Website founded by
Milan Velimirović
in 2006

6:14 UTC



Remember me

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

Rating lists


MatPlus.Net Forum Internet and Computing [poll] Composing in the Computer Age
You can only view this page!
(1) Posted by Hauke Reddmann [Friday, Jul 11, 2008 16:37]

[poll] Composing in the Computer Age

Personally, I *only* compose sitting at the computer
(it's very improbable that I have a brilliant idea
AND a chessboard AND no computer in vicinity :-).

What does this change?

- No more cooks. (Still, duals :-) But I usually run
Popeye over it at the end. [I use MateMaster for
better handling] ) A rather trivial issue.

- But note that this also means I compose more
audaciously, so to say. Don't have a particular
example at hand, but there were tasks I probably
haven't tried at all.

- MateMaster allows shifting and rotating a position.
This comes very handy for, I mean, for
optimizing a position. :-)

- Computers somewhat ease to interpret a modern theme
into your problem. This is irrelevant for me since I'm
an antimodernist, but if someone finally programs an
automatic themespotter even just for 2#, that would
be a revolution.

- More random thoughts?

(Read Only)pid=2443
(2) Posted by Juraj Lörinc [Friday, Jul 11, 2008 22:34]

I have started using computers for composing back in 1993. In these days I have still owned Sinclair ZX Spectrum with playing program that managed to solve h# and s#, but veeeery slowly - one test of twinless h#2 took about an hour, h#3 was virtually impossible to test. Helpmates in
were my first problems immediately tested by computer.

Then came Popeye in 1994, if I remember correctly, and I have been relying on them, alternating with VKSACH, till about September 2001, when I got new computer and new program WinChloe. I prefer using WinChloe now, but as I have it installed only on my main workstation, anywhere else I am quite satisfied with Popeye and Fancy front-end.

Nowadays, having spent 15 years composing using computers, I have already established method of composing.

First phase is on paper or just in my mind. Thinking about the aim, combing lines, weighting possible key motives.

Then the board, where I put main pieces and try to find working scheme, later trying to fill in unthematical gaps, often finding difficult points in construction, returning back to previous stages and so on... until I have position that may be tested with computer after all. This paragraph may be done using wooden or virtual on-screen pieces, it depends on circumstances.

And then work with computer - adding pieces, turning and moving position... you know.

I strongly prefer composing in genres where I am able to test my works. Recently I have even refused to compose for a section proposed by Slovakia in the international match as it was using complicated fairy condition untestable by any known computer program, I have rather concentrated on other sections so as not to lose time composing something wrong. By the way, submissions of other country for the section were all cooked.
(Read Only)pid=2444
(3) Posted by Siegfried Hornecker [Saturday, Jul 12, 2008 16:09]; edited by Siegfried Hornecker [08-07-14]

Positions are given as FEN codes, sources are given in HvdH code, {c} means "correction". Hopefully this is not misunderstood as propaganda for EG.

Same as Hauke, I also compose almost only at the computer. However, I found a basic matrix for one of my best studies without a computer but completed it with it (I won't go into detail here, the final study was this one: 8/1q3p1p/2p1P1kP/1pP3Pp/1P5P/8/P6P/4K2R WTM + tk69#U530 2007).

Normally, however, I use computers to compose. There are two kinds of studies (yes, I admit!). One is when I set up a position that comes into mind, look if there's something in it, and adjust if necessary. I don't know how that works, but it does. One of these was 5k1K/4Np2/5P2/5P2/p1P5/P1r4P/P1B3pP/2R5 WTM + ts10#15392 2003 ({c} ts10 08/2004)

The other kind is when I have an idea and use the computer to work on it, most of the time to add foreplay. Of course, introductions have to be found manually, and only the ones I find can be tested with a computer then for correctness. A good example for that way is the one by Richard Becker I cited in a recent article in MP Review where I talked about development of old ideas (I think, in Spring 2008?). By the way, he told me to have found Gunst's introduction without knowing Gunst's study. 2B5/p2p2k1/1P6/5r1P/8/8/N7/7K WTM + 1.p tm04 2004 Richard Becker

So these were my personal opinions. Now let's take a look at the whole "war" of the pro-EGTB-study faction against the con-EGTB-faction. The arguments of the pros are good enough (according to Iuri Akobia in an article in MPR Spring 2008, reprinted version of EG 173). Note that most of the things are my opinion and only loosely based on the article.

