Purpose of correspondence chess in modern era

[MuguJack]

Thanks, jefk! Excellent info. Since I'm not good enough to improve on Stockchess, it seems silly for me to bother echoing moves (aka, playing) in ICCF.

[Cornfed]

well you (Cornfed) could also ask what's the purpose of 'normal' chess;
(or poetry, but that's another subject

For professional chess players, chess is their profession
(although only the top, aka superGM's seem to earn enough
(or for the absolute top, quite a lot)
For hobby players, chess is their hobby, whereby using
your brain for analysis is beneficial both for kids/youngsters,
as for elderly; for the adult in between group, it has an element
of competition, but also social element (despite the high difference (*) in
playing ability and an element of distraction; some people like to learn
a new language, to broaden their view of the world; others
like games, or computer games (especially during corona times)

etc...

The 'Purpose' of 'normal' chess is what is always has been - one on one competition....for fun and/or a challenge...and for profit if one is good enough.

You just don't get that (in the current age of computers) in "Correspondence". It's a dying art and the engine is what has been killing it. The numbers who find it a worthwhile pursuit continue to drop and drop.

To Team Komodo's credit, they have added this and that to their product. Most people use engines to help perfect their opening repertoires...it's largely the GUI's that help with that. Whomever can succeed in converting simple line point evaluations to into winning percentages (ala what Nibbler does for LC0) - for Stockfish, Dragon etc...will have hit upon a goldmine for the average to better than average player, even professional player. Mark my words.

[jefk]

You just don't get that (in the current age of computers) in "Correspondence". It's a dying art and the engine is what has been killing it.

well if you think that, i wasn't clear enough in my messages; i do see the draw problem, and only want to think with LK (and mr Nickel) about changing
rules to lessen this draw problem.
But... i think of it, also with rule changes, the engines will be
very strong in such (correspondence) chess.

The numbers who find it a worthwhile pursuit continue to drop and drop.

including me, for the time being at least (in ICCF)
with my slow machine...

As for your mentioning the need for deep analysis of opening
lines, see (if you hadn't seen it) the video review of GUI's.
Whether you use win percentages and indicate the best moves
with arrows on the board (as in Nibbler) or use centipawn evaluation,
doesn't make so much difference. But multi-pv works best only
with Mcts type of analysis, whereby nowadays such engines
are not the strongest (yet?).
Meanwhile a lot could be improved in Aquarium (with features
as you indicated) and maybe we will see such improvements.
Chessbase certainly has more active programmers (a better
financial situation i suspect) but whether they will go
for such features, remains to be seen; maybe in Chessbase,
(rather than Fritz). Meanwhile, i use some different tools,
depending on the situation, for the cerebellum'merged' book i
use ScidforPC, for Komodo-Mcts i look in an rather old Fritz GUI,
and for combining analysis from different engines in a tree, i
use Aquarium, which sometimes is a bit cumbersome exercise;
Not only because it's not so user-friendly (even although you
look at the manual) but because different engines give different
evals; the resulting tree becomes almost 'brute force' in nature
(not alfa-beta) and thus can expand rather fast. There's a lot
of interactive analysis, and human judgment needed to come
to a final move, in such process, but whether that helps in
the game of chess to simply 'beat' ST may be doubtful indeed
(maybe it's possible in the Catalan or so, we'll see (other thread)

[jefk]

in addition (to my reply to 'cornfed'):

can you specify nrs ? how many are quitting ICCF approximately ?
and in comparison to earlier years ?
And how many are joining ? over a year or so ?

some days ago i brought up the draw issue on Iccf,
and got a response from someone who's now one year on ICCF,
and he wrote that he thinks it's still the individual that matters (*)
he still has difficulty to win or even draw higher rated players,
and he's always happy if he can get a draw against a higher player.

