Website founded by
Milan Velimirović
in 2006

6:57 UTC
ISC 2024



Remember me

Forgot your
Click here!
to create your account if you don't already have one.

Rating lists


MatPlus.Net Forum General Variant Divergence and Atomic Chess
You can only view this page!
(1) Posted by Kevin Begley [Saturday, Apr 25, 2009 13:27]; edited by Kevin Begley [09-04-25]

Variant Divergence and Atomic Chess

There are three principle agents currently causing divergent skews in variant chess rules:
1) Variant Players,
2) Variant Problemists,
3) Variant Programmers.
(Perhaps Dick Cheney might be added as a fourth V.P., but the situation is not quite so grim -- yet!)

Unfortunately, in many cases a lack of inter-working causes problems with consistency and disputes about versions of variants.
I am lately concerned how this may have impacted Atomic Chess Rules, which may have been badly hacked by Programmers and Players.

From what I can gather from the internet, Nassah Bey Taher invented Atomic circa 1949.
The very basic idea is that units within the blast-radius (within a 1-square radius of a capture) are also annihilated (including the capturing unit)...

There are some indications (not sure how credible) that players changed the rules, as explosions relate to pawns, in order to avoid some "virtually forced win."

Can somebody please post Taher's original rules?

Furthermore, it seems the rules for check were further hacked by ICS programmers, in settling on a working sub-version of the official rules. Unfortunately, this resulted in a highly contrived set of rules, which requires a plethora of new stipulations which are, as yet, exclusive to this condition!

Essentially, there are three possibilities for check:
A) Checkless,
B) Check fully applies,
C) Hybrid (some claim this is the "official version" based on GICS having first implemented this version of the game on an ICS, though it is unlikely this adheres to the inventor's definition, and does not follow conventional precedent for variants).

To illustrate:
(= 3+2 )

Black to move.

Version A: Checkless [playable on ICC, and some other ICSs.]
Black has 4 legal moves (Pa3-a2, Ka2, Kb2, Kb1), any of which allows white to win (read: explode black King) in 1 move.

Version B: Normal-Check [not certain this is availible anywhere.]
Black has 0 legal moves (stalemate!). If check fully-applies, any move clearly results in an illegal self-check.

Version C: Hybrid [playable on GICS, and some other ICSs.]
In this absurd hybrid version, where reason goes pantless, Black has 1 legal move (Pa3-a2).
How is this self-check legal? Half-Ass Check Kontrivance (HACK) is the best answer.

Undoubtedly, this was a work-around to make a sub-version of Atomic playable on the GICS interface, which could not properly detect indirect self-checks (such as after ...Pa3-a2), without substatial alteration of the interface code.

Thus, some very sloppy, and contrived rules were hatched, due to lmiitations of the chess interface. The more logical work-around (checkless), unfortunately, escaped their gaze (or perhaps this too was considered a burden).

Just ask the original programmer if the Hybrid Version C would have been coded as Full-Check Version B, had this change been easier to implement on his server. Ask how much more work is required to recognize indirect self-checks (such as ...a3-a2). I suspect you will receive an honest answer.

This not evidence that the hybrid version is poor. Indeed, Loser's Chess was similarly hacked, to create a workable sub-version of Giveaway Chess (same story -- interface limitations w/2 check detection prohibited a fully correct version), and I find no fault with this game.

The problem here is that Hybrid Atomic supporters will not aknowledge the game's true history, and have sought to redefine check such that it excludes "indirect" attack on an enemy King, rather than aknowledge their version's honest rules.

A variant may adapt their rules for when checks apply, but redefining a check is not an option -- the term's meaning has survived thousands of fairy conditions before this hack appeared.

Hybrid-Atomic pages do not aknowledge that indirect checks are discounted by their condition. The wikipedia page on Atomic Chess (writen by a GICS member) claims that "check still fully applies." Nick Long himself -- a TD in the Hybrid Atomic World Championships -- fails to even mention the point in his definition of Atomic Rules ("";).

