Freestyle Chess is Coming Home

[Ozymandias]

There are unsporting things. Friends could give each other points away. Maybe you can design the draw after the qualification so that friendly players (where this is known) have to play against each other in the first round (14 rounds = 28 games).

The only way to prevent friends who give away points can be by changing the competition.

I believe that the competition should be simply about beating pure chess engines.

Every participant play against some engines(stockfish,Dragon,Lc0 and more players) and the winner of the tournament is the participant who get more points againt them.

That'd be fine by me, but I don't expect to see anything of the sort in this life. There's also the question about how many wins (if any) could be scored against top engines, which is why your proposal sounds interesting ("more players").

[Steamhammer]

Freestyle only against pure engines could be another interesting project, but it aims in a different direction., I as an organizer want to strengthen the human factor in computer chess, which is why I prefer freestyle without pure engines. Apart from that, the alternative project would have some difficult questions to clarify:
Which hardware for the engine? Which book? Which settings? Time control? And there is another problem: If it will not be possible that all players compete at the same time, those who play later would have an advantage as they have more information (due to the games already finished).

Here is some more information on the postponed qualification tour (starting on 13th November - see www.infinitychess.com). I quote from the latest announcement:
" ...we have postponed the start to November 13/14 for scheduling reasons and because the number of registrations so far has been too low (14 + "x", i.e. some are still considering ...) and have modified the further schedule.
However, it does seem that a number of factors stand in the way of the hoped-for greater participation:

1. the exclusion of engine-only players, i.e., those who want their engines to play automatically. It was interesting to compare how the "pure" engines did against the centaurs in the earlier freestyle tournaments, but now that I think the goal is to strengthen the human factor in conjunction with engines, this would be counterproductive.

2. many probably believe that engines are too strong today for centaurs to be able to set noteworthy accents and score successes as opposed to pure engine play. Such objections are not easy to dismiss, but there are counter-arguments: a) the endurance test in long games (in our case 60 min + 30s per move) has not been provided yet - we are just bringing it; b) similar to correspondence chess, the different human playing strengths (and that includes all aspects of a chess match) should be an interesting factor. We are practically faced with the dilemma, to put it bluntly, that many chess computer afinados don't play chess themselves anymore - or don't trust themselves chess-wise either - and that, on the other hand, strong chess players often have no relation to computer chess but use the engines only as an analysis tool.

3. the former relatively high cash prizes in freestyle tournaments have attracted many gamblers who don't want to make the effort for smaller amounts.

Well, I think we will try, even with a lower participation if necessary, to get the project "Freestyle League" going, on the one hand because the participants have fun with it, and on the other hand to demonstrate to the chess world that many interesting chess matches are put on the board, which can be found neither in pure computer chess nor in pure human chess and which radiate a liveliness all of their own, especially when the participants also report regularly about their experiences. It is not about highest perfection at any price, but about the old Laskerian idea of chess combat transported into the digital age.

Everyone who wants to participate is welcome, and a maximum of 7 weekend dates spread over the year should not be a problem for most people."

Kind regards,
Arno Nickel
(TD JoJo)

[Ozymandias]

Hi Arno

First let me tell you I will be most likely taking part, just today I bought the parts to repair the PC that died this summer; as soon as I have everything working I'll register. (DB is up to date already).

As for your questions:

- HW: TCEC performance level, but avoiding their wild expending.
- Book: very short, no more than 8ply. Preferably forum-agreed.
- Settings: as close as TCEC as possible.
- TC: 90+30. A human TC any player is familiar with.
- Earlier dates could give you tie-break points.

Lastly, about the article:

- "the endurance test in long games (in our case 60 min + 30s per move) has not been provided yet" Performance in the ICUC was similar between centaurs and engines. For example, without the unexplained loss, CoffeeOne had one of the top performances (even ahead of Zor or Akhtar).

- no mention of the draw problem? I'd say this is the single most important reason for the low expectations raised so far. Sure a bigger prize-bag brings people over, but with draws being what they are, it could probably attract the wrong kind of player, if you get my meaning.

[Steamhammer]

Hi Juan,
Thank you for your very specific feedback and your support for our project. I am also in conversation with a few others who still want to register, whereby various aspects come up. Mainly there are scheduling problems. Some players fear that they will not be able to promise any dates for the Freestyle League 2022 in the long term.
We are therefore thinking about allowing more alternative dates before the respective round (I suggest up to 25% per player). Another possibility to reduce the deadline pressure would be that the players would be allowed to play at least some games (as a rare exception) "engine only". In my opinion, the total amount of such matches should not exceed 10%.
As far as the problem of the high draw rate is concerned, for the time being we can only address this within the framework of the tie-break rules (which have to be updated, same as the tournament rules alltogether), i.e., as in some previous tournaments, reward stalemates and material victories (against the bare king) with bonus points. Unfortunately, our tournament software does not recognize 3/4 points. However, we could possibly go a step further with the league launch in spring.
(btw, in a correspondence chess world championship round, in addition to a regular win, I was able to forcibly end a game with K + N vs. K in order to demnstrate an advantage. I will also succeed with K + L vs. K. So it doesn't have to be be so rare when you already have a certain advantage.) So much for today, Arno.

