Media comments on chess

[Brendan O'Gorman]

What's the solution ? I don't get it.

The white pawns lock in Black's heavy pieces and can't be captured by the black bishops which by themselves cannot mate the white king.

[Christopher Kreuzer]

You could probably come up with an extension to that position where the White king walks into the a8 corner and is trapped and forced to move the white pawns. Might even be a way to construct some way for White to deliver mate in a clever way if you tweak things a bit... (e.g. replace the Black queen with a Black bishop and it is mate in 2, no I have no idea if it is legally possible to get a fourth dark-square bishop, or indeed why there was a need to have three bishops in the first place other than to make a rather obvious point).

[IM Jack Rudd]

In the above position I have yes Black & White can castle but the moment I play 1.0-0 it cannot reply 1...0-0-0.
It does not recognise the exact nature of the position. We have fooled it.

You've lied to it, which is not the same thing. It's not the move 1.O-O that renders 1...O-O-O illegal, it is the fact that 1.O-O is legal; if the position was reached in a real game with 1.O-O being possible from here, then 1...O-O-O would have become illegal at least three moves ago (I think), and the computer would correctly determine that black cannot castle, and that both 1.O-O and 1.Rhf1 are mate in two.

[Kevin Thurlow]

"ah yes. I was distracted by work "

One of the problems with work!

I looked at the Penrose institute and couldn't see any mention of the test. It only takes a few seconds to see that you stay on white squares, and there is no point over-exerting yourself by stretching too much, so Kd1, Ke2, 25 times each should suffice.

[Geoff Chandler]

Hi Jack,

How can you lie to machine?

Set up this position.

r3k3/1p1p2p1/2pP3p/8/8/5R2/PPPP4/4K2R w Kq - 0 1

Tell it White can 0-0 and Black can 0-0-0 it's Ok with that.
It's accepts that both sides can castle. We know that is not correct.

It is accepting and analysing an illegal position.

It cannot see that the f3 Rook can only get out from a1 if the King has moved.
It cannot see that if the f3 Rook is an under promotion how did it get from f8-f3
or a8-a3-f3 without the a8 Rook or the Black King moving.

It just does not get it.

Same again. Enter this position.

8/8/8/8/8/8/3K1kPP/4R1b1 b - - 0 1

Tell it that is Black to move and it happily starts analysing that position.

Only trouble is it cannot possibly be Black to move.
Blacks only possible last move was Kf1-f2.
It can only be White to move. The computer cannot see that.

[Christopher Kreuzer]

Computers (or algorithms or AIs or engines or software or whatever you want to call it) can only follow its programming (well, AI and some heuristic processes will learn and develop their own rules). It would be possible to program a computer to be aware of illegal positions (most computer programs will not let you tell it that it is Black to move, when White is in check, as a simple example). Not sure where you would start with programming this kind of retrograde analysis, though. It would require computer programs that don't just analyse forward from the position they are given, but analyse backwards as well. Not sure if that has ever been done (surely it must have been?).

[NickFaulks]

How can you lie to machine?

Set up this position.

Tell it White can 0-0 and Black can 0-0-0 it's Ok with that.

You lied to it. It could have worked out that the position is illegal but you told it not to bother and to accept your input. Then you laughed at it for doing so.

[Geoff Chandler]

Hi Chris,

There are proof game programs out there. Not sure if they can retrace a long game in a short time.
It would possibly need the final move number.

I have an American friend who specialises in retro analysis, I'll ask him.

Hi Nick,

It cannot work out from that position that it is illegal for both White and Black to castle.
If it could it would not allow you to say that both sides can castle.

In the original position try telling it that White can castle queenside or Black Kingside, it will say no!

Basically what you and Jack are saying it needs the whole game to see whether or not it can castle.

That is what I am getting at. It needs the whole game. A human does not.

I am not lying to it (lying to a machine...good grief... which part Hell do I sit in for doing that?)

[E Michael White]

Same again. Enter this position.
8/8/8/8/8/8/3K1kPP/4R1b1 b - - 0 1

Tell it that is Black to move and it happily starts analysing that position.

