SPEAKER:  Deep Blue is the child of IBM's investment and
  research, which takes place at seven sites around the world.
  That research ranges from basic science, to working on people's
  everyday problems.  And Deep Blue technology covers that whole
  spectrum.  I would like to introduce to you Dr. Paul Horne, the
  senior vice-president of IBM, and heads up the IBM research
  division.
MR. HORNE:  Thank you, and welcome.  IBM research has pulled
  together what I believe is one of the strongest collections of
  technical talent ever assembled.  You may know that that's the
  group that invented the disk drive, the D-RAM, relational
  database, RISC architecture.  The list goes on and on.  Five
  Nobel prizes, numerous national medals for technology and
  science.  And over the last eight years, some of the best of
  that technical talent has been focusing on a problem that's as
  old as computer science:  The problem of computer chess.
            Actually, computer chess goes back to 1833.  The
  first chess schematics were built -- were pulled together by a
  guy by the name of Babbage in that year, never constructed the
  machine, but believed he could build something that would be
  able to play chess versus a human being.  And in 1915, the
  first Torries built the first chess computer, and that chess
  computer could actually do a rudimentary endgame, a rook and a
  king vs. a king.
            The first real chess computer was built actually by
  one of the fathers of modern information theory, Claude
  Shannon.  And we actually have that computer here.  It was
  built in 1949.  It could do a rudimentary endgame.  Actually,
  the technology, as you might guess, pales compared to what we
  have available today.  What we're talking about today and
  thinking about today goes far beyond the ability of a computer
  to supplement man's ability to play chess.  It's one of the
  fundamental issues of the 21st century.
            IBM, just one company, has 170,000 pages, home pages
  on the Internet.  It would take a fast reader, operating seven
  days a week, full time, five years to read it all.  We talk
  about the 21st century as the information age.  But all that
  data is going to be useless unless it can be boiled down into
  something useful, transferring random pieces of information
  into useful information, and ultimately into knowledge and
  potentially wisdom, is the key challenge we all face for the
  21st century.
            And it's that technology that we're testing here.
  That technology is our ability to interface with the data and
  the world that's around us and pull in key pieces of
  information that are going to be important for the future of
  all our societies.  So with that background, we believe we're
  asking the computer and ourselves a much more difficult
  question than whether it can play chess, but whether or not it
  can start the beginnings of understanding.  So with that
  introduction, let's get on with the game.
SPEAKER:  Next I would like to introduce to you Monte Newborn.
  He is chairman of the ACM Computer Chess Committee, and he is
  responsible for the officials and commentators for the game.
  In fact, Monte will oversee the match.
MR. NEWBORN:  I'd like to begin by saying what a great pleasure
  it is to be a part of this incredibly exciting chess match.  On
  behalf of the ACM, I'd like to thank IBM and the Garry Kasparov
  team for inviting us here to serve in an official capacity.
  This event is going to be a tremendous landmark in the history
  of computing.
            And I just want to point out one interesting fact.
  The first game that was played by a computer, first chess game
  was played in 1958 on an IBM 704, in I believe the same
  Yorktown Heights where the computer is playing today.  30 years
  ago it was played on a computer that was one million times
  slower than the computer that's playing today.  Imagine an
  automobile a million times faster.
            Last year the score in the match between Kasparov and
  IBM was 4-2 in Kasparov's favor.  Both sides are going to be
  stronger this year.  Kasparov is playing the best chess of his
  life, and IBM's program is clearly stronger than it was last
  year.  So it's a tremendous challenge between the two
  competitors.
            I'd like to at this point introduce the commentator
  team and some of the officials, and then we'll begin, but I'll
  end my own personal remarks by wishing both sides the best of
  luck.
            The match arbiter is Carol Jarecki, who is
  sequestered upstairs on the 35th floor, and she has been the
  arbiter of a number of major world championships in the past
  and will add her expertise to making sure that everything goes
  smoothly at this time.  You won't see her, but you may see her
  on television.  She's maybe got her back to us now.
