Amazing Chess Game: Kasparov’s quickest defeat: IBM’s Deeper Blue (Computer) vs Garry Kasparov 1997

Amazing Chess Game: Kasparov’s quickest defeat: IBM’s Deeper Blue (Computer) vs Garry Kasparov 1997

Morning all. In this age of engines and
very powerful chess computers I thought it would be interesting to check out the
game six – the final game of the deep blue versus Kasparov match in 1997 so this
was by IBM who constructed Deep Blue nicknamed ‘Deeper Blue’ actually by Kasparov – it was an upgraded version so it’s a rematch
. The name ‘Deep Blue’ by the way was after ‘Deep thought’ which was in Douglas Adams ‘Hitchhiker’s Guide to the galaxy’ – the computer set out to find the answer to
life the universe and everything and took quite a few years – millions and millions of years and came up with ’42’ but anyway ‘Deep Thought’ was renamed to ‘Deep blue’ and you have a lot of engines after that with the prefix ‘Deep’ in them – okay so deep
blue game 6 a very memorable game and so it marked a very important event really –
so let’s have a look at this game game 6 so 1.e4 from Deep Blue – Kasparov
played the caro-kann which is usually you know enjoys a rock-solid reputation
after d4 d5 we see Nc3 black takes on e4 Nxe4 Nd7 which
is safely going to exchange off the Knight here without incurring any
structural damage usually – now we see this aggressive looking Knight g5
already probing some sensitive spots like f7 – black plays Ngf6 and now
Bd3 okay so far so good … Kasparov
plays e6 and now after the N1f3 the usual move would have been like a
bishop move but Kasparov played something which is very untypical of him
he’s usually a great opening theoretician but I believe he was
annoyed from an earlier game in the match where he thought something was
really suspect about Be4 which we’ve covered in an earlier video on
this channel actually – he suspected something was not quite right but he was
already a bit upset by the stage so but here yeah he seemed to forget his his
opening theory and it’s needed for great precision here – instead of playing the
Bishop move which would mean you know the King’s ready to castle and it’s got
you know f8 to go to in certain circumstances he plays the move h6 – I
wonder if you can spot how the computer punished this move ?! It gives black a
difficult position this next move – I wonder if you can spot if i give you 10
seconds starting from now … Okay not a normal Knight retreat anywhere
but Nxe6 and okay it’s a knight sac for just one pawn – but it’s
very very dangerous – you know if it fxe6 okay Bg6+ – the King is in
the center white castles and we have a an imprisoned rook like a previous video
recently well this looks quite unpleasant – how does it get out ?! But I
believe actually this position has occurred in a game of Murray Chandler
against Michael Adams which I witnessed that Lloyd Bank masters many years ago
which I think Adams managed to wriggle out but he was complaining that his rook
being in prison for much of the game okay but um so in this position Kasparov didn’t take there – he played actually Qe7 but it doesn’t really
help things much white just castles here so he’s got the skewing option if Qxe6 then Re1 skewing the Queen to the king so he’s in big big trouble
against the computer here already he plays fxe6 – he’s slightly
improved the state of affairs that the King doesn’t have to go to e7 – it can go
to d8 now but it’s a bit depressing to have this position against computers
– Kasparov was trying to play against the computer style and research
computers a lot their strengths and weaknesses and would have been horrified
with this position because it’s going into the computers major strengths you
know it’s a very very tactical position tactical liabilities for black for the
computer to exploit but here after Bf4 he makes things lot lot worse
than perhaps what they needed to be – we should really get an engine evaluation
of this position in the second pass through this game but this next move is
described as quite a bad blunder by several grandmasters – basically in
terms of King safety you really want open lines but secondary to that –
that the defender shouldn’t create open invitations to
open lines but this move shows maybe Kasparov’s huge frustration with the match
and is basically after b5 is played here but played
it’s an invitation open invitation to peel open lines not just with a4 but
with c4 if lines are opened up like the C file
the A file then access routes the black King are going to be generated so it’s
really an open invitation for access routes to the black King here and the
first way that’s used is with a4 okay which is very very nasty – black
plays Bb7 trying to hold things together at least protecting his rook
and maybe if takes you know he’s got a nice Bishop on the diagonal. The computer
doesn’t need to take on b5 to liberate the bishop though – it just plays Re1 so
another target is set sight on look at this imprisoned rook (on h8) – here so the knight sacrifice is very very powerful in this position – if blacks pieces are just in
prison it’s not just the rook it’s the bishop that’s in prison so Nd5
okay an active piece you might think Bg3 and the King running a bit to
safety in inverted commas Kc8 but there is that C file and that lever c4
to make use of pretty soon but white first plays axb5 seemingly allowing
a potentially useful Bishop on the diagonal but it’s pretty cruel now
because Qd3 hits b5 – how is b5 actually defended?! And it also sets up
some possibilities if the knight wasn’t on d5 for Qc3+ and also c4
for this Queen to join forces with the Bishop which is cutting across that very
sensitive diagonal so very unpleasant position after Bc6 – now this pawn
is picked on even further with Bf5 exploiting the pin on the E file so we