[carldaman]

I'm afraid you'll be swimming and likely drowning in a sea of draws, unless you set it up as Centaur vs Pure Engine.
However, there's a twist I recommend - the pure engine(s) should be some 100+Elo below the top current engine!
This will both reduce draws and reward skill.

If today's top engines (SF in particular) have a serious flaw it is that they overvalue a safety-first approach.
This serves them well against similar strength opponents but allows too many draws vs weaker ones, and
much more so with Black, from the normal starting position.

A skilled centaur should be able to overcome this limitation and produce more wins against lesser engine opposition
than an unassisted top engine.

[Ozymandias]

Hi Juan,
Thank you for your very specific feedback and your support for our project. I am also in conversation with a few others who still want to register, whereby various aspects come up. Mainly there are scheduling problems. Some players fear that they will not be able to promise any dates for the Freestyle League 2022 in the long term.
We are therefore thinking about allowing more alternative dates before the respective round (I suggest up to 25% per player). Another possibility to reduce the deadline pressure would be that the players would be allowed to play at least some games (as a rare exception) "engine only". In my opinion, the total amount of such matches should not exceed 10%.
As far as the problem of the high draw rate is concerned, for the time being we can only address this within the framework of the tie-break rules (which have to be updated, same as the tournament rules alltogether), i.e., as in some previous tournaments, reward stalemates and material victories (against the bare king) with bonus points. Unfortunately, our tournament software does not recognize 3/4 points. However, we could possibly go a step further with the league launch in spring.
(btw, in a correspondence chess world championship round, in addition to a regular win, I was able to forcibly end a game with K + N vs. K in order to demnstrate an advantage. I will also succeed with K + L vs. K. So it doesn't have to be be so rare when you already have a certain advantage.) So much for today, Arno.

It's not difficult to predict multiple players being tied with the same points, so using a different scoring system, if only for tie-break purposes, is a much bigger improvement than what it would appear. Happy to hear about this.

As for playing at different days, I would give as much freedom as a player asks for, but use the number of "asks" as a secondary tie-break rule. Number of games played as engine should be the 1st tie-break rule if someone insists on playing that way.

[Steamhammer]

I have just published a 10-point paper with changes to the concept of the Freestyle Chess League 2022 on the InfinityChess website:
"After intensive discussions and some interesting suggestions for changes from freestyle chess players, we have decided the following changes to the concept of the Freestyle League 2022:
1. The thinking time will be doubled to give centaurs more time for analysis during the game and to reduce the time pressure on the players. Each player will get two hours for the game and 30 seconds per move.
2. Only one game per day is played. Regular game days are Sundays only, in order to avoid scheduling problems whenever possible and so that not always a whole weekend is affected.
3. As a consequence of the new time regulations, changes in the schedule and possibly in the class division are also necessary. It will not be possible (as previously planned) to play double round robin with 15 participants in one class.
4. Which tournament format will finally be chosen, on the other hand, also depends on how many players participate in the qualification tournament and then later register for the league.
5. It will most likely be a tournament format with aximal 14 and minimal 10 games with equal number of white and black games.
6. The match dates will be evenly distributed over the year 2022. However, it will be possible to move up to 25-30% of the games forward by agreement with the opponent.
7. In order to get a reference value for the comparison engine-only and centaur, an account with the current Stockfish version will play in each class. The exact details for hardware, settings and opening book still have to be worked out.
8. To further reduce the load on the players, everyone will be allowed to play up to a maximum of 4 of his games per tournament in engine mode (=engine-only). However, he accepts that in case of a later tie-break he will receive a worse fine rating in relation to players with the same number of points who have played fewer or no games at all in engine mode.
9. The mentioned changes under 1-7 do not affect the qualifying tournament, unlike point 8, which is also intended to provide relief here.
10. In view of the extremely high draw rate to be expected, our top tie-break criterion in all tournaments will be the number of possible "stalemate wins" and "material wins" against the bare king, as we have already applied this rule in the previous "Centaur Weekend Tournaments"."

http://www.infinitychess.com/Page/Publi ... spx?id=346

[Eduard]

Translate:
In my opinion, 10 to 14 games are not enough to determine a worthy champion. The draw rate is high and whoever loses a game of 10 has hardly any chances. With 28 games it looks different! With more games (e.g. 28) you can take more risks than with just 10 or 14.

