Would you like to inspect the original subtitles? These are the user uploaded subtitles that are being translated:
1
00:00:08,566 --> 00:00:11,566
hi everyone this is Grandmaster Eugene Polsky
2
00:00:12,033 --> 00:00:14,500
and welcome to today's lecture
3
00:00:15,100 --> 00:00:19,933
now today I will talk about something very interesting
4
00:00:20,066 --> 00:00:24,100
I'm gonna talk about my match with the strongest
5
00:00:24,633 --> 00:00:26,766
chess program in the world
6
00:00:27,166 --> 00:00:28,600
called Rybka
7
00:00:29,166 --> 00:00:30,800
over the Labor Day weekend
8
00:00:30,800 --> 00:00:33,166
I came down to Maryland
9
00:00:33,366 --> 00:00:38,200
and played the two game match against this monster
10
00:00:38,433 --> 00:00:41,600
it was running on 8 CPU
11
00:00:42,233 --> 00:00:44,166
8 processor system
12
00:00:44,266 --> 00:00:50,366
and Larry Kaufman was kind enough to offer this match
13
00:00:51,600 --> 00:00:54,100
and uh I thought that was
14
00:00:54,100 --> 00:00:55,833
there would be a great opportunity for me
15
00:00:55,833 --> 00:00:57,966
both to practice against the strongest
16
00:00:58,300 --> 00:01:00,000
chess engine in the world
17
00:01:00,066 --> 00:01:02,466
and also just to see with my own eyes
18
00:01:02,466 --> 00:01:05,766
you know how powerful computers really are
19
00:01:05,766 --> 00:01:10,100
especially Rybka running on 8 8 way processor machine
20
00:01:11,600 --> 00:01:12,866
so
21
00:01:14,700 --> 00:01:16,633
I played the two game match
22
00:01:16,633 --> 00:01:20,100
and the first game I was white
23
00:01:20,500 --> 00:01:25,333
and I managed to lose pretty much a one sided game
24
00:01:25,633 --> 00:01:27,700
I had some chances in the opening
25
00:01:27,700 --> 00:01:28,600
but basically
26
00:01:29,633 --> 00:01:32,300
I was unfamiliar with the opening a little bit
27
00:01:32,433 --> 00:01:36,066
and Rybka played truly flawlessly
28
00:01:36,700 --> 00:01:39,733
according to her 3,000 or so rating
29
00:01:40,766 --> 00:01:43,000
and the second game I believe
30
00:01:43,000 --> 00:01:43,966
I finally
31
00:01:44,200 --> 00:01:48,333
was able to find one single weakness of the computer
32
00:01:48,466 --> 00:01:51,200
and you're gonna be able to see what what it is
33
00:01:51,600 --> 00:01:54,266
and just a little background for you
34
00:01:55,000 --> 00:01:58,133
if you are not quite familiar with what Rybka is
35
00:01:58,766 --> 00:02:02,866
Rybka is the strongest chess engine in the world
36
00:02:03,033 --> 00:02:05,766
it's probably much stronger than Fritz these days
37
00:02:05,766 --> 00:02:08,133
especially the latest version Rybka 3
38
00:02:08,900 --> 00:02:10,833
now Ripka as you know
39
00:02:10,833 --> 00:02:13,600
have played many matches against grandmasters
40
00:02:13,866 --> 00:02:16,800
and to a lot of grandmasters
41
00:02:16,800 --> 00:02:19,466
Ripka gave pawn and move odd
42
00:02:19,466 --> 00:02:21,300
so grandmaster would be white
43
00:02:21,300 --> 00:02:22,900
and Ripka would be down a pawn
44
00:02:24,433 --> 00:02:26,633
and most of the matches Rybka won
45
00:02:26,633 --> 00:02:31,900
except one match Rybka drew against my coach
46
00:02:31,900 --> 00:02:33,300
Roman Janjgashvili
47
00:02:33,900 --> 00:02:39,333
but this time we decided to play on equal ground
48
00:02:39,900 --> 00:02:42,700
so this is gonna be completely even
49
00:02:42,700 --> 00:02:46,366
there are no pawns there are no you know exchange
50
00:02:46,433 --> 00:02:48,600
it's just simply chess
51
00:02:49,100 --> 00:02:53,366
except I'm gonna have 90 minutes plus 30 seconds
52
00:02:53,366 --> 00:02:55,366
which is the normal fide time control
53
00:02:55,466 --> 00:02:58,133
and Rybka is gonna have 10 times less
54
00:02:58,233 --> 00:03:00,100
9 minutes and 3 seconds
55
00:03:00,666 --> 00:03:02,433
now according to Larry Coffman
56
00:03:02,433 --> 00:03:07,400
he still thinks that Rybka plays about 3,000 rating
57
00:03:07,400 --> 00:03:09,066
which is unbelievable rating
58
00:03:09,066 --> 00:03:10,900
stronger than any world champion