have these two very very dangerous bishops on these diagonals. Kasparov in desperation well it’s a pretty hopeless position now because how do
these defend actually e6 – if he plays Nc7 – maybe they can just be snapped
off and then white can take on e6 – it looks
fairly fairly unpleasant so where would the Queen go here and there then there’s
possibilities of Rxc6+ and Qe4+ and winning the rook
over here – there’s all sorts of nasty possibilities after this but will engine
check this in the second pass through the game so here though Kasparov decided he gave up his queen so in theory because it was already a piece up it
doesn’t seem technically as though you know maybe there’s a hope for a balance
of forces here because it’s rook and two pieces now but the extra piece from
before but it doesn’t really balance the Queen because in this position black’s
King safety it’s pretty bad it should be noted if black had one move you know maybe f4 looks that looks good to do something with this Bishop to win
this bishop but c4 here is a very powerful punch in the position
unfortunately and it compels actually Garry Kasparov to resign and I think he
was really really annoyed with match generally and you know wanted evidence
of a previous game in the match for where the computers seem to play human
move which I’ll put a link for that video in the description of this one but
here yeah he resigned so it marked actually the first time a computer had
defeated a world champion basically in a match of several games very serious
games and he was obliterated – you know 19 moves so it’s a miniature – let’s have a
look what goes on in this final position with the dreaded Houdini engine which
we’ve all got access to on our computers now so bxc4 Qxc4 okay
hitting the bishop if it if black tries to defend with that (Nb4) so an incisive move here apparently is Re1 although Ra6 also looks
tempting let’s go with Re1 because I suppose it introduces Re6 – I wonder just in fact d5 should be strong – that seems simple and strong as well
no Nb6 and it could get tricky okay so white would would have to be
still accurate because black has quite a few pieces so let’s see – Re8 Bd6
winning further material in this position Qxg7 and white’s significantly
up and I think these pawns are going to start being useful but it’s a
very unpleasant position for black to try and coordinate a lot of loose pieces
against the Queen and he starts lose even more material
now I think just Qxf5 – I think it’s a big advantage – so let’s go back on Re1 is there actually anything else?! so if Bf6 what is the major threat ?!
Is it Re6 ?! no Qxb4 – okay so this Re1 is actually targeted to
b4 to winning more material with Rxe7 maybe and Qxb4 – so here
that this is a back row defense of course Re8 – Rxe7 Rxe7 Qxb4
there’s there’s um even this though is good for white so that’s why the final
position seems fairly convincing after the simple move Re1 in fact – that
seems pretty pretty crushing now if we go back earlier in the game – so we will go
back earlier so in this position the actual evaluation on this position after
Nxe6 – it’s given as a move but you see Houdini liking Ne4 so it’s
not that convincing that Nxe6 is the strongest move at this depth but um
I think Kasparov himself may have indicated Nxe6 as refutation – it’s a
deep refutation but um it’s to do with the imprisoned rook again scenario which
is difficult for engines to evaluate so we’ve got this imprisoned rook on h8 as
an implication of Nxe6 and so actually fxe6 is given and
initially we seem to have this this situation where it’s just slightly
better for white – I was checking this out on and actually there’s
there’s a few games but statistically it favors White – it’s not as if there is an
immediate mate or something – it’s just very unpleasant for black to try and
defend but technically you know it looks as though initially at least it doesn’t
it doesn’t seem that bad but it just gets worse and worse I think I think
with the rook on h8 .. Nxe5 dxe5 Ne7 – Is there compensation?! Is that actually winning if a computer was playing black obviously this is a much
more solid approach than Kasparov’s approach so really I mean that that’s
why actually perhaps a lot of people thought – a lot of grandmasters thought b5
later was actually the more crucial mistake rather than h6 so even even fxe6
seems seems a move here Qe7 seems a
slightly inaccuracy compared to that with White actually being
slightly better technically according to this modern Houdini engine okay but b5
yes that is increasing White’s advantage technically in one move – it’s damaging
black’s solidity inviting line opening with a4 and c4 so that I believe is
the crucial one of one of the major crucial contributing factors to the
disaster black has here – b5 so we see now I think white really getting a huge
advantage pretty soon now after Kc8 maybe a more determined resistance was
b4 but after Kc8 we see line opening and then further line opening on
the cards after this Queen sack now c4 is very very powerful move here –
it’s creating a lot of loose pieces if Nb4 I think just Qxf5 then we’ve got some major threats like
Ne5 – keeping the pin here – well okay so
that that’s pretty nasty going into heart of black’s position hitting the bishop
and things like d5 on the cards Bd8 cxb5 Bxb5 – check .. check …
it’s blasting through now on these lines it starts to look very unpleasant starting to really tear black apart in
these variations so c4 is a very very powerful line opening move and now it
looks as though Re1 would be the real nail in the coffin in this position
okay so that was the classic game six which if you haven’t seen
before it’s like a it’s a very very significant event in man versus
computers back in 1997 comments or questions on YouTube – thanks very much.