Why should the level be increased to 2h? On the one hand you want to have more time, but on the other hand you agree that 4 games can be played in engine mode. I do not like it. If engines are allowed in auto mode (I am against it but I would still participate), then those players who want that should only play that way. All others should only be allowed to use the Centaur mode.

We don't play correspondence chess or engine chess.

Freestyle is a discipline in its own right, and there must be time pressure. Level 60 min is good. Such a game actually lasts 2.5 to 3 hours. 4 games are possible on the weekend. This shows the mental strength of the players. If you can't do that, you'd better play correspondence chess or leave engines running. I am already over 60 years old and sometimes I still play non-stop on Lichess for up to 14 hours. I also analyze non-stop for up to 10 hours without any problems. I even know people over 70 who play tournaments that last up to 9 hours a day. If I can do that at my age, then 2 games a day (approx. 3h + 3h) should be possible for everyone. This also allows more games, and with 28 games, everyone has the chance to catch up even after defeat.

More and more concessions are made, just so that the pure engine player as well as long-term thinkers corr. Players come their wishes. Just so that you get more players. The spirit of freestyle is sacrificed (time pressure, fight, Centaur mode).

Original message:

Meiner Meinung nach sind 10 bis 14 Spiele zu wenig, um einen würdigen Champion zu ermitteln. Die Remisquote ist hoch und wer eine Partie von 10 verliert, hat kaum noch Chancen. Bei 28 Partien sieht es anders aus! Bei mehr Partien (z.B. 28) kann man mehr Risiko gehen als bei nur 10 oder 14.

Wozu sollte die Stufe auf 2h erhöht werden? Einerseits möchtest Du mehr Zeit haben, andererseits aber stimmst du zu, dass 4 Partien im Enginemodus gespielt werden dürfen. Ich finde das nicht gut. Sollten Engines im Automodus erlaubt sein (ich bin dagegen aber ich würde trotzdem teilnehmen), dann sollten solche Spieler die das wollen, nur so spielen. Alle anderen sollten nur den Centaur Modus nutzen dürfen.

Wir spielen weder Fernschach noch Engineschach.

Freestyle ist eine eigene Disziplin, und der Zeitdruck muss vorhanden sein. Stufe 60 min ist gut. Effektiv dauert eine solche Partie 2,5 bis 3 Stunden. Es sind 4 Partien am Wochenende möglich. Hier zeigt sich die mentale Stärke der Spieler. Wer das nicht schafft, sollte besser Fernschach spielen, oder Engines laufen lassen. Ich bin schon über 60 Jahre alt, und spiele manchmal auf Lichess trotzdem nonstop bis zu 14 Stunden. Ich analysiere auch problemlos nonstop bis zu 10 Stunden. Ich kenne sogar über 70 Jährige, die spielen Turniere die am Tag bis zu 9 Stunden andauern. Wenn ich das schaffe in meinem Alter, dann sollten 2 Spiele am Tag (ca. 3h+3h) möglich sein für jeden. Das erlaubt zudem mehr Spiele, und bietet bei 28 Spielen, jedem die Chance, auch nach Niederlagen noch aufholen zu können.

Es werden immer mehr Zugeständnisse gemacht, nur damit auch die pure Enginespieler sowie Langdenker corr. Spieler auf ihre Kosten kommen. Nur damit man damit mehr Mitspieler bekommt. Der Geist des Freestyles wird geopfert (Zeitdruck, Kampf, Centaur modus).

[Eduard]

I have no problem playing long games. I analyze a lot. I enjoy it. I analyze almost every day. The other day I forgot the time, at 3 p.m. I had started analyzing and at 10 p.m. I looked at my watch and was amazed at how quickly the time passed.

I would like to say in advance that there will almost only be draw games if the time is increased. I am already desperate to find a way to win with much less time. Does anyone really believe that with 2 hours to think about it, they will win more games? Not me. Better to play 2 games at a time with 1h + 30s.

In 10 to 14 games there will be many first places, and then those who have bonus points (fine scoring) win, earlier winners have an advantage. I think it is possible that all games will end in a draw.

To be honest, I don't feel like doing that. I've heard that one or the other correspondence chess player would like to take part, but allegedly has time problems. Then why not play correspondence chess straight away?

[Eduard]

My strategy will be this:

Because, with only 10 to 14 games (28 were planned), as a loser in a single game you have hardly any chances, the tournament is almost over. Or you play hara-kiri chess because you have to! It is better to play a few rounds of draws first (I know a lot of safe variants), and then see where you stand. Why take risks right away?

What do you think about it?