59
00:03:11,400 --> 00:03:12,466
dead or alive
60
00:03:13,066 --> 00:03:14,266
so imagine
61
00:03:14,433 --> 00:03:16,166
how it feels like
62
00:03:16,166 --> 00:03:18,966
playing against the strongest entity in the world
63
00:03:19,966 --> 00:03:22,266
so let's see what happens in the second game
64
00:03:22,266 --> 00:03:23,166
I'm white
65
00:03:23,766 --> 00:03:25,966
so I play knight of three
66
00:03:27,233 --> 00:03:29,766
in the first game I played D4
67
00:03:29,766 --> 00:03:34,600
and went into one of the variations of the Nimzo
68
00:03:34,600 --> 00:03:39,800
Indian with G3 but now I decided to kind of play solid
69
00:03:40,266 --> 00:03:45,366
and not to go for forced positions
70
00:03:46,233 --> 00:03:47,800
so D5
71
00:03:49,866 --> 00:03:52,200
alright so now again I have a choice
72
00:03:52,433 --> 00:03:56,266
do I play G3 C4 or D4
73
00:03:56,766 --> 00:03:57,800
and again
74
00:03:57,800 --> 00:04:02,733
if I play G3 giving up the center to the strongest
75
00:04:04,166 --> 00:04:06,000
chess computer in the world
76
00:04:06,066 --> 00:04:08,133
it's probably not a great idea
77
00:04:09,900 --> 00:04:11,566
and by the way the expectations
78
00:04:11,566 --> 00:04:13,266
of me going into this match was
79
00:04:13,266 --> 00:04:14,866
I was gonna lose two games
80
00:04:16,166 --> 00:04:19,666
but I managed to draw the second game
81
00:04:19,833 --> 00:04:22,100
and actually
82
00:04:22,966 --> 00:04:27,133
discovered quite an interesting weakness of Rybka
83
00:04:27,166 --> 00:04:30,600
so I played D4 fighting for the center
84
00:04:32,033 --> 00:04:35,366
knight of 6 c 4
85
00:04:36,833 --> 00:04:40,833
e 6 and now e 3 this is not you know
86
00:04:40,833 --> 00:04:43,266
very aggressive way to fight for advantage
87
00:04:43,266 --> 00:04:46,033
if you're white but again
88
00:04:46,033 --> 00:04:48,566
my goal is to play solid
89
00:04:49,033 --> 00:04:51,366
try not to go into crazy positions
90
00:04:51,366 --> 00:04:54,866
because I'm playing a 3,000 rated opponent
91
00:04:55,166 --> 00:04:58,066
who is virtually almost like god
92
00:04:58,066 --> 00:04:59,233
in open positions
93
00:04:59,233 --> 00:05:02,766
there is no mistake that I can expect from rip
94
00:05:02,766 --> 00:05:05,900
my only chance is to try to get some kind of closed
95
00:05:05,966 --> 00:05:07,700
or semi closed positions
96
00:05:08,200 --> 00:05:10,800
where human understanding and experience
97
00:05:11,233 --> 00:05:12,866
will be dominating
98
00:05:14,300 --> 00:05:17,800
compared to sheer calculation and
99
00:05:17,800 --> 00:05:19,733
and brute force by the computer
100
00:05:21,100 --> 00:05:23,966
so I play e 3 bishop e 7
101
00:05:24,800 --> 00:05:27,566
so all of this you can say is book
102
00:05:27,566 --> 00:05:30,100
so Rybka has its own book especially
103
00:05:31,566 --> 00:05:34,900
constructed against me and now again
104
00:05:34,900 --> 00:05:40,400
I play a very solid knight b d two of course
105
00:05:40,400 --> 00:05:44,366
I can play night c 3 which is totally normal move
106
00:05:44,666 --> 00:05:46,433
but I wanted to
107
00:05:46,433 --> 00:05:49,900
deviate as much as I can from the mainstream theory
108
00:05:50,300 --> 00:05:53,933
and my goal is simply to develop my light square bishop
109
00:05:54,300 --> 00:05:56,866
castles and play b 3
110
00:05:56,866 --> 00:05:58,666
bishop b 2 kind of simple
111
00:05:59,000 --> 00:06:01,433
easy development especially
112
00:06:01,433 --> 00:06:04,300
this is for those D4 players who really
113
00:06:04,600 --> 00:06:06,100
don't wanna learn any theory
114
00:06:06,100 --> 00:06:09,966
this is practically the easiest way to develop
115
00:06:11,600 --> 00:06:13,766
alright so computer plays castles
116
00:06:14,433 --> 00:06:20,133
so far Rybka spent zero time on the moves
117
00:06:20,233 --> 00:06:21,966
because this is all book
118
00:06:22,866 --> 00:06:25,366
here I had the first dilemma of the game
119
00:06:26,000 --> 00:06:31,100
should I play Bishop e 2 or Bishop d 3 or even b three
120