            The other officials are Ken Thompson, Mike Valvo, and
  myself.  We will serve as officials in the sense that we're an
  appeals committee if there's any problems that go beyond the
  match arbiter.  Our commentary team is a tremendous team who
  will bring this match to life.  Maurice Ashley.  Maurice is one
  of the commentators.  Maurice is one of the top players in the
  New York area, on his way to the title of Grandmaster.  He has
  served as the commentator in the Philadelphia match last year,
  and brings it to life as well as any football announcer on CBS
  and NBC.
            Yasser Seirawan is our second commentator.  Yasser
  has been three times the U.S. champion, is a tremendously
  articulate chess whiz and will assist Maurice in the
  commentary.
            Mike Valvo is our third commentator.  Mike has served
  as an arbiter and commentator at countless computer chess
  events and a number of major human matches as well.  Mike is
  one of the best blindfold chess players in the United States,
  so don't try and fool him by thinking he doesn't see what's
  going on.  I wish everybody a very enjoyable afternoon and in
  fact an enjoyable week of chess.  And again on behalf of the
  ACM, I'd like to thank IBM and Kasparov for inviting us here.
  Best of luck.
MAURICE ASHLEY:  Hello, and welcome once again to the IBM
  Kasparov vs. Deep Blue rematch.  I'm Maurice Ashley, along with
  International Grandmaster Yasser Seirawan and International
  Master Mike Valvo.  And this match I guess is for bragging
  rights as to who has the best chess mind in the universe.  At
  least that's what it's been billed as.
            Garry Kasparov says he's here to defend humanity
  against the ever-encroaching wave of computers.  He's about
  ready to start, not quite yet in the room.  Before we get to
  that, I'd like to get a few comments from my associate, Yasser
  Seirawan.
            Yaz, what do you think?  Has the time come?  Have
  computers gotten good enough yet to whip the best human being?
MR. SEIRAWAN:  No.  (Audience laughter.)  Next question?
MAURICE ASHLEY:  Next question.  Let me mention over here, to
  Mike Valvo, you've been arbiting these computer tournaments for
  15 years now, since 1980.
            Before we do that, Kasparov has arrived.  And he's
  looking quite chipper.  The guy doesn't look worried at all.
  There is Feng-Hsiung Hsu of the Deep Blue team, all prepared, a
  lot of photographers and a lot of press.  USA Today, CNN, New
  York Times is providing practically hourly coverage, and this
  is going live over WebTV and many museums around the country
  also live.  Kasparov is now adjusting his pieces.
            Mike, what is your opinion of this match?  Does the
  computer have a chance?
MIKE VALVO:  I think the computer definitely has a chance.  I
  think people are just too down on computers.  They think that
  computers are going to eradicate the human race, like "2001" or
  something.  But as Ken Thompson said to me, "Do you think that
  you can compare a footrace to a car running?  Is it the same
  kind of thing?"  They are enhancing to us.  They aren't the
  enemy.  They are helping us.  That's my position.
MAURICE ASHLEY:  Why are we so terrified, then?  Why is it that
  everyone is saying no?  A friend of mine, Josh Waitzkin, who is
  the subject of the movie "Searching for Bobby Fischer," says
  this is one of the most terrifying things he can think of.  He
  doesn't want the computer to win, thinks the computer is going
  to take over everything.
MIKE VALVO:  A lot of people think the game is going to get
  solved and no one is going to play anymore.
MAURICE ASHLEY:  Is that going to happen?
MIKE VALVO:  I doubt it.  Checkers has been mastered by a
  computer, and there's still checkers.
MR. SEIRAWAN:  I think that absolutely with chess, regardless of
  whether or not Deep Blue wins this particular match, or a
  future match, I'll always want to beat my brother -- or
  Maurice!  I guess we're getting ready.
MAURICE ASHLEY:  I think we are getting ready.  Kasparov has put
  his game face on.  (Audience laughter.)  All the smiles are
  gone.
            He has drawn the white pieces.  The drawing of lots
  was made two days ago, and Kasparov got the white pieces,
  unlike in the first match where he had the black pieces.
