11 tactical positions computers can't solve

389 views
Skip to first unread message

Marc-Francois Baudot

unread,
Jul 11, 1994, 5:56:28 AM7/11/94
to
It seems the positions didn't come through so here they are :


# Position: 1
# Move: W

r...qb.k
.b....p.
p..pr..p
...n....
Pnp.N.N.
......RP
.B...PP.
.B.QR.K.

# Position: 2
# Move: W

r....rk.
pp.n.p.p
.nqP..p.
..b.P.B.
....NQ..
.B...P..
PP..K..P
..R.....

# Position: 3
# Move: W

r..qk..r
ppp.b.pp
..n.p...
...pP.n.
...P..b.
..PB.NN.
PP....PP
R.BQK..R

# Position: 4
# Move: W

r.b.kb.r
.p.n.ppp
p..ppn..
......BB
..qNP...
..N.....
PPP..PPP
R..Q.RK.

# Position: 5
# Move: W

r..qrb.k
.p.b..p.
p..ppn.p
........
...NP...
.BN.....
PPP...QP
.K...RR.

# Position: 6
# Move: B

rnbqk..r
.p...ppp
p.......
.NpPp...
QPP.P.n.
P....N..
....KbPP
R.B..B.R

# Position: 7
# Move: W

.r.bk..r
..R..ppp
p...p...
.b..P..q
....QP..
....N...
.B....PP
...R..K.

# Position: 8
# Move: W

r...rbk.
ppq..ppp
..b.pB..
........
......Q.
.P.B...P
P.P..PP.
R..R..K.

# Position: 9
# Move: W

r....r.k
....bppb
..n.p..p
p.n.P...
.p.p.BNP
...P.NP.
qP..QPB.
..RR..K.

# Position: 10
# Move: W

r.b..rk.
.p.nbppp
pq.p....
...B....
P..NP...
..N.p...
.PP...PP
R..Q.R.K

# Position: 11
# Move: W

r.b...k.
p..p.nP.
..pqr.Rp
.p..p..P
..B.PnQ.
.P......
P.PP....
.K....R.


Before you ask : yes, I have the answers to all these positions. I'll post
them later if there is enough interest in the challenge.


Marc-Francois Baudot.

Marc-Francois Baudot

unread,
Jul 11, 1994, 4:01:26 AM7/11/94
to
P. Nolot challenges supercomputers with 11 difficult tactical positions.

Computer chess expert Pierre Nolot, in the may issue
of the French chess mag Gambisco, challenged readers to solve these
11 tactical positions. He claims computers won't be any help as the
positions have resisted the best micros for weeks of computation!

For the ten first positions, readers were asked to give at least the
first 3 moves for each side. In the eleventh position
readers also had to give the outcome of the game, assuming best
play for both sides.

Some readers resorted to databases and books, but there was at least one trap,
so they were easily uncovered!

Some of the positions are well known (pos 1, pos 5 for instance), others are
from obscure sources. All of them are rather difficult, but were solved by
humans under tournament conditions.

I'm sure Pierre Nolot has tested them very seriously on Chess Genius and
Tasc R30. I'm also sure he did not try them on Mchess nor Zarkov, so if
any of you want to give it a try...

But what he's interested in is the performance of the fastest machines on
these positions. He assured me none of them could solve any of these
positions within reasonnable time (while humans solved them at the board!).

So if Feng, Bob Hyatt or Hans Berliner want to take up the challenge,
Pierre Nolot would be much thanksful to them for posting their results.

Feng-Hsiung Hsu

unread,
Jul 12, 1994, 8:03:57 AM7/12/94
to
In article <2vqu8m$4...@sheckley.cnam.fr> m...@cnam.cnam.fr (Marc-Francois Baudot) writes:
>P. Nolot challenges supercomputers with 11 difficult tactical positions.

DT2 is not really a supercomputer, but what the heck.

>But what he's interested in is the performance of the fastest machines on
>these positions. He assured me none of them could solve any of these
>positions within reasonnable time (while humans solved them at the board!).

So Pierre does not think DT2 is a machine then (I assume that tournament time
control is considered reasonable?). When did it pass the Turing test?

I am not going to say anything further for a few days so that others have
the chance to try them out. The first position, by the way, was from the
Kasparov-Karpov match in New York. In fact, we were in the press room during
the very game. None of the grandmasters on site got it during the game.
Well, they saw the move, but did not think that it works.

Marc-Francois Baudot

unread,
Jul 12, 1994, 10:49:56 AM7/12/94
to
f...@watson.ibm.com (Feng-Hsiung Hsu) writes:


>So Pierre does not think DT2 is a machine then (I assume that tournament time
>control is considered reasonable?). When did it pass the Turing test?

Tournament time control is more than reasonable! Of course, you have
to find the good move for the good reasons. For instance, the Novag Diablo
finds the good move in the first position after 7,5 months (Pierre
uses the diablo to detect power failures!), but because it thinks white
has a bad position and can draw with kasparov's move. this would not be
considered a correct answer. But I guess everyone here is aware of this
problem.

I can't wait to see your results!

Marc-Francois Baudot

engelkes

unread,
Jul 15, 1994, 4:32:45 AM7/15/94
to
There is some discussion in this group on computers not being able to
solve difficult chess problems. May I add that computers often can't solve
very easy problems?
Some time ago, the Dutch Grandmaster Donner said that the only use for chess
computers is that they can be used as alarm clocks. Just give them any
position to solve late in the evening and they will beep you awake when you
have to go to your work. That they suggest the wrong move doesn't matter.
It's the beep that counts.

White: Kh1, Bd1, pg3
Black: Kh8, ph4
White to play and win. I've never seen a computer that didn't play gh4.

White: Kf4, Bd6, Na1, Nf6
Black: Ke6 pe7
White to play and win. Every move other than Be7 will win.

Many commercial programs don't even know the rules properly. Test this
position against your computer:
White Ke1, Rh1, Ra2, Bb1, Na1, pawns a4 b3 c2 h2
Black Kh8, Rg8, Ra8, Be7, pawns a5, b4, c3, g7, g3.
(the last moves before this position were: b2-b3 and h4xg3)
Now these moves are being played: 1. hg3 Bh4 2. Rg1 Be7 3. Rh1 Bh4.
This is **NOT** the second occurrance of the position. White has lost the
right to castle, so according to FIDE rules, the position is different.
4. Rg1 Be7. Now ask your computer for a move.
Will it find the easy checkmate in 2 moves or will he play a losing move?

Excuse me for not giving more examples. I have to set my alarm clock.
I hope it awakes me at a time computers are really playing chess instead
of brute-forcing the noble game. When I play my computer now it's just as
exciting as playing tennis against the wall.

--
W.E. Engelkes
Spaarnrijkstraat 26
2024 EK Haarlem (NL)
Tel. 023-259279

Rick Ganong

unread,
Jul 15, 1994, 9:10:48 AM7/15/94
to
"will he play" is cute, but should probably be "will it play"
(a losing move).

"occurrance" should be "occurrence".

Nice post. Thanks.

Reply all
Reply to author
Forward
0 new messages