Pro 1: Iuri mentions the use of checking for soundness of a position. I can agree to that, but I additionally believe that complex wins should be avoided. KRB-KSS is no complex win since it is almost always won, I talk about positions where the win is unclear without EGTB, not those with long wins far beyond human understanding. It is unimportant for a composer if anyone actually could win it on the board - nobody needs to give all 35 moves for KBS-K if such an endgame comes, to prove it is won; nobody needs to give all 100 moves in KSS-KP since the Troitzky line is well known today etc.

Note 1: The seemingly big size of EGTB is neither pro nor con. In a few years, there will be HDs with 5 Terabytes (I read an announcement by a big company recently) and hopefully internet will be usable then, not staying at some Kilobyte or Megabyte per second but hopefully be fast enough to download a terabyte in less than two weeks. I doubt so, though, since there were no breakthroughs in the last years (so it may take another 15-20 years for such a connection if it stays to grow as slow). It only proves that telecommunication companys are too capitalistic to give full optical fiber access to everyone (yes, we talk about more than one terabyte per second download speed here - if nobody else uses the connection - practically it would be extremely expensive, of course, to have such a connection) or at least offer it. I think, some universities have optical fibre, though, so it may be a good idea to download tablebases there. Harold van der Heijden gave in Iuri's article the link I anyway wanted to add here:
Note that this is fully legal, although some industry idiots want to tell you, all peer-to-peer is illegal! It will take three months for a standard connection to download all 6 piece tablebases (under best circumstances), though!

Contra 1: Composers will be more likely to let their studies end in positions that are unestimatable for a solver. In my opinion this is the fault of the composer, not of the computer.

Pro 2: A list of mutual zugzwang positions is available and opens door for beautiful constructions...

Contra 2: ...or use of the position for something that is not artistic. Again, in my opinion, if the composer is too bad, no good ending will help him.

Contra 3: ...or use of a position that can't be understood by humans. I believe, this is the same as in Contra 1.

Note 2: At this point I have to mention that computer generated databases do not count as anticipations, even if they were published on paper. Of course, and I agree to HvdH here (in fact, I had the same thought), all positions are quite anticipated by a mathematical formula to create all positions. Please let me add a personal question here: Would the work of for example ND Grigoriev of MA Zinar be less admirable if they had a computer with EGTB for all positions (with up to 64 pieces)? I don't think so! They only would've had an easier time to check the positions for hidden traps (although at the level they compose at it is extremely unlikely there is an incorrection anyway). In fact, I think it would have been at least the same admirable, since EGTB tempt to just believe it without checking the positions. So if they would've not understood the endings they wouldn't have come up with the positions and not been able to check it. Of course, this only holds true for endgames that can be understood (like pawn endgames - and I've heard about an OTB grandmaster who refused to exchange pieces since was unable to estimate the pawn endgame he would have been confronted with). This doesn't hold true for all compositions, though, some modern ones (by other composers) look like they have been done without the composer understanding what he did. What - however - is so bad about five piece endgames when a composer knows what he does? Pervakov won the first prize with a pawn study with only five men in a russian magazine 2000, and it's very famous today. It seems, even five piece endgames - especially with pawns - have some surprises left! Gosh, I found a small one in a four piece endgame some time ago! And it is without mutual zugzwang! 8/k2p4/8/8/8/6K1/6P1/8 WTM + (Internet, 2006/ix/24)

Offtopic: Regarding a comment by Iuri Akobia in the article: While it is true that the database of HvdH had 67000 studies in 2004, when the last version was released (this is what Iuri wrote in other words), he's still collecting! I expect the next publication to have at least 75000 studies and improvements to old ones (like, new duals found or refuted, etc).

Pro 3: A good composer will always try to show the EGTB mutual zugzwang for both sides if the mutual zugzwang is the main part of the study. Nobody will say someone who composes a task that leads in a variation to a mutual zugzwang that his Babson study is bad because the zugzwang only appears for one side. However, that is perfectly acceptable to say it if the zugzwang is the main point.