00:06:33,033 --> 00:06:35,800
and again I'm not trying to play according to theory
121
00:06:35,800 --> 00:06:39,400
I'm just simply trying to play chess right now
122
00:06:39,400 --> 00:06:40,500
against computers
123
00:06:41,266 --> 00:06:44,133
closing the position is recommended
124
00:06:44,300 --> 00:06:48,366
so C5 kind of went through my mind
125
00:06:49,400 --> 00:06:51,666
but the problem with C5 is the phone
126
00:06:51,666 --> 00:06:53,500
black simply plays B6
127
00:06:54,466 --> 00:06:56,733
B4 and A5
128
00:06:58,066 --> 00:07:00,400
and I don't have the time to play a 3
129
00:07:00,400 --> 00:07:03,800
because my rook on a 1 is not defended
130
00:07:04,066 --> 00:07:07,500
and basically my queen side falls apart
131
00:07:07,500 --> 00:07:11,333
and so does my position so see if I was too premature
132
00:07:12,066 --> 00:07:14,700
and I play bishop to d three
133
00:07:16,800 --> 00:07:19,066
and now Ripka play C5
134
00:07:20,700 --> 00:07:24,100
Castles Night C6
135
00:07:26,300 --> 00:07:28,500
so here I really
136
00:07:29,700 --> 00:07:31,700
didn't like c t x
137
00:07:31,700 --> 00:07:34,000
d coming up with idea to
138
00:07:34,233 --> 00:07:37,200
basically have an isolated pawn on d four
139
00:07:37,633 --> 00:07:39,366
and my knight on d
140
00:07:39,366 --> 00:07:41,866
2 is not well placed in these types of positions
141
00:07:42,233 --> 00:07:44,800
so I don't really wanna get an isolated pawn
142
00:07:44,866 --> 00:07:47,600
I wanna keep my pawn structure nice and fluid
143
00:07:48,000 --> 00:07:50,133
so I take on d 5 myself
144
00:07:51,066 --> 00:07:54,300
and Rika plays the unusual looking queen
145
00:07:54,300 --> 00:07:57,566
takes d five again with zero
146
00:07:59,033 --> 00:08:00,800
time spent on her clock
147
00:08:01,800 --> 00:08:04,333
this is all preparation so far
148
00:08:06,000 --> 00:08:07,466
now if pawn takes
149
00:08:07,466 --> 00:08:09,600
then I have a very comfortable position
150
00:08:09,600 --> 00:08:12,666
after pawn takes c 5 Bishop takes
151
00:08:13,200 --> 00:08:17,566
and now either night b 3 going after the d 4 square
152
00:08:18,000 --> 00:08:20,766
or b 3 with a bishop b 2
153
00:08:23,500 --> 00:08:25,766
white has a very easy play
154
00:08:25,766 --> 00:08:28,866
and he's gonna play against the isolated pawn on D4
155
00:08:29,166 --> 00:08:33,666
so these types of positions are pretty solid
156
00:08:35,066 --> 00:08:38,066
and yet black probably would have played this
157
00:08:38,066 --> 00:08:39,566
if this was a human game
158
00:08:40,666 --> 00:08:44,866
so Queen takes d 5 is a big surprise to me immediately
159
00:08:45,466 --> 00:08:46,666
and
160
00:08:47,400 --> 00:08:49,866
the move that I was actually thinking here
161
00:08:49,866 --> 00:08:51,800
for a long time is E4
162
00:08:53,366 --> 00:08:55,100
because why don't you you know
163
00:08:55,100 --> 00:08:56,833
the thought of going through my mind is
164
00:08:56,833 --> 00:08:58,466
why don't I attack the queen
165
00:08:59,100 --> 00:09:00,900
get my opponent to the center
166
00:09:01,066 --> 00:09:03,800
and now if the queen goes to h 5
167
00:09:03,966 --> 00:09:06,466
I can drop back to e 2 with my bishop
168
00:09:07,166 --> 00:09:10,200
threatening knight e 5 and hunting down the queen
169
00:09:10,766 --> 00:09:12,200
and I looked at this variation
170
00:09:12,200 --> 00:09:13,133
pawn takes
171
00:09:13,700 --> 00:09:17,100
Knight takes queen e 5 knight c 6
172
00:09:17,100 --> 00:09:21,900
pawn takes and now black does have a weak pawn on c 6
173
00:09:22,166 --> 00:09:26,300
but my knight on d 2 has to defend my pawn on e four
174
00:09:27,100 --> 00:09:29,533
and I was a little bit unsure
175
00:09:29,633 --> 00:09:31,300
if this is the good position
176
00:09:32,200 --> 00:09:33,966
to have against the computer
177
00:09:36,766 --> 00:09:37,866
so
178
00:09:41,833 --> 00:09:45,733
probably I made the correct decision now
179
00:09:45,766 --> 00:09:49,100
and I play Bishop to C4
180
00:09:51,033 --> 00:09:51,266
again
181
00:09:51,266 --> 00:09:54,433