  Kasparov has said that the first couple of games will be sort
  of like a reconnaissance mission.  He's often complained that
  he does not know anything about this new version, the sort of
  big brother to little brother from last year, and he wants to
  know what's better, what has improved, what could be improved.
            Well, one thing that we know has improved is the
  calculating speed.  It's double what it was last year.  I mean
  that's frightening.  200 million moves a second.
MIKE VALVO:  Positions.
MAURICE ASHLEY:  200 million positions a second.  Even more
  frightening.  You just played in the Aegon tournament, a
  tournament in which you did very well, five and a half out of
  six?
MR. SEIRAWAN:  Five and a half out of six.
MAURICE ASHLEY:  Last year you were 6-0, the tournament winner.
MR. SEIRAWAN:  Right.
MAURICE ASHLEY:  What's the strategy against computers?  What do
  you do to beat this kind of thinking?
MR. SEIRAWAN:  First of all, you have to just start from the
  perspective that it's truly amazing to think that a game of
  chess, which is such a complex but structured world, you can
  have a chess computer play against the world's best.  And in
  this particular tournament and in Aegon, it featured the best
  computer -- microcomputer programs.
            I should hasten to say "microcomputer."  Deep Blue,
  as you perhaps well know -- we have a mock-up on the stage
  here.  This is half of what Deep Blue looks like.  It's got 16
  nodes, and 32 processors, so this is half of what Deep Blue is
  like, and this is not a microcomputer.
            But I was playing against microcomputers, and they're
  extremely good, and you have to adjust your way of thinking.
  And computers have forced us to adapt.
            In a sense what you were saying earlier, Mike, I tend
  to look at the computer as a great aid, as opposed to, you
  know, the terrifying thing that's going to replace me.
MAURICE ASHLEY:  Sorry for interrupting, Yasser, but it seems as
  if Kasparov just wants to get it on, and you've got all these
  press people in the room, and they're trying to get them out
  right now.  (Audience laughter.)  "Let's get busy."  And you
  know he knows exactly what he wants to do today.  There won't
  be any surprises as far as his initial moves are concerned.
            Last year, he did figure out what to do against the
  computer with the white pieces, and by game six he dominated
  the computer completely.  It's going to be interesting to see
  what he tries today.  Is he still waiting for the room to more
  or less clear so that he can get started?
MIKE VALVO:  They're really in his space, aren't they?  They put
  a camera right in his face.
MAURICE ASHLEY:  They want some good pictures of him.
MIKE VALVO:  Last year for the first five minutes we couldn't
  control the photographers.  They just clamored all over the
  place, and he just sat there as though nothing was going on.
  That was part of the agreement.  We had to let them do it.
  After five minutes we just chased them away.
MAURICE ASHLEY:  You were in the room; you were the arbiter.
  What did you see about Garry's moves at the beginning of games,
  especially at the beginning of the first game as it is now?
MIKE VALVO:  He was very confident.  His watch was on, which is a
  very important sign, and his coat was on.  Those are important
  signs.
MAURICE ASHLEY:  We will tell you about the coat and the watch
  that will be going on.  Garry Kasparov has certain tell-tale
  signals.  When the watch goes off, that's something, and when
  the coat is off, "Uh-oh, I'm in trouble.  I've got to get
  myself out of this."  Trust us, he does that every single
  game.  We'll pay attention to that.
            He's very, very visual.  He shows his emotions quite
  a bit, and you will know how he feels at practically every
  moment, because he doesn't hide his emotions on the board.
            And he has played a move.  The game has started.  The
  match has begun officially with the move 1 Nf3.
MIKE VALVO:  He predicted that.
MR. SEIRAWAN:  How many of you were around --
MAURICE ASHLEY:  Deep Blue has responded with d7-d5.  Kasparov
  has replied instantly g2-g3, and the computer has also
  responded immediately -- these moves will come very quickly --
  Bc8-g4.
MIKE VALVO:  This is new.  The computer didn't play this last
  year.