Note 3: In my opinion, endgames from practical games are not anticipations of studies, even if the endgame is the same position. This does not hold true if also the moves played there were the same! For example, Farago took and endgame position of some national OTB championship where he judged and published it as study since he immediately saw how to play it and the players didn't. 8/2p5/8/2kPKp1p/2p4P/2P5/3P4/8 WTM + ts23#07 1958 - after 1.Kxf5?? Kxd5 and a few more moves the game ended in a draw.

Note 4: Judges that downgrade the most economical studies are inacceptable! Judges that downgrade obviously computer generated studies are not! However, a judge must be very careful not to be wrong here - a good judge, in law and chess, always believes in "in dubio pro reo". -> Note, Emil Vlasák also mentions this, saying that Mario Matous (sorry for misspelling!) cannot be distinguished from a database "miner", while being surely none.

After this, I'll talk about some opinions on things Emil Vlasák mentions in his article in the same EG issue. As above, this is only my personal viewpoint.

Note 5: Databases do not cause problems. Humans do. Especially those who abuse databases but that is not the fault of the database. It is not the fault of planes that they can crash into houses like in 1992 in Amsterdam. In that case, the fault was at the missing maintenance - which was a human fault, not the fault of the plane. Nobody wanted to forbid all planes after this. See for more details on the catastrophe.

Contra 4: Composers don't need skill anymore but can mine databases. As Iuri Akobia shows in his article, it is still hard work to make a good study from the mined positions. If the study is bad, it shouldn't be honored anyway so it is the judge who can fail, as well as the composer, in my opinion, but not the database.

Contra 5: Theoretical positions can be published as studies. That's unfortunate, although not a database related problem. Even Troitzky did so. Often it's difficult to distinguish so I'd as well accept it as a study. 8/5K1k/8/4N3/8/7p/7N/8 WTM + td20 1906 (A. Troitzky)

Note 6: According to German law, "Urheberrecht" (which is similar to copyright) is a human right so we will not accept a copyright for computers. It will be completely legal here to break copyrights of computers. Then EV mentions the publication Love and Sex with Robots. I believe, that "full valuable sex" also consists of other things. It may be fully satisfying, but the valuability only can be full if there is a deep spiritual connection, in my opinion. It is a philosophical question that was raised several decades ago that would have to be discussed here: Can robots/androids have a soul? My time is too limited and the space here, too, to solve this riddle (I doubt, the space of the whole universe is enough to explain exactly what a soul is etc. - the only thing we can do is believe in the sciences or religions but they also don't explain it - even the most essential thing, if and how souls can be created and destroyed - although many idiot psychiatrists speak of damaged souls when maybe only body and psyche may be damaged -, is so difficult that it may be unknown forever. It'd be essentially the one question about the meaning of life, the universe and everything (and serious scientists don't consider it to be 42, by the way).

Note 7: There was this issue with FIDE album 2001-2003. In fact, the judge (I don't want to name him here) was legitimate in his views as long as there was no codex addendum clarifying that EGTBs don't anticipate compositions. Afterwards, his behavior, in my opinion, was understandable but not appropiate. And: I don't want to be an album judge! Even one award takes ages to me (but it is of high quality then). I misunderstood the software I used (it's an excel sheet by Michael Roxlau) so I'll wait for the article of it which will delay work even more. So the 2006-2007 award of that certain magazine will not appear before 2009 or very late 2008, I'm afraid.

Contra 6: Programs that mine databases for certain things will improve and may do all the work for the composer. I disagree to HvdH who says that one needs an idea first to mine, since it is possible to take ideas by other people (the lists of mutual zugzwangs are available on internet and don't have to be generated by everyone for himself, for example), and it won't be long until orthodox problems (!) may be overwhelmed by someone near to genius and madness with computergenerated compositions of all possible positions (nobody wants to see a mate in 55 in the endgame KRB-KR, as an example).

Note 8: I disagree with EV on the point why one should compose. Composers should not compose for solvers nor for titles and honors (although these are nice) but for himself. If you compose the worst crap but have fun doing so, do it, but make sure it is never published! If you compose something good, publish it! Of course, one can compose for honors if that is he believes the honors are good for himself. Wotawa didn't want honors but was a good composer. Senkus wanted honors and was not a composer at all. On the other hand, many a classic composer won many prizes etc. and was a brillant composer. The studies, however, don't look like composed just to win prizes but rather for this unexplainable feeling (Ladislav Salai jr. calls it "euphoria"?) that makes us thinking that the world should see this and other people we don't even know should have a moment of joy when experiencing the idea. Again, the magazine then maybe has been chosen for the hope of honors (else he could've written a book, for example), but this does not mean he only composed for the honors. But the composition was made for himself and for those who will see it.