my job is not to go for some kind of complications
182
00:09:54,433 --> 00:09:55,733
because I am playing
183
00:09:56,266 --> 00:09:58,100
the computer that can see
184
00:09:58,600 --> 00:10:01,166
millions of positions per second
185
00:10:01,166 --> 00:10:04,166
this is really not a match
186
00:10:04,166 --> 00:10:06,066
in terms of calculation ability
187
00:10:06,300 --> 00:10:09,000
but my goal is simply to develop my pieces
188
00:10:09,166 --> 00:10:11,100
and try to get a comfortable game
189
00:10:12,366 --> 00:10:16,366
and Ripka plays which is probably the strongest move
190
00:10:16,366 --> 00:10:21,600
queen f 5 and immediate question for you
191
00:10:21,633 --> 00:10:26,933
can white go e four and win a piece after bishop Z 3
192
00:10:28,300 --> 00:10:30,066
well don't forget
193
00:10:30,066 --> 00:10:32,100
you're playing against the strongest
194
00:10:33,100 --> 00:10:36,566
computer in the world and now night X
195
00:10:36,566 --> 00:10:43,066
F2 is the tactic that I had to find over the board
196
00:10:43,166 --> 00:10:45,100
so this is not good for white
197
00:10:45,633 --> 00:10:48,366
if I take knight takes queen
198
00:10:48,633 --> 00:10:49,800
if bishop retreats
199
00:10:49,800 --> 00:10:53,400
Knight goes to e 3 and black is totally winning
200
00:10:54,700 --> 00:11:00,433
a bishop h 7 king takes rook takes CD again
201
00:11:00,433 --> 00:11:01,866
white is totally lost
202
00:11:02,400 --> 00:11:06,933
so basically bishop d 3 and e 4 doesn't work
203
00:11:09,000 --> 00:11:10,300
alright so
204
00:11:12,100 --> 00:11:14,066
how do I do all my pieces down
205
00:11:14,066 --> 00:11:15,733
that's a good question to ask
206
00:11:16,833 --> 00:11:19,133
and again I try to play simple
207
00:11:19,300 --> 00:11:25,466
I take on C5 and I'll play an a B3 preparing bishop d 2
208
00:11:25,466 --> 00:11:30,300
bishop c 3 in the future and also possibly
209
00:11:30,300 --> 00:11:32,666
I am the d 4 square for one of my knights
210
00:11:33,566 --> 00:11:36,266
rook d 8 queen a 2
211
00:11:36,666 --> 00:11:39,700
bishop d 6 now by the way
212
00:11:40,866 --> 00:11:43,266
Ripka started playing on her own
213
00:11:43,266 --> 00:11:45,266
after I played Bishop C4
214
00:11:45,266 --> 00:11:48,566
which is a novelty in her book
215
00:11:48,900 --> 00:11:52,400
so now I'm actually playing the computer
216
00:11:53,033 --> 00:11:57,400
without any type of databases or opening theory
217
00:11:59,466 --> 00:12:02,366
and looking at this position
218
00:12:03,100 --> 00:12:07,733
it's not so clear how white can try to
219
00:12:07,766 --> 00:12:11,266
you know fight for advantage but you know what
220
00:12:11,400 --> 00:12:12,666
I'm not even concerned
221
00:12:12,666 --> 00:12:14,233
about trying to fight for advantage
222
00:12:14,233 --> 00:12:17,566
I'm looking for maintaining Equality
223
00:12:17,766 --> 00:12:21,700
which is in itself very hard against computers
224
00:12:22,233 --> 00:12:24,500
the human kind is doomed
225
00:12:24,766 --> 00:12:28,966
that world champion can no longer win against computers
226
00:12:29,266 --> 00:12:31,133
maximum we can do is draw
227
00:12:31,466 --> 00:12:33,066
so if we draw one of the games
228
00:12:33,066 --> 00:12:37,300
this is already a huge huge success for humans
229
00:12:39,100 --> 00:12:41,133
so now let's think
230
00:12:43,066 --> 00:12:45,600
what to do with the bishop on c 1
231
00:12:46,000 --> 00:12:49,000
well bishop d 2 is possible
232
00:12:49,000 --> 00:12:52,466
but I really didn't like either knight e 4
233
00:12:53,766 --> 00:12:55,733
or possible e five
234
00:12:57,833 --> 00:13:00,133
so that's why I play rookie one
235
00:13:00,200 --> 00:13:02,166
with idea to play e 4 myself
236
00:13:02,800 --> 00:13:04,366
followed by bishop e 3
237
00:13:05,700 --> 00:13:08,966
now Rybka simply drops back to g 6 with the queen
238
00:13:09,700 --> 00:13:13,200
setting up a little trap if I play e 4
239
00:13:13,466 --> 00:13:15,200
the knight to g four
240
00:13:16,600 --> 00:13:18,900
and I really have nothing better to do
