“In the late 1970s, I circulated among the members of the now defunct Fairy Chess Correspondence Circle some ideas and illustrative problems for variants arising from different effects of a piece moving to an occupied square. By 1987, many of these ideas and problems had appeared in a series of articles in George Jelliss’s magazine Chessics (issues 10-12, 17 and 29/30) under the name of ‘Chess Reactions’.”
..skipped.. ⇒ see the full article in PDF, 23 pages
“The present version of this account contains 105 problems belonging to 17 distinct variants (4 of them established ones, the others my own) plus another 12 more miscellaneous problems in an appendix. Almost all the problems have positions that are ‘fairy-legal’ from the normal game array; 76 of them are CT originals/corrections of variable quality, some being mere sketches to illustrate possibilities. Only a few are computer-testable, but although the majority have been looked at by someone other than myself, in the very nature of things it is likely that some will prove to be unsound. It is hoped that any defects in the problems will neither detract too much from their interest nor discourage composers from tackling some of the variants described.” – Chris Tylor
I would dearly love to see some comments (even negative ones) about this piece. I have put quite a lot of work into it over the years, and still feel that there ought to be something in it to interest somebody!
Dear Chris,
I like fairy conditions where a lot is happening in one move, and I have read your article with pleasure. It seems to me that I noticed one potential common problem for a number of your conditions, namely, that infinite loops may occur. What is your idea of how to treat this problem? (I guess that this is not a real problem for those conditions in which only pieces of the same color play during one move, but in those conditions where the controlling side can be switched during the move, it seems that this could be a problem.) I do not have any particular example currently in mind, but if I am not clear what I am talking about, please say so and I will try to construct an example for illustration.
Also, I analyzed in particular your Auto-Kick problems, and I have some questions about B5.3 and B5.4. In B5.4, I do not understand the intended solution. After 2.{Rc7,Sc1}, there is no rook anymore on c1, and thus I am not sure how it could be kicked to a2. And in B5.3, I understand the intended solution and agree with it, but could you please check whether this is a cook: 1.Re4! ~ 2.{Be4,Rg6}#.
Thank you very much for your interest, Bojan!
In B5.4 there is (I hope) just a simple documentation error: move B2 should begin with Rac7 – i.e. the bRa7 makes the first kick; the move seems to work all right after that.
I accept your cook to B5.3. I think the problem can be corrected by moving wSc7>h8 and wBh7>c8 – which seems to give a better-looking position anyway. Do you agree.
(I did write an answer to your infinite loop point, but something went wrong and I lost my text. So I will post this and send the rest of my answer later.)
Dear Chris,
Then, I guess, there is a diagram error in B5.4, there should be bRa7 instead of bSa7? And concerning your proposed version of B5.3, I think that the intended solution doesn’t work after this modification: 1.Re3 g2 2.{Be3,Rf2}+ {Ke5,d6,e5,Ke4}! (the same defence works for the line 1… h5).
Here is an example of an infinite loop I was talking about, in (e.g.) Free-Tag Chess. I will present an example with a fairy piece, because I find it the easiest way to illustrate the problem (but a similar example can probably be constructed with orthodox pieces only): white rose d1, black rose b2. In a direct play, can White move the rose to b2? Then bROb2 is tagged, and Black is allowed to play bROb2xb2 (assume that the rose null-move is allowed), tagging wROb2; then White could play wROb2xb2, tagging bROb2 etc. Therefore, if we are not in a help-play but in a direct play, when exactly this chain stops (if both players, for some reason, want to have a rose on b2 after the move ends)?
Yes, diagram error in B5.4, which I missed; sorry! (It must have crept in when I changed all the diagrams from one format to another.)
And yes, your defence to my B5.3 works; well spotted! Can you see a way to correct the problem without using a flight-taking key or losing a variation? It may not be easy.
As to your rose example, I would say that if Black allows his rose to finish up on (say) a4, then White may play wROxb2, but if Black insists on playing his rose back to b2, he thereby prevents White from playing the move. Will that do?
The last proposal seems fine, but it is interesting to think how should it be formulated in the general case. I thought a little about this, and does this formulation captures your intentions?