MAURICE ASHLEY:  And already, Bc8-g4.  We should mention on the
  Deep Blue team there are very many programmers.  They also have
  an acknowledged expert, another Grandmaster and former U.S.
  champion, Joel Benjamin.  He has basically dedicated a year of
  his life to avenging the loss from last year, and he has no
  doubt worked on all the opening niceties, that he needs to
  check the weaknesses the computer has.
            And so Garry very much feels, Yaz, that he's playing
  against the mind of Joel Benjamin as well as against Deep
  Blue?
MR. SEIRAWAN:  Certainly for the opening stages, certainly.  And
  let me just go with that.  Once again, Philadelphia, how many
  of you were there?  That's a very good representation in the
  audience.  As you'll recall, Maury and I, Mike, just about
  everybody in the world of chess was totally stunned when Deep
  Blue massacred Garry in game one.  And it was really a shock.
            By the time they got to the end of the match, game
  six, Garry looked like he was just conducting an orchestra, and
  he in turn clobbered Deep Blue in game six.  And he clobbered
  Deep Blue with the opening that he chose already in game one.
  One of the problems that Garry has had and will have in this
  match is he doesn't really know anything about Deep Blue, or,
  how do you say, the new, improved Deep Blue.  And so IBM has
  kept its research rather secret.  Certainly Joel Benjamin has
  been in with --
MAURICE ASHLEY:  I was trying to get some stuff from him.  No
  information.
MR. SEIRAWAN:  His eyes lit up when you showed him the five
  bucks.
MAURICE ASHLEY:  They did.
MR. SEIRAWAN:  No.  But the truth of the matter is, the openings
  were -- will be crucial throughout this entire match, because
  what we know today about computers is in sharp, calculating
  positions, they can outplay everybody, including Garry
  Kasparov, the world chess champion.
            Now, in other types of positions, closed or blockaded
  positions, the thinking is that the human will outplay Deep
  Blue.  So Garry has adapted to that.  As white normally Garry
  plays the sharpest, most attacking opening lines, but against
  Deep Blue his style, as he discovered in the first match, was
  to adapt, to play a little bit cagey, to play closed
  positions.  And since he did so well in game six, he's going to
  find out how Deep Blue is going to respond.
MAURICE ASHLEY:  Why is it so hard to program the strategy of
  closed positions into computers?  It would seem as if the
  humans would have solved this problem by now, but computers
  still, from what we've seen, do not play closed positions -- as
  positions with pawns still on the board, locking up of lines --
  still don't play very well.
MIKE VALVO:  You've got to realize that computers think not the
  same way that humans do.  A human looks at the position and
  sees the potential in the position.  A computer looks at it and
  just sees the position.  He answers all the positions, says,
  "I'm ahead," "I'm behind," and when it makes a continuation,
  it projects all the moves and compares the sum of that position
  with the sum of some other position.  So it's not really
  thinking the way humans move.
            We have a new move?
MAURICE ASHLEY:  We do.  And Kasparov has played an unusual move,
  Yaz, b2-b3.  And in the first two moves Deep Blue had responded
  instantly, but after this new twist Deep Blue has decided to
  give the position a little bit of a think.
MR. SEIRAWAN:  Well, let's just talk about the opening for a
  second.  One of the things that we hope to do, first of all, is
  to involve our audience in our discussions, and at the same
  time not talk completely encyclopedic about chess, but to give
  you some insights about what's going on.
            What we have in the opening move is the developing
  move Nf3.  Deep Blue took the center with the move d5, then
  came g3.  The idea is to put this bishop, to fianchetto this
  bishop on the long diagonal by putting the bishop on g2 and
  castling behind the bishop.  This is known as building a
  house.
            When Deep Blue responded with the move Bg4, what he
  was doing was putting direct pressure upon this knight and
  threatening to capture it.  And the normal response would have
  been to defend the knight with the move Bg2 so that the pawns
  would not have been doubled.  From a positional consideration,
  it's considered a disadvantage to have your pawns doubled.
            Garry's move, b2-b3, simply ignoring Deep Blue's
  threat, is a surprise.  I would not have predicted this move,
  and so he, Garry, certainly has an idea of inducing Deep Blue
  to give up the two bishops.