56 thoughts on “Amazing Chess Game: Kasparov’s quickest defeat: IBM’s Deeper Blue (Computer) vs Garry Kasparov 1997

  1. I thouhgt the first game of this match where Kasparov saced the exchange to get a couple of passed pawns for the win was really interesting. I'd love it if you'd video annotate that some day. Just a suggestion

  2. Wow, the opening looks like something Karpov would play! Preparation against the tactically superior machine? Or did the aggressive Kasparov play this rock-solid system on a regular basis even against humans? Nice video, btw!

  3. In that particular game DB played a move that would have allowed kaspy to draw a lost game if he had seen it. That's where the controversy was as the machine it'self had other moves that did not allow for a draw and would not have missed it.

  4. Kasparov Playlist:
    Join me for a game:

  5. Kasparov never played this opening in serious match play and it shows. He made a mistake that a GM who regularly plays this would not make. Too bad as the match was even after 5 games and after this loss IBM dismantled the machine so Kasparov could not get revenge.

  6. Perhaps the most decisive blunder: ( this according to crafty 23.4), after 16. Qd3, the position is almost equal  ( -0.01). However, after 12. ..Bb7, black is at a disadvantage (-1.54), eg..  17. Bf5 exf5 18. Rxe7 Nxe7 19. Qc3 Nb6 20. Ne5 Kb7 21. Qa5 Ned5 22. Qa6+ Kc7 23. Ng6+ f4 24. Nxf4 Nxf4 25. Bxf4+ Kd7 26. Re1 g5 27. Bg3 Bd6 28. Bxd6 Kxd6  Crafty likes 12. ..Nc7 instead. 