241
00:13:18,900 --> 00:13:20,700
than play rook f 1
242
00:13:22,700 --> 00:13:25,733
and now of course rook e 1 doesn't make any sense
243
00:13:26,233 --> 00:13:29,366
I just simply lose time if I play bishop e 3
244
00:13:29,633 --> 00:13:31,333
this is just simply bad
245
00:13:31,466 --> 00:13:35,466
I get to double up my pawns and then I lose d 4 pawn
246
00:13:35,900 --> 00:13:37,900
so rook f 1 is the only move
247
00:13:38,233 --> 00:13:41,366
not something that I wanna play
248
00:13:42,366 --> 00:13:44,266
so after queen g six
249
00:13:45,666 --> 00:13:48,200
my best hope is try to
250
00:13:49,766 --> 00:13:51,966
attack the queen hopefully
251
00:13:51,966 --> 00:13:53,800
the queen will leave a good square on g 6
252
00:13:53,800 --> 00:13:54,966
with knight h four
253
00:13:56,033 --> 00:13:59,800
Ripka plays queen h 6 and now night of three back
254
00:14:00,233 --> 00:14:03,300
I'm basically saying I don't mind the repetition of
255
00:14:03,300 --> 00:14:04,466
of the position
256
00:14:05,300 --> 00:14:08,133
because draw against such a powerful beast
257
00:14:08,366 --> 00:14:10,133
is a victory for me
258
00:14:11,000 --> 00:14:17,833
and of course now I wanna play E4 and Rybka says I
259
00:14:17,833 --> 00:14:22,200
I really don't wanna repeat the position
260
00:14:22,400 --> 00:14:23,700
and I wanna win
261
00:14:24,466 --> 00:14:29,733
so the next move is very strong Knight to G four
262
00:14:32,466 --> 00:14:37,166
so if I get to play H3 which is what I played
263
00:14:37,233 --> 00:14:40,100
the knight gets to reposition to E5
264
00:14:40,866 --> 00:14:43,366
so far there are no fancy tactics
265
00:14:43,366 --> 00:14:47,100
but Rybka is outplaying me little by little
266
00:14:47,166 --> 00:14:51,400
because now I get to lose my B5 bishop
267
00:14:52,266 --> 00:14:54,700
and as we all know the bishop pair
268
00:14:54,700 --> 00:14:57,100
is slightly better than bishop and knight
269
00:14:57,466 --> 00:15:02,733
especially in symmetrical half open you know positions
270
00:15:03,400 --> 00:15:06,866
and again I have to kind of grind my teeth
271
00:15:06,966 --> 00:15:08,866
and look for possibilities
272
00:15:09,366 --> 00:15:12,600
to equalize and make a draw
273
00:15:15,000 --> 00:15:15,466
of course
274
00:15:15,466 --> 00:15:18,566
bishop b 5 now is not such a great possibility
275
00:15:18,566 --> 00:15:21,400
after a 6 bishop a four
276
00:15:22,166 --> 00:15:24,133
my bishop is totally dead
277
00:15:24,233 --> 00:15:28,733
the g 3 square is weak and this is really bad position
278
00:15:31,433 --> 00:15:34,566
so I play bishop d 2
279
00:15:35,166 --> 00:15:37,800
knowing that I have to part with my very nice
280
00:15:37,833 --> 00:15:39,100
light square bishop
281
00:15:39,766 --> 00:15:43,466
knight takes queen takes and yet I'm pretty optimistic
282
00:15:43,466 --> 00:15:45,800
if I get my bishop on c three
283
00:15:47,266 --> 00:15:50,100
I had I still maintain a very solid position
284
00:15:50,400 --> 00:15:53,333
I'm sure I would do well against the grandmaster
285
00:15:53,633 --> 00:15:55,133
but against computer
286
00:15:55,433 --> 00:15:57,366
whenever computer has a slight edge
287
00:15:57,366 --> 00:15:59,833
such as the 2 bishops it is very
288
00:15:59,833 --> 00:16:01,900
very hard to hold
289
00:16:01,900 --> 00:16:04,900
because computers never make mistakes
290
00:16:04,900 --> 00:16:06,133
so think about it
291
00:16:07,000 --> 00:16:08,466
for the duration of the game
292
00:16:08,466 --> 00:16:10,566
I have to play near perfect
293
00:16:10,566 --> 00:16:14,500
if I make one mistake in a slightly inferior position
294
00:16:14,766 --> 00:16:18,300
I'm basically gonna lose so this is very hard
295
00:16:18,300 --> 00:16:21,866
psychological battle that every human faces
296
00:16:22,266 --> 00:16:23,266
against computers
297
00:16:25,166 --> 00:16:26,700
alright so let's keep going
298
00:16:27,500 --> 00:16:31,700
so now Rebka wants to open up the position so she plays
299
00:16:33,266 --> 00:16:37,366