MIKE VALVO:  Let's just think about that for a second.  Garry has
  another motive in mind here.  He wants to get the machine out
  of book.
MR. SEIRAWAN:  What do you mean by "out of book"?
MIKE VALVO:  Computers play openings very badly, and what has
  been done is we programmed into the machines all of the games
  that have been played before as much as we can, and we call
  that "book," prior knowledge.  So the computer doesn't have to
  think on its own in the beginning; it just plays moves other
  Grandmasters have played.
            Here it's obviously on its own.  It would have
  replied right away if it had something in its book, so it's
  thinking.  This is to Garry's advantage.
MIKE VALVO:  Another advantage Garry has, he wants to find out if
  in fact bishop-takes-knight is going to occur.  The computer
  thinks bishops are significantly better than knights, so it
  probably won't take it even though it's going to get a doubled
  pawn back as an advantage.
            Garry has two advantages here.  He wants to get out
  of book, and wants to find out what the computer is going to
  do.
MAURICE ASHLEY:  The computer has in fact ignored capturing the
  knight on f3 and instead has developed a night from b8 to d7,
  and it did not take it too long to make this response.  Garry
  in the meantime looked very confident.  He was aware of what he
  had done by playing the move b2-b3, and now has responded
  immediately with the move, the very natural developing move,
  Bc1-b2, controlling the e5 square.  What do you think about
  this position?
MR. SEIRAWAN:  A normal, average position.  We can't look too
  deeply into the game right now at its earliest stages.  We'll
  probably in a few moves' time come back and start rethinking
  what has taken place.  I just wanted to add a few things and
  get Mike's comments too.
            Essentially we now know today that Deep Blue is doing
  this incredible calculation of 200 million positions per
  second.
MIKE VALVO:  Think about that for a second.  (Audience
  laughter.)
MR. SEIRAWAN:  Every five seconds, Deep Blue is looking at a
  billion positions.  And if it thinks for a minute, then it's
  looking at 60 billion moves a second.  Have I got that right?
  No, it's six billion.
MIKE VALVO:  I was never very good at math.  (Audience
  laughter.)
MR. SEIRAWAN:  12 billion moves --
MIKE VALVO:  It's a big number.
MR. SEIRAWAN:  It's a big number.  The average length of time --
  The time control is 40 moves in two hours for the first
  session.  To be clear, it means that Deep Blue has two hours of
  thought to make its first 40 moves, then one hour to make its
  next 20 moves, and then a half an hour for the rest of the
  moves for the third time control.
            Is Deep Blue thinking when Garry is thinking?
MIKE VALVO:  Definitely.  And you should build into your formula
  that the computer has to reserve maybe 20 minutes because it
  might break down and they have to replace it with another
  machine, and that all counts against them, so they don't really
  have the full two hours in terms of thinking.  They have to
  have that reserve time in case something goes wrong.  But yes,
  they are going to be thinking on Garry's time.  What they're
  going to be thinking about, however, is they're going to guess
  the best move they think Garry can play, they're going to
  analyze that.  If Garry doesn't play that move, then all that
  thinking is wasted.
            So the next question is, how often do they guess
  right?  How often do you think they guess Garry's move?
AUDIENCE MEMBER:  50 percent.
MIKE VALVO:  50 percent is correct.  The computer guesses Garry's
  move 50 percent of the time.
MR. SEIRAWAN:  That's pretty amazing.
MIKE VALVO:  Pretty amazing.  That means they get twice as much
  time for that move.
MAURICE ASHLEY:  Kasparov has written down the last move Deep
  Blue has played, which is e7-e6, has advanced a pawn one
  square.  And the position again, as you can see, remains
  blocked.  No locking of pawns as yet, no exchanges of pawns in
  the middle of the board to open up lines, and it's a pretty
  quick opening going on.  I mean you can see that Kasparov is
  kind of keeping it at arm's length, just making sure that he
  plays very carefully.  Doesn't want to blunder; doesn't want
  any kind of crazy tactical melees starting too soon.