  7. Kasparov obviously knew this line, as even I did when I was a kid.
    He merely mixed up the move order, simple as that.
    That said, I expect that eventually it will be demonstrated that black can survive
    the strategical mess of this line and a mass of theory will have to be memorized.
    Similar to the Fried Liver Attack in the Two Knight's Defense.

  8. black had a good counter attack at 9.06 could of attacked his queen then his roock then the main pwn got his peices out black would of had a good chance to win with all open peices

  9. I don't think it's fair that the computer has an opening library and endgame library. Those advantages are not the product of its 'thinking' but of its storage. The human should have access to a similar opening and endgame library. Also, as time goes by computers play more and more by brute-force calculation rather than strategic algorithms. There should be a limit to the computer's memory size. That would lead to a more interesting challenge. That is, can a computer beat a human by reasoning alone, without the advantage of pre-loaded solved positions.

  10. A flawed match. Gm 2 was very suspicious, as most know. Machine has opening library, doesn't get tired or frustrated, by their own admission IBM played psych warfare. Big deal – what did it really prove?

  11. I just tried to analyze the position with Stockfish 6, and it too goes very quickly to knight takes e6 as the preferred move (and keeps it so even after analyzing for many minutes and hundreds of millions of calculated nodes). The advantage it calculates for that move isn't very big (oscillates around +0.2), but it still considers it the best move for white.

  12. Still something in this game does not sit right with me. This was lazy chess by Kasparov. I still don't believe the initial blunder was unrecoverable at least to a draw. There is something strange here and it's not the "human move".

  13. Really . . . this is not an especially interesting computer game at all.  After Kasparov blundered by allowing the night sac at E6 (what was he thinking??), I could have cleaned up with White just as easily as the computer did . . .

  14. I see basically no computer engines that do either Ng5, or later Nxe6 from a bit of analysis with various engines. And was not surprised as seeing those moves, they are just uncharacteristically aggressive moves for a computer to do in an unclear situation. Computers generally only go aggressive when they have clear, forced tactically advantageous moves. Was this computer more advanced than basic engines as we have them now, given technology advance, probably not. So how was it making these moves? Yes Stockfish is one that seems to suggest Nxe6, but it straight away chooses it as if its already had that specific variation programmed into it, probably by its programmers being aware of this very game from the past. Interested in people's thoughts, if anyone has any inside knowledge to explain why this computer was making such aggressive moves in unclear situations?

  15. White has a several tricky knight sacs just 4-5 moves into the Modern variation. It's highly theoretical because of these hidden tactics which even grandmasters have fallen for.

    The crazy looking 5.Ng5!?, …h6 actually allows to 6.Ne6, which can't be taken any time soon or all sorts of mating patterns show up. Georgiev lost this way to John Nunn in Linares 1988.

  16. Please can you do the very controversial game 2 , which is obviously what got Kasparov so upset because he said that had to be a human move controlled by a human as the computer refused the offered up pawn , thanks

  17. Houdini 2 sucked so bad. Stockfish 8 nowadays sees Nxe6 very quickly and with an actual advantage for White. Not Houdini's equal eval … heh.

  18. Replayable game link:
    Join me or other Youtubers for a game: – Cheers, K

  19. How's Rook is inprisoned at H8?
    Pawn to H5 ….if white bishop(G6) take pawn then black knight (F7) take bishop at H5……a smooth exchange……what u say?

  20. Whatever is stored in computer memory….it just plays one move by one move. If u know ur counter play…it is tough for even stock fish to beat you easily.

  21. I believe I remember that the computer(s) team(s) having access to all of Kasparov's games whereas Kasparov was not given any access to the computer(s) games. This analysis really really helps the opponents. I remember Bobby Fischer saying he had all of Spasky' games & that information together with the fact that Bobby knew he was better than Boris at that time was the "icing on the cake". They should have provided Kasparov with that information. I think he would have done better by "getting into the head of those computers". Would you agree with this? Thanks. JAMES.

Leave a Reply

Your email address will not be published. Required fields are marked *