Note 9: Unless you have a laptop, it is unpractical to take a computer with you into the bed and read a chapter before sleeping. Even with a laptop it's unpractical since sometimes I fall asleep while reading. This does however not mean the compositions there are bad. At the moment the mobility and ease of use is the biggest advantage books have over computers.

Now onto the Ladislav Salai jr. text, published in the same article.

Note 10: Censorship is evil!

Note 11: I agree to HvdH that an editor should be able to explain a composer why the studies are not good. Also, it is common to publish rejected studies somewhere else (although I do so in internet sometimes).

Contra 7: A composer who has no contact to solvers misses something. Often solvers are also valuable for help. There's a solver I have good contact to, who always is a valuable help, although I must sometimes be annoying. ;-)

Contra 8: I have to agree with Salai that too many computer generated sidelines may kill the idea. An editor once had me creating tons of analyses to prove that a certain variation was drawn but it didn't kill the study then, I hope. I think, it's the base of the study that may be wrong but a nice idea can be only wounded - not killed - by such variations if there are not too many such lines. An ideal study wouldn't require such lines but - citing a dutch author - "sometimes the material resists". I agree that studies with zugzwangs not understandable to humans should get subtractions in the rating. However, one must be careful with the evaluation what humans could unterstand - most people also wouldn't have found the combination of many Morphy or Anderssen games but this doesn't diminish their value.

Note 12: The Dvoretzky 60 JT was for studies with value to practical players! Is it wrong to send gamelike tablebase endgames to such a tourney? Maybe, maybe not. One still has to find a starting point for it, or at least come up with a position. I think, LS chose a bad example, many other tourneys would've done. He complained about composers doing so. I agree, though, that the artistic value is not as big often. Here I also like to reveal a secret: In MatPlus an endgame was published that was composed in such a manner! It was the one in Summer 2007 with KP-KPP by some study-anticomposer named Hornecker! ;-) (Ok, so you can tell me I'm biased towards EGTB studies)

Note 13: Studies are my life-love, too!

Pro 4: EGTB studies can be a pleasure for solvers, if a good composer works on them. That's what I said above, and LS shares the opinion, it seems.

Contra 9: EGTB studies may not be a pleasure for solvers. I agree to HvdH that it is not up to the composer to choose what should be used in solving competitions etc. but the composer should always be careful to send too difficult studies not to magazines that have solving lists, in my opinion. That is why some of those ultra-complicated endgames that I consider barely solvable appear in EG and MatPlus (as well as others, though, I don't treat EG and MatPlus as recycle bin).

Note 14: Unclear sidelines for which EGTB would be necessary also appeared in earlier publications, far prior to EGTB. Also, for a long time KBB-KS was thought to be a draw in a certain position which prove wrong with EGTB. It's still impressive if one can win with KBB-KSS like in some studies.

Note 15: LS says, Emil Melnichenko let his study, 1st prize in Rossi 60 JT, compose from beginning to end with a computer. HvdH highly doubts this. In my opinion saying this without a proof is inappropiate. He should've said, "I believe, ...", but not said as if there would be a proof. However, I find the comments of HvdH - some kind of discussion between him and the author - highly interesting! LS says, a computer doesn't know how to finish lines at the correct moment. In my opinion, a composer that composes for so long as Melnichenko would know to do so even if he'd use a computer.

Note 16: LS mentions a study with 18 pieces where he says, the human touch is missing. Such compositions existed long before the computer era, too. I believe, studies without an idea should not be honored. Luckily, most studies have ideas!

Finally, the number of pros, contras, notes in my statement does not say anything about their value etc but are just given for easier reference when replying. Also, there were no such numbers in the articles of which I borrowed the ideas of what to talk about. It took me several hours to write this (would be 17 KB in a textfile), as you may see, so hopefully I didn't disappoint you too much!


PS, July 14th: Somebody told me, that EGTB issue is different from twomovers, moremovers and studies. What then to say, my reply is for studies!
(Read Only)pid=2445

No more posts

MatPlus.Net Forum Internet and Computing [poll] Composing in the Computer Age