            Do you think, Yasser, that he's probably satisfied
  with an early draw in the beginning?
MR. SEIRAWAN:  Oh, no, no, no.
MAURICE ASHLEY:  Would he be happy to draw just to get rid of the
  nerves?
MR. SEIRAWAN:  No, I think that chess, very similar to, say,
  tennis, having the opening move, that is to say "the right to
  serve," is quite an important advantage, especially amongst
  chess professionals.  And Garry will definitely be at a
  disadvantage when he has black in this match.  I mean Deep Blue
  will really put a lot of pressure on Garry throughout the match
  when Deep Blue has white.  So Garry knows that his best
  opportunities of winning the match are when he's white.
            So I think he was perhaps a little disappointed to
  have white in the first game, because I agree with you, he
  would like to make an early draw, but with the black pieces,
  not with the white.  I'm sure he feels pressured today to do
  his utmost to win.
MAURICE ASHLEY:  The other thing is if he loses this game, he'd
  be in a huge hole.  Last year when he lost, he lost with the
  black pieces, and he could come back with the white pieces and
  be very aggressive, do what he needed to do.  But this time, in
  this game, that could be a big hole.  May not get another
  chance until game three.  He's now moved his bishop from f1 to
  g2.  Normal developing move, getting all the pieces out.  Soon
  he will be castling.
            And, Mike, you made a comment about all the positions
  that the computer is analyzing.  You say 200 million positions
  a second.  Chess is such a rich game that what happens with
  those moves, or even 90 percent of these moves?  They can't be
  really good chess moves?
MIKE VALVO:  They're junk.  Most of the positions are junk.  And
  a human can easily say, "Well, I won't even bother looking at
  this because it's no good."  That's the advantage humans have.
  They can disregard a whole bunch of moves.  Computers got to
  look at absolutely everything.
MAURICE ASHLEY:  What's the advantage of 200 million positions a
  second that turns out to be junk?  I know myself only two or
  three moves --
MIKE VALVO:  They're moving quickly again.
MAURICE ASHLEY:  No doubt Deep Blue anticipated the last response
  and has quickly played Ngf6.
MIKE VALVO:  It varies, obviously, but there are lots of
  positions with good moves.  One of the things computers use to
  figure out what to look at, they have a thing called the
  alpha-beta search, and they have a thing that says, "Well, gee,
  unless a move is better than a particular move, I'm not even
  going to look at others in this area."  That saves them about
  50 percent of the moves that they look at.  So they do have
  some ability in that area, but not always.
MAURICE ASHLEY:  What did you use successfully in the Aegon
  tournament, playing against computers?
MR. SEIRAWAN:  Let's see.  Pulling out the power cord helped a
  lot.  (Audience laughter.)  Why are you laughing?  It worked.
  Bribing the operator.  You know, simple things, simple things,
  really.  And really closing the position down.
            Again, what Michael was saying is that the computer,
  when the computer is faced with an opportunity of making a
  capture, well, that's where it's going to focus all of its
  attention, and all of those hundreds of millions of moves a
  second suddenly become very, very useful to have.  So, by
  creating as large an opportunity of moves as possible, then the
  computer starts looking at a lot of things that aren't
  significant.
            We have another move.
MAURICE ASHLEY:  Kasparov has castled --
MR. SEIRAWAN:  Bringing his king.  This is known as building a
  house.  What Garry has done is fianchettoed his bishop; that is
  to say, putting his bishop on g2 directly in front of his king,
  his knight on f3, a very nice little pawn shield, so that
  white's king is nice and comfortably ensconced in his own
  little home, so that he will later turn his attention to the
  center.
            And I just wanted to kind of set up ourselves here
  with a bit of our props.  What we have behind us in the middle
  screen is of course the two-dimensional representation of the
  chess game.
            And in the lower right-hand corner, you see those
  nice little green and red bars?  The chess engine called Fritz
  is giving its evaluation.  So you see the evaluation, a
  plus-over-equals.  In chess-speak, that means that the computer
  believes that white has a slight advantage.  Right next to it
  you have an evaluation of .22.  What that means is that the
  computer believes that Kasparov's position is 22 hundredths of
  a pawn better; that is to say that Garry has an edge of
  two-tenths of a pawn.
