of my better students was playing a G/45 game and fell
behind a good opponent in a complicated position, 37 minutes to 7
(!).  Needless to say, even though his position was about even at
that point, he eventually lost.  I will give a big tip: in any long
game where the battle still remains in doubt, try to have 15
minutes left when your opponent has 5 minutes left.  I asked
several players how well they would do against their clone if they
had to give 5 to 15 odds.  Most said they could only win about
25%.  This seems correct.  Since 25% indicates about 200 rating
points, then if you can get 15 minutes left to 5 in an even position,
it is like adding 200 points to your rating, or raising a 50-50 chance
to 75-25!
4. Take (almost) all your time every game!  Go to any Open
tournament and you will see that the best players are the ones who
use almost all their time, every game.  Just as in the previous note,
I also asked some of my students, "If I cloned you and you had to
play your clone, but you took 5 minutes for the game and your
clone took 90 minutes, what percentage would you win?"  Most
give the reasonable answer of 0-5%; this means that if you are a
very fast player and some of your opponents take their time and
beat you 19 games out of 20, then they might not be any better
than you would be if you took your time!  There is only one
exception to the advice of taking almost all your time: If you are
completely lost, but are playing on only because your opponent is
short on time, it is not good strategy to play slowly and let him
think of how to win on your time; you should play quickly and
hope for the best (because otherwise you lose).  However, don't
confuse this with a game where you are NOT hopelessly lost and
your opponent is short on time; in that case it is very wise to make
use of your extra time to think.  Remember, your opponent may
anticipate your move and make some use of your time, but you
surely know which move you can/will make, and thus can use that
same time (when your clock is running) much more efficiently.
5. In a sudden death time control, speed up a little if the game is
very even and it looks like it is going to be a long game.  You may
need that time later if things get complicated or it does become an
exceptionally long game.
6. In a sudden death time control, start speeding up when most of
your time is done (but don't wait until almost all of your time is
done).  For example, if you are playing G/90 and you are getting
down to less than 20 minutes and the game looks like it is far from
over, start speeding up then, and not when you have 3 minutes
left 
7. If you are playing a short sudden death time control, like G/30,
don't take a lot of time over subtle moves which are unlikely to
effect the evaluation much.  For example, if nothing much is
happening and you want to play Rad1 and Rfe1, don't take two
minutes to decide which you want to do first.  Just look at any
possible tactics and make a quick decision.
8. On the other hand, if the game is very tactical, that is when you
want to use your time (no matter what the time limit).  A tactical
opening, like the Traxler variation of the Two Knights Defense
(my book on computer analysis of this opening is due out in March
2000), is likely to be decided within the first 20 moves, so it makes
good sense to use most of your time trying to ensure that you are
the one who is winning at move 20.  A general guideline would be
"save you time for when the game is most complicated; the player
who outplays his opponent when the impact of errors is greatest
(during the tactical phase) is most likely to win.  For example,
when I am analyzing a game with the computer, the difference
between the first, second, and third best moves is often well less
than 0.1 pawns during the development phase, but can be as much
as a pawn or two during a tactical melee!  So if you think the 2nd
best move might be a lot worse than the best, that is a good place
to use more time.
9. When in time trouble, if everything else is equal, make "safe"
moves.  Put your pieces that are on squares that are protected,
move pieces two squares diagonal from Knights, make "luft" for
your King, put pieces on the opposite color of your opponent's
Bishop, etc.  That way you can arrive at safe decisions more
quickly.  Try not to run your clock under 1 minute except in an
emergency, or when playing with a time delay clock.  If you are
not playing with a time delay clock, then at some point you have to
realize that it is more important to move fast than it is to make a
good move!  In this circumstance, sometimes the side which
moves faster, but not better, just wins on time.  For example, when
you have 1 minute left and no time delay, it is hardly ever worth
30 seconds to figure out if you can save a pawn, except in the deep
endgame.  Similarly, very few moves are worth 90 seconds when
you have 4 minutes left.
10. Practice at a mixture of time controls.  Play slow games to pace
yourself and to learn good analysis techniques.  Play fast games to
practice your openings and get time pressure experience.  And
instead of playing the traditional G/5-minute with your friends for
fun, play G/2 minute with a five second time delay!
11. Know the rules!  There are different rules for non-sudden
death, sudden death, non-sudden death with less than 5 minutes
left, and sudden death with less than 5 minutes left!  For example,
FIDE recently passed a rule that you have to play with one hand all
the time - it used to just you had to just when in time pressure. 
Also, both sides have to keep score until either side has less than 5
minutes left.  And if you are playing without a time delay, the US
Chess Federation will not allow you to claim "insufficient losing
chances" until you have less than 5 minutes left.  Your national
chess federation's rules can usually be found at their web site, or
you can buy a copy of the rule book.