During a tag-chain, it is forbidden to make a sub-move that leads to a position that has already occurred (including the information on the tagged piece) in the same chain.
In my example, this would mean: wROd1-b2 (tagging bROb2) is allowed; then, moving the tagged bROb2 again to b2 (tagging wROb2) is also allowed; but then White cannot again move his tagged wROb2 to b2, tagging bROb2, because that would lead to the same position as after the first sub-move (however, White is allowed to move his tagged RO to b2 with capture of bROb2).
On B5.3: If I think of some fix, I will let you know, though I guess it won’t be so easy. Meanwhile, I managed to find also the following cook to your last version: 1.Bb6 ~ 2.Bd8#
On B5.4: OK, I agree that the solution works with bRa7 instead of bSa7, but in that case there is no need for such a complicated chain in the second move, isn’t is easier just 2.{Ra2,Bd2,Kh6}? Or I am missing something?
And one more quesiton about those conditions in which a side can move opponent’s pieces during the move. This had been asked in the Problemist in the comments to C1.7, but I don’t know where nor whether it was answered: in Oppo-Additive Chess (but a similar dilemma exist in some other variants of yours), “If the bK is on the eighth rank, can Black move a piece combined with a wP to that rank? That is, who chooses promotion in such a case?”
Thank you, Bojan; I will answer your points on Auto-Kick first.
I accept your new cook to my B5.3 version. I don’t know what else might work. moving wR to (say) a3 with bPa4 fails to 1.Rd3! (threat {Bd3,Rb6#}) e4 2.Bd4#.
On B5.4, I accept that your shorter B2 chain would work, spoiling the problem. A cure might be possible, but I wonder now if it is worthwhile looking for one. I thought at the time that the complex chains of kicks in this and B5.5 were interesting, but now they just look meaninglessly complex, with no artistry or elegance in them (unlike the tag-chains in B3.5 and B3.6 – which I hope you found nothing wrong with, seeing that you like moves where a lot happens!). So I wonder if it might be better to change the Auto-Kick conditions, either to allow single kicks only or to have kicks followed by passes as with Oppo-Kick. What do you think about that? I will try to answer your other points later.
On B5.4, Bojan, moving bRa7>b7 might be a simple fix if the problem was thought worth saving, but the possibilities are so numerous that it would take a lot of analysis to be sure of anything.
On the oppo-tagging issue, your formulation looks good, but my preference would be to avoid making a ruling on the point myself but to leave it to anyone who composes a problem in which the issue comes up; that composer would naturally choose the interpretation that works best for that particular problem. (I think that is the normal procedure in such cases.) And it might never happen; the situation is a very specialised one, and I cannot visualise how it might actually occur in a problem. So if you happen to have such a problem, be my guest and choose your own interpretation!
The related issue of pawn promotion in Oppo-Additive and similar variants is another matter; the situation could easily occur even if it was not part of the problem’s theme. My ruling here would be that the side making the move should choose the promotion; the concept of control changing in the middle of a move it such an alien one that it would be best avoided – unless, as with Oppo-Tag, it was an essential feature of the variant.
You have been very industrious in looking up the Problemist readers’ comments on C1.7. My memory of these comments is that they covered points like pawn promotion that did not occur in the problem, and that none of the comments even mentioned the problem theme itself with the (as I see it) very interesting moves of the combined piece fighting a duel with itself. I think it was largely these comments that led me to avaoid making any other attempts to publicise my work until now.
Dear Chris,
Thank you for your answers. I will try to respond to the points you raised.
it is certainly clear that reducing the condition to single kicks only would make it easier to grasp, but from my point of view, it would also take away most of the fun from it. 😀 And about the idea of having kicks followed by passes, I feel it as less natural than the current definition of Auto-Kick. Therefore, in my opinion, it would be the best to leave it as it is. (But, of course, these are your inventiones and you are free to define them as you feel most appropriate, I am just answering your question with my subjective opinion here.)