MAURICE ASHLEY:  Sorry to interrupt you, Yaz, but Deep Blue has
  played another move, c7-c6.  A very solid move indeed, and this
  would not be typically the strategy you would expect a computer
  to employ.  You would expect much more sharp moves.  Right now
  it's playing a very solid position.  One thing I heard, though,
  is that this year they've tweaked it a little bit and given it
  more chess knowledge, so to speak, given it a certain ability
  to play these types of positions that it really did not have in
  the past.  How successful do you think it could be at this,
  Mike?
MIKE VALVO:  I think it could be the critical factor as to how
  well Deep Blue does this year.  I know from behind-the-scenes
  kind of stuff that the whole positional evaluation that they
  had before has been improved.  And if they are able to solve a
  lot of these pinning situations -- they had a problem with pins
  last year -- if they could solve those kinds of situations,
  they could be a real problem for Garry.
            One of the questions that I didn't finish answering,
  you had asked before, why can't computers play well in blocked
  positions.  Well, in open positions they have a thing called
  move extensions.  Which means, if I capture something, I'm not
  going to stop analyzing there.  I'm going to see what happens.
  If he can recapture, then I will exhaust all capture situations
  down to their end, which is something I'm not going to bother
  doing in some blocked positions, not even have the opportunity
  of doing.
            In tactical situations that could be a big edge to
  the computer.  Tactical situations in fact are to a computer's
  advantage.
MAURICE ASHLEY:  Right, we saw that in game one last year where
  Kasparov --
MR. SEIRAWAN:  Got massacred.
MAURICE ASHLEY:  Got crushed.  He sacrificed a pawn thinking that
  he had a great king-side attack, but the computer had
  calculated it to the finish, and admittedly humans would not
  have taken that pawn, unless you were drunk or something, but
  in that situation the computer had worked everything out to
  Kasparov's being crushed, instead of itself being mated.
            And so Garry ended up losing that game, and he
  learned his lesson well and avoided sharp, open, tactical
  positions from that moment on.
MIKE VALVO:  Can I tell you something from last year's game that
  wasn't well known?
MAURICE ASHLEY:  Please.
MIKE VALVO:  Garry left his score sheet behind.  That's the only
  game he forgot his score sheet.  He was pretty upset.
MR. SEIRAWAN:  You're talking about the game he lost in game
  one?
MIKE VALVO:  The game he lost, yeah.
MR. SEIRAWAN:  Probably didn't want to keep it.
MIKE VALVO:  If you are familiar with chess collectors, score
  sheets are worth about 50 to $100 a score sheet.  He asked me
  later, "What happened to my score sheet?"
MAURICE ASHLEY:  Well, there are two perspectives on this match.
  Some people think it's a total mismatch.  On one side you have
  Garry Kasparov, world champion, acknowledged being maybe the
  best player of all time.  Just had a great tournament, Las
  Palmas.  Massacred everybody.  His rating is now 2820,
  something frightful, up in the stratosphere.
            He knows everything -- eats, drinks, talks, sleeps
  chess.  You don't want to be near the man on a chessboard
  because you're going to get killed.  And in addition to that,
  he devastated the computer last year, 4-2, three wins.  After
  the first game, didn't lose another game, looked very
  confident.  People are saying, "Well, it's all Garry."
            On the other side, Deep Blue has improved to 200
  million positions a second, double what they did last year.  On
  a key note, it played at least even with Garry three of those
  games last year -- drew two and won one.  Maybe its chances are
  not so bad, especially with all the help it's gotten over the
  last year.
            Kasparov has been thinking about other things other
  than Deep Blue since last year, and all the Deep Blue team has
  been thinking about is this match, the be-all, end-all of their
  existence to the moment.