e 5 opening up the light square bishop
300
00:16:37,866 --> 00:16:42,100
bishop c 3 hitting the e 5 pawn and queen g 6
301
00:16:42,766 --> 00:16:44,900
now if I take the pawn on e 5
302
00:16:44,900 --> 00:16:47,400
Bishop takes h 3 and again
303
00:16:47,400 --> 00:16:49,600
this is not something I wanted to do
304
00:16:50,800 --> 00:16:56,300
I wanted to defend my h h 3 pawn I play queen to h four
305
00:16:58,666 --> 00:17:01,400
f 6 and now a 4
306
00:17:01,866 --> 00:17:04,100
with idea to play a 5
307
00:17:04,433 --> 00:17:07,766
and try to get some counterplay on the queenside
308
00:17:08,766 --> 00:17:10,066
bishop e 6
309
00:17:10,600 --> 00:17:13,633
so finally we get to the first test
310
00:17:13,633 --> 00:17:15,366
where we can test the computer
311
00:17:16,666 --> 00:17:19,533
opposite color bishop endgame down a pawn
312
00:17:19,866 --> 00:17:21,733
so I play a 5
313
00:17:21,833 --> 00:17:25,866
trying to lure the computer into taking my knight
314
00:17:26,433 --> 00:17:27,666
if bishop takes pawn
315
00:17:27,666 --> 00:17:30,900
takes pawn takes black is up a pawn
316
00:17:32,366 --> 00:17:35,100
which is a big advantage in computer chess
317
00:17:35,700 --> 00:17:38,600
because computers are very materialistic
318
00:17:38,800 --> 00:17:41,366
but Rybka is an advanced kind
319
00:17:42,000 --> 00:17:46,800
this is if you wanna compare Terminator 1
320
00:17:47,100 --> 00:17:50,866
this is Terminator 2 type of position
321
00:17:51,166 --> 00:17:53,166
so Rybka understands that
322
00:17:53,166 --> 00:17:55,100
this is the next generation chess
323
00:17:55,600 --> 00:17:58,200
opposite color bishops are likely drawn
324
00:17:58,600 --> 00:18:00,166
extra pawn means nothing
325
00:18:00,166 --> 00:18:02,733
especially this is a double pawn
326
00:18:03,233 --> 00:18:04,533
b 6 b 7
327
00:18:04,866 --> 00:18:07,066
and Rybka makes the first correct choice
328
00:18:07,066 --> 00:18:08,800
that every human would make
329
00:18:09,000 --> 00:18:13,766
is to not get Learned by this material advantage
330
00:18:13,833 --> 00:18:15,400
and she plays Bishop C7
331
00:18:15,966 --> 00:18:19,933
so you see this is the power of modern computer chess
332
00:18:20,633 --> 00:18:22,866
five or maybe 10 years ago
333
00:18:23,600 --> 00:18:26,500
almost every single computer would go for the pawn
334
00:18:26,966 --> 00:18:29,700
and I would make a draw without any problems
335
00:18:30,166 --> 00:18:32,600
but now I really have to suffer
336
00:18:32,833 --> 00:18:36,733
because the bishop pair still dominate I did too
337
00:18:39,700 --> 00:18:41,200
well actually one second
338
00:18:43,233 --> 00:18:45,800
she even played a better move than Bishop's C7
339
00:18:47,166 --> 00:18:49,566
she played bishop to d 5 first
340
00:18:49,700 --> 00:18:51,700
forcing the exchange of Queens
341
00:18:52,233 --> 00:18:55,866
she could have played bishop c 7 and then knight d 2
342
00:18:56,066 --> 00:18:59,266
bishop e 4 but this is a much bishop d 4
343
00:18:59,266 --> 00:19:01,166
I mean bishop d 5 sorry
344
00:19:01,166 --> 00:19:02,766
but this is a much better way
345
00:19:02,766 --> 00:19:08,766
because now she's threatening to take on G2
346
00:19:10,000 --> 00:19:12,266
so you see this is very kinetic
347
00:19:15,566 --> 00:19:17,100
now I play
348
00:19:19,400 --> 00:19:20,733
Queen G4
349
00:19:22,300 --> 00:19:28,300
when take g 4 pawn takes bishop c 7 and I d 2
350
00:19:28,566 --> 00:19:32,200
so this is the main position
351
00:19:32,566 --> 00:19:36,200
so let us think if black has any chance to win
352
00:19:43,166 --> 00:19:45,633
of course black has chances to win
353
00:19:45,633 --> 00:19:47,366
because black has a bishop pair
354
00:19:47,466 --> 00:19:51,700
all black has to do is try to create pressure
355
00:19:51,700 --> 00:19:53,900
and open up the position for the bishops
356
00:19:54,233 --> 00:19:58,166
the only problem is I'm just way too solid
357
00:19:58,166 --> 00:20:01,166
I'm gonna play f 3 king f 2
358
00:20:01,500 --> 00:20:03,933
and place my knight on e 4
359