Sorry to say, but this seems as another cook: 1.{Rc1,Sc2} {Be7,Sf6} 2.{Kc1,Rc2,Sa2,Bc1,Kh6} {Kf6,Se7,Bc8,Rh3,e3}#
I concentrated on Auto-Kick Chess for no particular reason: I had some spare time, and problems on that page somehow were the first to caught my attention. Today I had some more spare time, and I decided to dedicate it to problems with Free-Tag Chess (seeing that you like them the most, and this would have been my preferred choice anyway if Auto-Kick Chess had not occupied me first). I am sorry to bring bad news, but it seems to me that a number of your Free-Tag Chess problems are unsound. (Please check, I hope that I am wrong.)
I am sorry, Bojan; I have been too busy working on this Fairy Classification project of Julia’s to think very much about tags and kicks. But yes, your cooks and duals are all correct, and I am very impressed by your skill in finding them – especially as most of the mates they end in are better than mine!
As to cures, B3.1 might work by moving wS>f7 and bK>g8 with the same solution. The bK triangulation is lost, but otherwise the problem looks OK.
B3.2 is too complex for a quick answer, but it might be possible to rework the position somehow to keep the white play and end with a mate like yours.
B3.3 looks unsaveable – and not worth saving anyway.
With B3.4 your mate is a wonderful one, and it should be possible to rework the problem to include it. What about simply wKf3 wQg2 wBb1 bKa3 for 1.Kb4 Be4 2.Kc5 and then your mating move? The bK then has to move away from the the board edge in order to be mated!
The dual in B3.5 was a blow, the problem being my favourite. But placing the wK on the long diagonal might stop the dual (unless the K has some role on c3 that I had forgotten about). So try rotating the position 270 degrees, move wK>a1 or h8, and add wPc6 and bPc7. Alternatively rotate 180 degrees, move wK>a8 or h1 and add bPs g5, g6 and g7 (an extra unit, but the bR block on bP g5 is a plus). Would either of these work?
As to the kick problems, I have not looked properly at your B5.4 correction, but B5.3 seems just to have too many white pieces!
Absolutely no need to apologize. I fully understand that the fairy classification project is (and should be) much higher on your priority list than discussing whether some problems have some cooks. 🙂 Feel no pressure.
I managed to find this cook: 1.Kxf7 Kf5 2.Kg7 h6+ 3.{Kh6,Ph7} h8=Q#.
I played with this idea for a while, and concluded that White is way too strong in this setting, found even this h#1 in your position: 1.Kb3 Be4# (with a quite complicated tag-chain to see that Black is even checked in the final position, let alone mated :D). But this cook gave me another idea that I analyzed for quite some time, and really hope that it is sound.
White: Kc6, Pe4f7; Black: Ke5; h#2.
Intended solution: 1.{Ke4,Pe5} Kd7 2.{Ke5,Pe6} {Pd7,Ke8}# (threating {Pde8=Q,Kf7,Pe8,Qxe5}; bK cannot step to d-column because of 3…. {Pfe8=Q,Kd7,Pe8=Q,Qd7,Ke8,Qd7,Qxd?}, and the explanation is similar for f-column). A blemish is, however, the fact that these virtual tag-chains are not unique.
The second version does not work because Black is not mated in the final position, he can defend by {Ph6,Kh5}. And the first version seems to again have a dual: 5.{Re4,Sg5,Bd8,Kc8} Ra8#
Here is my attempt to fix B5.4: remove bRa7, move wBa3->d6, move wSe7->a7. The intended solution now is: 1.{Rc1,Sc2} {Sc8,Rd6,Bf6} 2.{Kc1,Rc2,Sa2,Bc1,Kh6} {Sd6,Rf5,Kf6,Bf5,Rh3,Pe3}#
Hopefully this is correct, but as you said, the possibilities are so numerous and one can’t be sure.
I have still not looked at your kick correction, but I have done a little work on the tags. I simply had not realised how powerful KBQ and similar combinations could be in tag. Your H#2 looks accurate as far as the play is concerned, but as you say the virtual tag chains are not accurate; in the threat the fP could promote to anything.
