Being an active ICCF player my experience is that time wasting within ICCF tournaments is in my opinion a real problem and ICCF don't seem interested in taking a hard line against this.

After 20 plus years of playing ICCF when I finish my current games I'm looking at other correspondence options. It is not my idea of enjoyment to be playing an opponent who playing down a mainline of the Petroff  takes 39 days to make ten moves before getting another 40 days to play the next 10 moves....and we are still playing a mainline!!

Hi.

Thanks for posting that. On the face of it there is a quite impressive number of older games that up the total.

Have nice day.
/ CbT

Hi.

In principle of course you would want as much information as possible. The problem for me is not exactly that the base gets to big or anything; the very marginal disadvantage in ease of usage from adding more games is not very concerning. It's more like statistics is the problem. First of all you may use it in decisionmaking, even if more commonly when I played at least, I based opening decisions more after what I found to be current best games in line. Secondly it is conceivable that I'd want to convey to others some statistics about some line and it does not feel credible to do that without a good idea of what is in the data.

In general I'd say a large chunk of games in the corr database are already not very relevant. Big engine improvements over time mainly (and hardware). To me there would appear to be a split in what you can do as a corr player with the invention of widely available AI engines. For this reason I'd rate fresh games to be volumes more valuable than ones before people started using AI engines.

If we go back to commercial corr databases. If not ICCF games or Lechenicher what other places are there where you could possibly find hundreds of thousands of qualitative games? (not post AI introduction qualitative more both players would be very likely to use decent engines qualitative) To me it seems more likely to be not very relevant ones that fill up these bases. If I am wrong though that would be very ok as then it would seem likely I'd learn about som new source of nice games. 

Have a nice day.

Hi.

LeeRoth wrote on 11/01/20 at 15:37:04:

Sure. Thanks for the tip. I still don't get what the added value is though, since those extra games may be just about anything

Have a nice day.

Hi.

Updated my homemade corr database with latest ICCF games now and actually bothered to make a spreadsheet with game sources.

ICCF
2004                              205
2005                              6442
2006                              17 356
2007                              28 251
2008                              36 880
2009                              39 600
2010                              42 978
2011                              43 686
2012                              49 809
2013                              53 900
2014                              60 220
2015                              60 984
2016                              62 479
2017                              65 502
2018                              68 056
2019                              64 282

(2004-2019                  =            700 630)

202001                        4998
202002                        4501
202003                        5937
202004                        6070
202005                        5269
202006                        5843
202007                        6155
202008                        5062
202009                        5781
202010                        6025

(202001-202010            =            55 641)

Lechenicher
Lechenicher startto201512            240 902
Lechenicher 201601to 202006      103 604
Lechenicher 202007to 202009      4133

(Lechenicher                  =            348 639)

Other
IECG                              166 659
Chessfriend                        22 675
ASPCCto201912                  694
DESCto201912                  612

(Other                        =            190 640)

TOTAL ALL = 1 295 550


It's nice that you can get a decent corr database without buying it but... What I still really don't get is how Chessbase corr database 2020 (over 1.6 million games - if anyone knows the exact number please share) and Opening Master (1 924 684) have so many games. Any ideas?

I also have ICCF old archives and Lechenicher No engine games that I did not include but that's only 96 586 and 32 858 games respectively.

Confused_by_Theory wrote on 06/28/20 at 08:31:40:

To come back to this. By now I've realised that in Chessbase 15 there is a function for this. Not sure how things were previously. Anyway you open the .cbh file in chessbase, select games you want, click "file" near top left and go to selection to text file then it says format and there are two places to check in PGN and if you do that the exported file becomes .pgn

Confused_by_Theory wrote on 09/26/20 at 15:39:09:

To update on tournament progress. Games progressing less slowly now. Managed to lose the black game where I was clearly worse out of the opening. Won two on time from the same opponent. Think I've gotten nearer to more or less clearly winning positions in two white games. Also my experimental game (as black) where the thinking was I was either gonna lose like an idiot or draw very straightforwardly seems to have gone in the latter direction. Edit: In that game I had the position that happened after move 22 in my analysis that was done before move 4 btw

Have a nice day.