Only trouble is it cannot possibly be Black to move.
Blacks only possible last move was Kf1-f2.
It can only be White to move. The computer cannot see that.

Why can the last moves not be:-

45 ..................Kf1-f2
46. Ra1xe1 there was a black rook there, leaving B to move ?

Black missed the mate in 13 with 45 ... Re1xa1

[LawrenceCooper]

Same again. Enter this position.

8/8/8/8/8/8/3K1kPP/4R1b1 b - - 0 1

Tell it that is Black to move and it happily starts analysing that position.

Only trouble is it cannot possibly be Black to move.
Blacks only possible last move was Kf1-f2.
It can only be White to move. The computer cannot see that.

I don't understand why it can only be black to move. Couldn't black's last move have been to d2 or e1 and white responded with a capture?

[NickFaulks]

It cannot work out from that position that it is illegal for both White and Black to castle.

We'll just have to disagree. I don't accept that it "cannot" work it out, I think you have instructed it not to waste its time bothering too much about that.

[Geoff Chandler]

Hi Chris,

Apparently there is not a program that work backwards to proof a game of this length.

He added: "Besides, the logic is clear enough to a human that they don't really need a machine to check it."

(Because a computer cannot do it.)

There is a programing challenge for someone. A program that prove or disprove a chess position is legal.

Hi Laurence and Michael.

Sorry, you are correct. A bad rushed example. (I need a computer to check it )

[Stewart Reuben]

Phil >Roger Penrose was interviewed on Radio 4's Today programme this morning, at about 8:20. His point seemed to be that computers are tremendous calculators but they don't understand chess.<

It dependxs what you mean by 'understand'.

Demis Hassabis programmed his computer to beat the World Champion at Go. But did it actually understand what Go is? He did not teach the computer Go strategy, unlike the machine which beat Kasparov at chess. Thst was programmed to play chess well. Even then, did it understand what chess is?

[Geoff Chandler]

Hi Stewart,

The computer does not even know it is playing a game.

Back to here.

8/p7/kpP5/qrp1b3/rpP2b2/pP4b1/P3K3/8 b - - 0 45

An American lad who too has seen the original Penrose article has proved,
just in case anyone had their doubts, the position could be reached legally.

I thought this would have been easy for a computer.

Give it a position and the computer works out the quickest way to reach it

If (and when) it could be done then that opens the door on all kinds of interesting chess things.

You could give it a famous final position. (Morphy at the Opera for example) and see if that position
could have been reached quicker. Just a thought.

1. d4 c5 2. Qd2 Qa5 3. Qb4 cxb4 4. d5 Kd8 5. Nc3 Kc7 6. Ne4 Kb6 7. Nd6 exd6 8. Be3+ Ka6 9. Bc5 dxc5 10. Rd1 d6 11. Rd3 h5 12. Ra3 h4 13. Nf3 h3 14. Nd2 hxg2 15. Nb3 g1=B 16. h4 Rxh4 17. Bg2 f5 18. e4 fxe4 19. Bxe4 bxa3+ 20. Ke2 Bxf2 21. Bf5 Ra4 22. Rh4 Bxf5 23. Rb4 cxb4 24. Kf3 g5 25. Ke2 g4 26. Kd1 g3 27. Ke2 g2 28. Kd2 g1=B 29. Ke2 Bg7 30. Kd2 Be5 31. Ke2 Bd3+ 32. Kxd3 Bgh2 33. Ke2 Bhf4 34. Kd3 B2g3 35. Ke2 Nf6 36. Kd3 Ne4 37. Kxe4 Nc6 38. dxc6 Rh8 39. Kf3 Rh2 40. Ke4 Rd2 41. Kf3 Rd5 42. Ke2 Rb5 43. Nc5+ dxc5 44. c4 b6 45. b3

[Nick Grey]

I was saddened to see the BBC report & read the web for the inquest into the death of Michael Uriely from chronic asthma.

It has been known from late 1970s that you cannot 'grow out of asthma'. I developed asthma in my teens (following hospitalisation from bronchial pneumonia). For ten years I was hospitalised at least a dozen times waking up with a mask & oxygen before finding a combination of cats plus alcohol were close to being lethal.

My heart goes out to his parents.