00:20:04,100 --> 00:20:08,366
because I'm not afraid to have double isolated pawn
360
00:20:08,833 --> 00:20:12,500
and this is the weakness that computers have
361
00:20:12,500 --> 00:20:16,733
even Rybka is that after king f 7
362
00:20:17,966 --> 00:20:20,566
f 3 rook to 8
363
00:20:22,566 --> 00:20:23,866
king f two
364
00:20:26,900 --> 00:20:34,366
Ripka plays now king e 6 rook f d 1 Bishop d 6
365
00:20:34,366 --> 00:20:38,800
Knight e 4 this is the test of time
366
00:20:39,000 --> 00:20:43,000
computers cannot avoid taking the night
367
00:20:43,000 --> 00:20:45,933
they just think this is too lucrative or possibility
368
00:20:46,500 --> 00:20:48,800
and this is the mistake
369
00:20:49,800 --> 00:20:53,100
well maybe not a mistake but this is an inaccuracy
370
00:20:54,833 --> 00:20:59,400
that rip Comix in the game after Point X
371
00:21:00,200 --> 00:21:01,966
this is very easy to hold
372
00:21:01,966 --> 00:21:05,133
you can almost see that this is gonna be fortress
373
00:21:05,400 --> 00:21:07,700
and computers really don't understand fortresses
374
00:21:07,700 --> 00:21:08,800
that well yet
375
00:21:09,100 --> 00:21:11,766
and this is what I mean by I discovered
376
00:21:12,433 --> 00:21:15,400
still there is chances for humans
377
00:21:15,400 --> 00:21:17,866
because computers don't understand fortresses
378
00:21:19,600 --> 00:21:21,733
now of course much better move
379
00:21:22,200 --> 00:21:24,433
from a perspective of complicating the game
380
00:21:24,433 --> 00:21:25,866
and trying to play for win
381
00:21:25,900 --> 00:21:27,133
is bishop to f eight
382
00:21:27,966 --> 00:21:28,500
now
383
00:21:28,500 --> 00:21:31,700
I have little doubt that I would be able to hold this
384
00:21:31,700 --> 00:21:34,366
in the long term because slowly
385
00:21:34,366 --> 00:21:37,766
slowly I have to make the best moves
386
00:21:37,766 --> 00:21:40,900
and yet computer will probably try to play for win
387
00:21:40,900 --> 00:21:45,300
in like 30 moves 40 moves I have to play correctly
388
00:21:45,600 --> 00:21:47,200
but I probably
389
00:21:47,433 --> 00:21:50,633
objective evaluation of this is that with best play
390
00:21:50,633 --> 00:21:52,666
white should hold nevertheless
391
00:21:53,800 --> 00:21:58,266
but let us think why computers
392
00:21:58,266 --> 00:22:02,000
computer cannot win this position so Bishop C5
393
00:22:04,500 --> 00:22:05,766
king e 2
394
00:22:07,433 --> 00:22:09,300
and then let's say we trade rook
395
00:22:10,300 --> 00:22:12,900
so starting from this move
396
00:22:12,900 --> 00:22:15,100
computer started to maneuver a lot
397
00:22:15,100 --> 00:22:17,566
without really showing a winning plan
398
00:22:18,466 --> 00:22:20,033
and if you think about it
399
00:22:20,033 --> 00:22:21,600
how is black gonna win
400
00:22:22,000 --> 00:22:25,066
black can try to move his king to G6 and G5
401
00:22:25,066 --> 00:22:27,100
and I'm simply gonna play king of three
402
00:22:27,966 --> 00:22:29,900
if black trades a pair of rooks
403
00:22:30,266 --> 00:22:32,133
how is black gonna make progress
404
00:22:32,633 --> 00:22:36,900
so we basically kept moving for about I don't know
405
00:22:36,900 --> 00:22:40,000
30 or more moves back and forth
406
00:22:40,233 --> 00:22:42,466
until Rybka started to make some progress
407
00:22:42,800 --> 00:22:46,366
she played a six and then we traded rooks
408
00:22:46,633 --> 00:22:50,366
and I wanna show you the final position
409
00:22:50,600 --> 00:22:53,266
where Rebekah still thinks she is winning
410
00:22:53,566 --> 00:22:55,700
but it's clear to all humans
411
00:22:56,033 --> 00:22:58,300
that the position is dead drawn
412
00:22:59,233 --> 00:23:02,800
so let's just make some some random moves here
413
00:23:05,666 --> 00:23:09,500
so basically we ended up trading rooks
414
00:23:11,166 --> 00:23:13,900
um then Ripka
415
00:23:14,000 --> 00:23:15,933
so I kind of go back and forth
416
00:23:16,233 --> 00:23:17,933
Ripka transferred the king
417
00:23:20,066 --> 00:23:25,433
to G6 and at some moment she played B5
418