Hi.

Ok. I will have to check the individual sources then and add them all up. Quite possible I've missed something in the merging process (or do you have other sources?). Prolly do it when I update though, i.e. when ICCF releases new games.

It's from lechenichers site. They hosted (host?) tournaments for these two other organisations ASPCC and DESC apparently. Only 694+612 games though so not the largest thing. Hopefully they are also not duplicated in the Complete Archive of Games played on LSS. Dunno that actually.

Archive of ASPCC Tournaments played on LSS upto Dec 2019      694      0.1 MB      0.1 MB      87      78      
1/20/20
Archive of DESC Tournaments played on LSS upto Dec 2019      612      0.1 MB      0.1 MB      96      62      
1/21/20

Chessfriend is, I think, an old non-operational site that Lechenicher provides a games archive for.

Agreed that ICCF and Lechenicher are clearly most important.

Have a nice day.

trw wrote on 09/28/20 at 19:35:37:

Cool. I also think it is a good initiative.

Confused_by_Theory wrote on 09/28/20 at 16:36:51:

Ok so I have some numbers now. Sadly don't have corr 2020.

Chessbase corr database 2018
1 431 813 games
of which:
Chessbase corr database 2018 years all games 1804-1990 (up to and including)
176747 games

Chessbase corr database 2020 (€190 )
1 600 000 games minimum ("over 1.6million")

My downloaded database comprised of:
ICCFto202008
Lechenicherto202006
IECGall
Chessfriendall
ASPCC&DESClechenichertournamentsto201912

1 279 611

Imo it looks like most of the discrepancy between what you can download yourself and the comercial corr-base could indeed be non-server events. There is obviously also postal events 1991 and later which didn't make the cut here due to my arbitrary cut off year 1990. That is sort of heartening as old postal games are not the most relevant to have in modern day corr database (at least ones for competitive use only) and it probably means I don't have to buy a new corrbase .

Have a nice day.

Hi again.

Just out of curiosity. I've now merged the ICCF games and Lechenicher (+IECG et al. found on Lechenichers site) into a single database. Can't remember the exact number of games that the database ended up with (1.2-1.3 million asfair) and I am currently on a bus so can't check but it seems to me like there is a few hundred thousand games more in the chessbase corr database (over 1.6 million).

What could these games be? 1804 to ca. 1990 postal games? Or is chessbase just better at finding games generally.
Edit: I actually have 2018 corr base so can check how many non server games myself. Will do later.

Have a nice day.

Perhaps a restriction on the number of games or tournaments that a player can have ongoing at any one time might help to lessen this kind of behavior.

Hi.

MW wrote on 09/26/20 at 18:54:58:

I can obviously only speculate but maybe he had to many games running and did not even want to think about the game in question.

MW wrote on 09/26/20 at 18:54:58:

The ICCF fondness for 4 or more days per move is kinda baffling to basically an outsider like me. Guess it has to do with tradition. Lechenicher has some shorter tournaments I think if you are looking for that kind of thing.

Confused_by_Theory wrote on 09/26/20 at 15:39:09:
Jupp53 wrote on 09/26/20 at 19:12:47:

Ok. So obviously I don't think ICCF has added Lechenicher games to their database that's why I asked. I did a few random samples and it did not appear like it.

Luckily very easy to find Lechenicher played games section on PC! Mobile not so much. Thanks for helping. It looks like I have been missing a lot of games in the corr database I've been using (oops ). There appears to be no less than:

344506 Lechenicher games
166659 IECG
022675 Chessfriend
000694 ASPCC
000612 DESC

bragesjo wrote on 09/27/20 at 11:46:56:

I have heard of the dead man's defense but also doing it in totally equal positions somehow feels even worse to me.

Confused_by_Theory wrote on 09/26/20 at 15:39:09:

1. Don't know.
2. If you are a member there is a menue point, where you can get all games played so far. I do it via laptop. Don't know anything about smartphones.

Hi.