            I do want to say this about those following over the
  Internet.  Some will see this visually, some with WebTV will be
  able to watch all the action as it's happening now.  Some
  won't, so we'll try to set everything up for you as Yaz did
  before.  We have three huge computer screens, one showing the
  board at all times, another showing Garry -- sometimes the
  board, and sometimes the operator.  Unfortunately, we'll never
  get a chance to see the Deep Blue screen.
            We do have a clock.  This is a new clock that Garry
  Kasparov has instituted, and it shows the time, what's going
  on.  Right now Kasparov has used 12 minutes, and Deep Blue has
  used eight minutes.  So it's a little bit ahead on the clock.
  On the second screen we do have Fritz 4, where we'll do all our
  analysis.  Right now, not too much happening, because pawns are
  not touching.
MR. SEIRAWAN:  Yes.
MAURICE ASHLEY:  What should be the plan here?
MR. SEIRAWAN:  We see Garry as kind of -- he seems to have his
  head in the sky a bit there, what he would like to do.
            Essentially what people miss about chess, as far as
  the opening is concerned, the entire purpose of the opening is
  to get a playable middle game.  That is to say, hey, the kind
  of position that the player would like to play.  And then as it
  kind of dovetails into the middle game, the idea is "What will
  be my overall strategy?"
            So Garry is not looking seriously -- he is not
  looking at one or two moves in distance acquisition.  What he's
  thinking is "What I want to do overall with my pieces."  Now,
  the essence of the question that he has to ask himself is, "How
  do I control the board?  What is happening?"
            And in chess, in great battles, like in warfare, is
  to control the center of the board.  And by the center of the
  board, I mean specifically the squares from c4, d4, e4, all the
  way from e4, e5, e6, back to e6, d6, c6, and back through c5
  and c4.  This area here is the center of the board.
            So what Garry could consider doing is bringing the
  c-pawn to the c4 square, bringing the d-pawn to the d4 square
  to control e5, or trying to advance his e-pawn later in the
  game.
            The point is, what Deep Blue has set its defenses
  around is the key square, the d5 square and the d5 pawn, very
  powerfully supported by the c6 pawn and the e6 pawn.  So what
  Garry has to say to himself is, "How am I going to make a dent
  in Deep Blue's position?"  And again, he's not looking at one
  or two moves.
            He's saying, "Overall, what will my key strategy
  be?"  And quite frankly, he has, I would say, three key
  strategies.  The first key strategy is to play the move c2-c4
  followed by Nc3 and try to play on the left flank, or the
  queen-side flank.
            A second strategy would be to try to play d2-d4,
  bring the knight on b1 to e2, and follow up with Nf3-e5,
  playing on the dark squares.
            And the third strategy is to play the move d2-d3,
  followed by Nb1-d2 and then e2-e4.
MAURICE ASHLEY:  We should note that the watch is off!  (Audience
  laughter.)  And that means another 20 minutes, probably.
MR. SEIRAWAN:  I must say that I have played Garry countless
  blitz games.  That is to say, "fun" games.  They weren't THAT
  much fun, but we played a lot.  (Audience laughter.)  And we
  have played together seven tournament games, including a game
  played in the Olympics.  My score with Garry is two draws, one
  victory for me (pumping fist) -- yay! -- now the bad news --
  three victories for him.
            And during this many hours of playing with Garry, and
  also not just playing against him directly, but also being in
  competitions with him, he reveals himself.
            Garry is very much an open person, and he wears his
  emotions openly.  When he doesn't like his position, he's
  shaking his head, and he's gnashing the teeth and -- grrrr --
  growling.  And when he's winning, he's beaming.
MAURICE ASHLEY:  Surprise, surprise.  The move has come quickly.
  He's played the move d2-d3, and instantly Deep Blue has
  responded with Bd6.  Now, this means one of two things --
MIKE VALVO:  This means one of two things.
MAURICE ASHLEY:  Kasparov has continued the bravado and played
  Nb1-d2.  And the action came pretty quick there with Deep Blue
  still continuing, as if saying, "I dare you to top this," he's
  castling.
            And Kasparov is instantly responding h2-h3, so a
  flurry of moves --