Long's page ("";), has plenty of history too... He talks about how Hyrbrid Atomic was introduced to the ICS world on (he claims on another page possibly before) November 27, 1995 on the German Internet Chess Server, after "Klaus Knopper collected the rules describing the game," and they were coded by the user with handle "connex" on GICS. In fact, he divides the historical periods into "golden, silver, and bronze age of Atomic", but somehow he mistakes the Hybrid implementation as the start of Atomic's history (discounting roughly 46 years -- from the time Atomic was created).

He does not provide the rules of the game as defined by the original inventor, nor does he mention why the meaning of check had to be altered from the original rules. This is a pattern you will find on many other Hybrid Atomic pages.

Version 1.0 of Atomic Chess should be reserved for the inventor's definition (if this is indeed what he called his game).
Sub-versions require modified names (maybe sub-atomic names), or version numbers...

As for Chess problems, in my view, the hybrid version is frankly unsuitable, and should be rejected as default or official.
Compare the aims possible:

Version A (checkless) has stalemates (though more difficult when checkless), and King-explosions, but no direct-mates.
Version B (normal check) has stalemates and checkmates, as normal, but no KIng-explosions, as normal.
version C (hybrid) is the ONLY version which would require the adoption of numerous combinations of stipulations merely to specify a victorious aim (e.g., win-endgame study, #/Xn-mate or explode the enemy King in n, #/!Xn-mate but don't explode in n, X/!#n-explode but don't mate in n, ...etc). To say nothing what this means for the confusion of existing stipulations (e.g., #/=n, requires a new stip, #/=/!Xn, then you'll also need #/=/Xn, !#/=/Xn, and the combinations of this nightmare abound).

For problem Chess, version B might be optimal, though A (checkless Atomic) is also quite interesting.
Version C does provide opportunities to exploit absurdities, and show how deep goes the rabbit hole these programmers have dug themselves into, but I would not advise this -- the folks who ducked their responsibility when it came time to implement this game will not have to deal with the nonsense they've left Problem Programmers to deal with here.

My appeal would be for some version control of chess variants, maintained by some official body (with members from all 3 principle agencies).

This would make availiable an honest history of the rules, history, and developments, plus allow for rulings on version disputes.
(Read Only)pid=3582
(2) Posted by Nick Long [Saturday, May 11, 2013 05:39]; edited by Nick Long [13-05-11]

Mr. Begley,

Seeing as how you've failed to reach out to anybody (especially me) regarding Atomic Chess and have received a lack of interest in this post for the past four years, I thought you'd be interested in knowing why (at least from my point of view).

There are actually three chess variants known as "atomic chess":

1 - Nassah Bey Taher (1947): It was reported that King Abdullah of Transjordan had an atomic set and Bey confided to the author that the actual inventor was the King himself. Uses a 12x12 board, addition of 2 "Tanks" and 2 "Aeroplanes". Tanks are knights with added power of nightriders, but only over two squares only (Nb1-a3/b3/c5/d2/f3); aeroplanes move like queens and can pass over any pieces but can only capture if there is a vacant square behind the target piece. Pawns promote to atomic bombs, which move like aeroplanes and can be detonated at the choice of the player when they destroy all men, friendly as well as enemy within a radius of six squares. If the king is destroyed, the next-highest valued piece replaces him. Aim is destruction, rather than mate. H.D. Benjamin adapted the atomic bomb for problem purposes to the 8x8 board in 1949... (also compare to Bomb Chess and Stratomic Chess). (D.B. Pritchard, Encyclopedia of Chess Variants, 1994)

2 - (As mentioned above, the conversion of Benajmin's variant to the 8x8 board as published in Fairy Chess Review in 1949, with atomic-bomb converted to a root-8 explosion from the square center, extended in 1970 by Michael Solomon to create "Radiation Chess"). Several variants involving bombs, bomb pieces, and the like followed. (J.B. Beasley ed., Classified Encyclopedia of Chess Variants 2007)

3 - What you call "Hybrid Atomic", which was also included in the 2007 Classified Encyclopedia of Chess Variants - its entry begins "Atomic Chess (at least the second game to have carried this title, originator of this version unclear). (J.B. Beasley ed., Classified Encyclopedia of Chess Variants 2007)