00:23:25,433 --> 00:23:26,033
B4
419
00:23:26,033 --> 00:23:32,300
which doesn't really make a difference and played H6
420
00:23:34,100 --> 00:23:38,933
king G5 so if you put it on any engine in the world
421
00:23:39,033 --> 00:23:42,100
it's gonna show that black is almost up a point
422
00:23:42,866 --> 00:23:45,000
so this is really remarkable
423
00:23:45,000 --> 00:23:48,500
because if you put this position in front of any Class
424
00:23:48,500 --> 00:23:49,500
a player
425
00:23:50,466 --> 00:23:54,200
then it's easy to see that black cannot break through
426
00:23:55,266 --> 00:23:56,033
the best black
427
00:23:56,033 --> 00:23:59,133
and hope for is to break through with G6 and F5
428
00:23:59,433 --> 00:24:00,966
but this doesn't achieve anything
429
00:24:00,966 --> 00:24:02,600
because my E3 point is gonna come
430
00:24:02,600 --> 00:24:03,566
to E4
431
00:24:04,400 --> 00:24:05,366
so for example
432
00:24:05,366 --> 00:24:07,900
let's just say I go back and forth just to show you
433
00:24:08,200 --> 00:24:10,200
now my bishop does have to be on d 2
434
00:24:10,200 --> 00:24:11,566
let's say I go here
435
00:24:12,466 --> 00:24:13,800
bishop here bishop there
436
00:24:13,800 --> 00:24:17,233
f 5 takes takes takes
437
00:24:17,233 --> 00:24:22,900
takes e 4 and again this is a dead draw position
438
00:24:24,400 --> 00:24:25,866
so you see
439
00:24:26,800 --> 00:24:29,100
this is a real problem in computer chess
440
00:24:29,100 --> 00:24:31,366
computers don't see far ahead
441
00:24:31,900 --> 00:24:36,133
because they don't understand that this is a fortress
442
00:24:36,633 --> 00:24:39,933
and the evaluation is not quite correct
443
00:24:40,100 --> 00:24:42,700
and the whole idea with bishop takes knight
444
00:24:42,900 --> 00:24:44,366
doubling up my pawns
445
00:24:44,833 --> 00:24:47,500
is the wrong plan if you're playing for a win
446
00:24:49,033 --> 00:24:52,066
and after about hundred and thirty moves
447
00:24:52,066 --> 00:24:54,000
of being tortured by Ripka
448
00:24:54,866 --> 00:24:59,666
Larry Kaufman made a very good judgment call and he
449
00:24:59,966 --> 00:25:01,466
he called this a draw
450
00:25:01,833 --> 00:25:07,200
and so the game was drawn now of course
451
00:25:09,033 --> 00:25:10,766
playing for you know
452
00:25:10,766 --> 00:25:12,000
49 moves
453
00:25:12,000 --> 00:25:14,266
moving back and forth with the bishop and king
454
00:25:14,366 --> 00:25:16,133
and then on 50th move
455
00:25:16,666 --> 00:25:19,066
Rybka plays a pawn move you know
456
00:25:19,066 --> 00:25:21,700
this is not a good spectator chess
457
00:25:22,033 --> 00:25:24,300
so programmers still have to work on
458
00:25:24,300 --> 00:25:25,133
you know
459
00:25:25,700 --> 00:25:28,700
very good techniques to try to avoid that
460
00:25:29,200 --> 00:25:30,666
but in general
461
00:25:31,300 --> 00:25:33,300
I was very very
462
00:25:33,600 --> 00:25:35,566
amazed how powerful Rybka is
463
00:25:35,566 --> 00:25:37,533
in calculation type positions
464
00:25:37,800 --> 00:25:42,400
and overall general play is almost
465
00:25:42,800 --> 00:25:46,033
you know any world champion type of quality
466
00:25:46,033 --> 00:25:47,033
probably even higher
467
00:25:47,033 --> 00:25:48,933
because Rybka never makes mistakes
468
00:25:49,366 --> 00:25:51,266
but in easy simple
469
00:25:51,633 --> 00:25:53,066
closed positions
470
00:25:53,366 --> 00:25:57,200
which are like fortresses like my game
471
00:25:57,766 --> 00:26:00,000
Rybka still has a lot to learn
472
00:26:00,466 --> 00:26:05,333
so at least humans still have a chance to make a draw
473
00:26:05,666 --> 00:26:08,000
or maybe even try to play for win
474
00:26:08,266 --> 00:26:11,300
if strategic ideas are very
475
00:26:11,300 --> 00:26:13,933
very long term in a closed position
476
00:26:14,433 --> 00:26:16,366
so I hope you enjoy this lecture
477
00:26:17,433 --> 00:26:19,766
and I talk to you later bye bye
32887
Can't find what you're looking for?
Get subtitles in any language from opensubtitles.com, and translate them here.