(*) the nature of the (ICCF) game(s) has changed ofcourse,
to me it looks opening theory has become even more important
and if that's not your thing, then you may not like it.
And it's also not simply a matter of looking it up.
Where do you want to look it up ? GM statistics ? well
that may not be good enough

PS btw ,do you have played yourself at ICCF ( just wondering) ?
(or just read here eg. that U.Blass already stopped with it
some years ago ? etc. and drew your conclusions?)

[Cornfed]

You just don't get that (in the current age of computers) in "Correspondence". It's a dying art and the engine is what has been killing it.

As for your mentioning the need for deep analysis of opening
lines, see (if you hadn't seen it) the video review of GUI's.
Whether you use win percentages and indicate the best moves
with arrows on the board (as in Nibbler) or use centipawn evaluation,
doesn't make so much difference. But multi-pv works best only
with Mcts type of analysis, whereby nowadays such engines
are not the strongest (yet?).

My Reply....just having morning coffee and can't figure out what I've done to pull it out of the part of your quote.:

MCTS - so what about their not being 'the strongest'? That is kind of irrelevant - like saying a Cheetah is faster than a Puma. Both are more than fast enough to 'get the job done'. They just have different environments they operate in. It sure seems that Komodo MCTS (or Dragon) is pretty darn strong. If that kind of program could actually give what Nibbler does for its program, Dragon (MCTS version) instead of kind of hiding the info and basically leaving one to 'guestimate' winning percentages, it would be invaluable to those who use it for analysis - particularly of lines to play in head to head (non-computer) games.

My final ICCF rating was 2399...right at about 20 yrs ago. Just as people were beginning to use engines...saw the writing on the wall. I have mentioned before I am 1.5/2 in head to head OTB against a well known Correspondence GM. I've not talked with him about Correspondence play these days but my record 'head to head' further cements in my mind that idea that any (not saying him) 1900 player can play 'like a GM' with their little helpers.

[jefk]

cornfed wrote

MCTS - so what about their not being 'the strongest'? That is kind of irrelevant - like saying a Cheetah is faster than a Puma. Both are more than fast enough to 'get the job done'.

depends on the job, if you like to quick assistance, to get
good drawing chances, ok, that seems true in the age of Nnue
(for an engine as SF).
But to win a game (even with White), you need more thorough
analysis, as known with the current correspondence players
who also are quite computer savvy (i.e. in the area of computer chess
and opening theory (with the aid of computers programs).

Anyway, you answered my question, namely, that you are not an ICCF player (anymore), clear enough.

As for Mcts, imho it can help to avoid the socalled 'horizon problem|'.
While Sf *seems* to calculate deeply, imho that's based
on too much selectiveness (they should improve this
for the current Nnue versions, i think). Example:
Uri B in the 'other thread' gives a Bogo-Indian line
at ply 59 or so, where White still plays Bd2. Komodo
(especially Mcts, at least on my 'slow' comp), seems
to like Nd2!> (in positions without ...d5),w hich is why i
went to the fianchetto position (with g3) to be specific.

From your text (for colour blind people or b/w screens)
it won't make a difference if you write in red, btw) it's
clear you're a very good chess player, but you're still human,
you know. Top Gms are still below 2900, top engines now around
3500 or so. The max might be approaching 4000 maybe,
just around 3930 or so, who knows.

In Shogi, under certain rules, they count material for
both sides after a draw situation, to decide upon a winner.
In future correspondence chess we might do the same
i.e. something similar) (whereby we should probably give
White less time, to compensate for it's first move advantage).
Also then It will become more like a computer (analysis)
'contest, and some (many?) don't like that, but anyway
you can't hold progress, i don't think ICCF is going to
ban the use of computers (because as some others also
wrote, it's hard to check for computer use anyway,
especially in such slow) conditions.

Meanwhile correspondence chess with the use of computers
still is at the forefront of chess research, for many GM's
opening theory is important, and they look at ICCF games,
at least at level 2400 + or so). Hey some even (frequently in fact)
write books about certain opening variations; do you think they
all are going to stop with that now, with SF Nnue ? don't think so.