My website (which will be undergoing another revision at some point) focuses on option #3 of the above. This game was/is played on the Internet using the rules as described on my website. The 'check' rule does apply, although you're correct - it is a sort of hybrid. You can move into a forced checkmate within western chess rules, but the opponent is not obligated to checkmate you. For example, 1.e3 a6 2.Bc4 b6 3.Qf3 a5 (forced checkmate). White doesn't have to play 4.Qxf7#. According to atomic chess rules, if hybrid check didn't exist, 2. ... b6 wouldn't be permitted. But this isn't quite as simple as you seem to think to resolve. You're concerned with preventing moves, but it was logically done in order to allow positions like this to not crash the server/end in mate. (I will add in a section describing how the 'check' rule is a sort of hybrid):

Too tired to try to figure out the diagramming code on this forum, so here's an image:

(White just played exf6 on the previous move. Double check.) In this diagram, the server allows five moves for Black, Re4, Qe7, Be7, Kd7, and Kf7. If the hybrid check rule didn't exist, Black would only be allowed to play Kf7. Black is mated, no matter what move he makes, if White plays properly. If Black played Re4 and White completely misses Bxd8, blowing up the Black King and plays something else, the game continues. You can consider this a sort of "skill check" in a way. Although I do get your point about how it 'messes' with atomic-chess problems. Consider this position (with white pawn on e5 and black pawn on f6) a "Mate in 4" problem. There's also a mate in 3, and three mate in 2s.

Back to the versioning - There was no "check", nor even king-explosion ending the game in the 12x12 board "version". My best guess is that modified the modified Benjamin rules of 1949, or heard about a version of this game from somebody else. The current 'atomic chess' game being played online is essentially its own variant, with no recorded games prior to its introduction to GICS. As my chess site is primarily about Internet Chess History, I feel it is within my rights to divide the "eras" of people playing atomic chess online. Theory moves fast, and each era has its own subdivided style of games based on insight gained by the players over time.

I won't spend much time going over some of your other points but will go down a brief list:

GICS is a server. The client and server are separate. There was no GICS interface-client. It was the server itself that had trouble with coding atomic in, because of all the inherent poor coding that was done in the first place to make chess playable. It's taken over two decades to get the code to where it is now on FICS/ICC.

ICC essentially did the same thing to atomic that they did with loser's chess. The problem is that checkless atomic and 'hybrid-check' atomic have differences in some of the openings and endgames due to this rule difference. So the games aren't exactly the same.

There have been some other "programming" differences built into the rules by some other people, especially recently. One is that buho21 (a Java-based game server) won't allow the Kings to move next to each other in the endgame. This changes the endgame and the game itself. In hybrid-check, if it is KQvK, but the Kings can become connected before the Queen can separate the two, the game is a draw. With the buho21 difference, KQvK always wins.

I actually have had conversations with ICS programmers regarding atomic chess programming. I haven't been lucky enough to talk with connex though. The issue was that the server enforces one set of rules, while some interfaces enforce their own set of rules (regardless of what the server tells them). This has slowly abated over the years, but the current 'hybrid-check' version is the original and entrenched version.

I find Loser's chess to be a poor imitation of Giveaway Chess. In addition, ICC's "giveaway" chess is actually different from the real 'suicide' chess played on all other ICSes. Yes, there are rule differences between 'suicide' and 'giveaway' chess. Yes, they matter. Yes, they change results.

I however, respectfully disagree with your list of three principal agents causing skews in variant chess rules. I propose the following people to be blamed for skews in variant chess rules:

1) People who don't do their own research
2) People who disregard history (or are ignorant of it)
3) ICC. It's always ICC.
4) Problemists*

* - That was meant as a joke.

I thank you for bringing this topic up, and I'll be including a summary of the 'hybrid-check' rule breakdown in the rewritten atomic chess site.

-Nick Long of
(Read Only)pid=10413
(3) Posted by Hauke Reddmann [Saturday, May 11, 2013 20:48]

Funny, my first instinct was "C is the most natural".
That's why I stay at twomovers :-)

(Read Only)pid=10415

No more posts

MatPlus.Net Forum General Variant Divergence and Atomic Chess