Games progressing slowly atm. Mildly losing interest in a lot of them. Two questions:

Are Lechenicher schachserver games included in the ICCF correspondence database? If not where can they be downloaded? I tried checking Lechenichers site via mobile but could not find

Is it just me or does Leela seem very careful about forcing the play a lot of the time? It likes marginal improvements a lot I feel.
Edit: this sort of ups the difficulty as assessing five different quiet continuations of (according to Leela) roughly equal value is harder than following concrete lines of play and assessing those.

Have a nice day.

Hello.

Quick update. Games are progressing. I have like two black games where I worry. In one I didn't find a good reply to the opponents opening play (sort of highly critical line) and in the other I chose a totally new system where I am either totally fine or totally busted; one can hope the former of these. Conversely I have probably two white games where I should really be better.

Somewhat feel like playing more games. As mentioned earlier correspondence is kind of stimulating.

Have a nice day.

bragesjo wrote on 06/30/20 at 15:58:41:


21 days fyi.

Confused_by_Theory wrote on 06/29/20 at 12:32:43:

I think I'm right in saying that the code enables you to run the software on up to three devices at any time, so you should still be able to use it if your clubmate had only installed it on one or two (that's assuming you have the code, of course; it'll probably be in the CD case rather than on the CD itself).

Hi.

an ordinary chessplayer wrote on 06/28/20 at 22:08:07:

We have a guy who reads the local paper including obituaries every single day so it is rare that chessplayers here dies without us knowing about it. If we know the person who died was connected to our club and would have had some kind of collection of chess items (usually old books...of which there may be many) we typically contact the estate. Families are usually happy if we can help clear the chess stuff from the things left behind.

an ordinary chessplayer wrote on 06/28/20 at 22:08:07:

Ye. Update will be hard I think.

an ordinary chessplayer wrote on 06/28/20 at 22:08:07:

Ok. Interesting. This copy is far after that anyway. The code will have been used already I suppose

Cheers.
/CbT

Hi.

I was thinking converting from .cbh to .pgn. The other way around I got how to do, but thanks anyway.

an ordinary chessplayer wrote on 06/28/20 at 19:46:25:

Got full version of Chessbase so yea. This is good advice for me and it works. Thanks.

Gl with the spare Chessbase copy.
Somewhat ot. Is there any value in old copies of e.g. Chessbase or Megabase etc? Or only with the code included (i.e. unused ones) like your Chessbase copy? We had a person who died of cancer and with that some were left at the club a couple of years ago.

Have a nice day.

Confused_by_Theory wrote on 06/28/20 at 08:31:40:



Yes, when I am gathering games each month. I gather them in pgn. Combine them in pgn. Then use chessbase to switch it to cbh as it works faster with the larger cbh files. Just right click on the pgn file and convert it to cbh format should be the top option.

You generate the tablebases and you point Stockfish to the folder from its options. You can tell it's working if SF starts showing TB evals. Special numbers that show distance to mate etc.

Yes that is the gist of the discussion, after all Magnus has been quoted as saying Leela's moves all make sense. Stock is very strong in latter stages. I think an engine that tries to meet VGA's idea is corchess, which is based on using Stock for a bit longer analysis idea. It's easily found by googling Ivan Ivec, and yes updating Stock is a good thing.

Stockfish (and its derivatives I suppose) can use Syzygy tablebases. The size of 7-man tablebases is many terrabytes but the 6-man databases are much smaller and practical to generate and use. Something like 70gb I think.

When it comes to Stockfish, every now and then I download fresh builds from

https://abrok.eu/stockfish/

So that I am not stuck with the last stable Stockfish. If there is a Stockfish derivative tuned for long analysis you should probably use that, though. And I really hope you can configure the engine properly for your hardware and purpose.

About "engine-strength" openings, I am not sure but take a look at this freeware opening book

https://sites.google.com/site/computerschess/perfect2019books

Disclaimer: I have never used it myself and I have never played an engine-allowed correspondence game, it just *seems* useful for your purposes.

One idea the guys over at Talkchess discuss is using Leela or Alllie until there are about 16 pieces left, then switching to Stockfish or Brainfish. They seem to think it is the best solution for now.