PS and while we waiting for possible new (endgame/adjudication) rules
(while not an expert on Nnue), i would suggest SF runs the testgames
(to develop better Nnue's) without endgame syzygy bases, As it
is now, SF tends to give 0.0 eval after a while (even at my 'slow'
comp). Komodo seems to do this better, and gives better differentation;
ofcourse with current rules you sometimes still can't win a 'better'
endgame, but that's not the point; (human) GM's ofcourse always
would prefer to come in a 'better' endgame, rather than defending
a worse endgame but in the post mortem seeing that SF considered
it a draw anyway; hope you see what i mean (otherwise some others)
although my purpose here is not to give lectures about computer chess.
(just reacted a while ago to the LK posting about drawing in Nnue times)

[Cornfed]

I’ll see if I can wrap up my thoughts here because we are kind of talking past one another. My words come from the viewpoint that ‘Correspondence Chess’ as a past time (pursuit) has largely…passed its onsale date.

When a (just to pick numbers- one could choose others) someone with the talent of an 1800 level OTB player can use ever more powerful CPU’s and engines and GUI’s like Aquarium, and essentially play like a 2800 and be playing someone who might have a 2300 OTB understanding/ability at the game who might even have less power toys to play with…the ‘sporting element’ by most any definition is largely gone.
One might as well dedicate their lives to racing electronic toy cars against each other and finding some satisfaction in pushing their new Ultra Primo Matchbox rig past someone's Mach V Sport rig to victory…and think they have accomplished something. YMMV…

My interest lie in using ‘toys’ to find good ideas (and the chance to steer my opponent towards ever worsening position!) in the (non-correspondence) games I play OTB and online.

THAT kind of help – which I think would be very fruitful in the guise of say a Nibbler/LCO for engines like Komodo or Stockfish (percentage chance to win…MCTS) is quite a bit different from their in ‘Correspondence Chess’. One can use the toys to prep to try to get an edge…but we still have to suit up naked and play the game with our own brains.

[towforce]

Method A: get an expensive computer and run it long enough to select unbeatable moves

Method B: anything different from method A

For method B players, the game lives on - but there are two risks:

1. Your opponent might switch from method B to method A

2. In the fullness of time, "expensive computer" becomes "cheap computer" and "long enough" becomes "not very long"

[jefk]

cornfed,

well i see your point, corresp. chess has been an interesting
distraction for me, and in the meantime i learned more about
opening theory; but Iccf chess may become meaningless,
and possibly later-on obsolete,
As for normal chess, no comments, while i'm working on
(an update of) a handicap system; most online nowadays
is blitz or rapid, whereby computer cheating is possible
as well, i guess, if you do it in a sophisticated way to
avoid the detection systems as eg with chess.com>
This doesn't mean online chess will become obsolete,
ofcourse; especially not (later) with handicaps !


towforce,

correspondence chess with handicaps may also become an
option. But then as human player (among 'cheaters') we
shouldn't have any ambition(s) to get a higher rating, i.e.
get into a higher handicap 'class', it would become more
like fun game, eg. to practice you analysis skills with more
time, (and possibly endgame skills and knowledge).
just my2cnts and ymmv indeed

We'll see how it goes, adoption of a new possible variant
(as eg. chess960 or zh) takes quite some time, and most
variants don't become popular at all. Personally i think
crazyhouse aka zh with handicaps is an interesting idea,
with also a 'default' handicap to compensate the firstmove
advantage (as i hinted (*) before, for White without the f2
and for Black, c7)
(*) with SF-zh removing the a3 pawn was enough but for
two better engines as Pulsar and CrazyAra my suggestion
(without f2/c7) looked more equal (as in 'normal' chess btw)

[MikeB]

The main question: What is the point of correspondence chess in the computer era?

To kill some time before we kick the bucket and say "hello" to our maker?