One way to keep some control would be to use this force in the mating move of a series helpmate such as B3.5. I thought I had such a correction to B3.5, and was in the process of typing it out when I saw a dual. I would very much like to collaborate with you on producing a tag or other problem that might be publishable – perhaps here on JF. But it would be easier for me to contact you by direct emails than by these posts.
Of course, if you think we could produce something together, I will be glad to give it a try. You may write to my Gmail address, the username is bbasic.
OK Bojan; I will retype the rest of my reply.
My answer to the potential difficulty of infinitely repeated loops of sub-moves is to state that any chain must have a definite completion; I say this on page 8 in the first bullet point under ‘general principles’ (though the statement there needs amplifying). The general point that moves including closed loops of sub-moves should not be considered as duals would need to be accepted. (There is a precedent for this in ‘Win’ Studies, where White is often able to repeat moves before embarking on the winning plan.) I had not thought about any difficulty arising through control being switched during the move, and am not sure that there would be any such difficulty, since the original control would be restored at the end of any loop. But perhaps you can think of an example.
Closed Tag and Kick loops are interesting in their own right, of course (though Skip ones are trivial and Pass ones fairly obvious). Some could be highly complex, and their formation could even be the aim of a problem!
I have one extreme scheme which requires that only self-capture is allowed (with exotic fairy pieces and neutral ones also). Still I do not know how to computer-test it.
Your scheme looks like my Auto-Capture Chess, though you may not have the same rules for check. Have you composed any problems in it? I am afraid that I do not really know anything about computer testing.
More like Cannibal chess mentioned in your article.
The idea was to have an full cyclic Zilahi with active “sacrifices” in a twomover.
If I remember correctly, it is even C+ in a sense (tested with neutrals on e3 e6 h6 and all invalid solutions were deletyed).
White : Kg1 Ph4g3f2 Gh7a3
Black : Kh3 Pg4 Be6e3 RHh6
Neutral : Bf4 RVe4 RHf6
h#2
G – Grasshopper
RV – vertical rook
RH – Horizontal rook
Priority captures (must capture if a legal capture move exists)
Cannibal chess(?)
b) h7 -> c8, c) a3 -> c8
a) 1.RHnxh6 Bnxh6 2.RVnxe3 Bnxe3‡
b) 1.RVnxe6 RHnxe6 2.Bnxh6 RHnxh6‡
c) 1.Bnxe3 RVnxe3 2.RHnxe6 RVnxe6‡
I found your problem difficult to grasp because of all the different fairy elements, and I still do not understand the Cannibal aspect, since no white or black pieces capture others of the same colour during the solution. And if such captures were allowed, part (b) would suely fail since Black could play 3.Kxg4. Also, if pieces were allowed to check their own kings, the bK would be in check (and mate) by the bPg4. Have I got this wrong?
The cannibal aspect obviously is that Black captures black pieces throughout – but with neutrals! I agree that the captures Kxg4 or Pxh3 seem to destroy the problem, unless the condition is taken to mean that neutrals only are allowed to capture pieces of the moving side.
The following interpretation was intended:
White can capture white or neutral pieces with white or neutral pieces (as neutrals may be considered white here).
Black can capture black or neutral pieces with black or neutral pieces (as neutrals may be considered black here).
These rules did not apply to kings which move and are attacked normally.
This seems a rather special interpretation, which may not make computer-testing easy. It is normal for kings to be exempt from being captured or checked in special ways, but very unusual I think for kings being themselves unable to capture in these special ways.
I am afraid I cannot help you with your problem, but I thank you for sharing it.
Dear Chris,
I am really curious to know whether you are the same Chris Tylor who invented Football Chess, as described in “The Encyclopedia of Chess Variants”? I learned the game from the first edition of the encyclopedia and have taught it to people in various countries over the years. It generally meets with a very positive reception.
The encyclopedia suggested starting the game with the ball on d4 and disallowing a goal on the first move. However, this enables white to win immediately with Qd4-ball e3.
I usually start with the ball on e4 and it seems to give a good game (although white has a huge initiative).
If you’re not the right Chris Tylor, do you know anything about the inventor of this game?