Breaking the 3800 Barrier

This is post no.3800 on this blog, meaning that it's time for another 'Breaking the Barrier' post. The previous post, Breaking the 3700 Barrier (February 2024), was a little more than a year ago. At that time I noted,

We have one engine [rated] over 3800 (Stockfish 16), 15 engines over 3700, and (not shown) 11 engines over 3600. [...] See you a year from now for the '3800 Barrier'.

The chart below shows the current ratings of the top engines as calculated by the same source.

CCRL Blitz Rating List (Feb 2025)
(computerchess.org.uk/ccrl)

Now we have two engines rated over 3800 (Stockfish 17 and Torch v2), 29 engines over 3700 (not all shown), and 14 engines over 3600. If I keep posting to this blog at the current frequency, I'll be able to do 'Breaking the 3900 Barrier' in a year. Given that Stockfish, the top engine (rated 3817), has only increased 11 points since last year, I don't expect to see any engines over 3900 anytime soon.

Human players have reached a so-far insurmountable barrier of 2900. Are engines facing a similar barrier of 3900?

Breaking the 3700 Barrier

The post from earlier this week, The AI Generator (January 2024), was no.3700 on this blog, which means it's time for another 'Breaking the Barrier' post. In the previous post for the series, Breaking the 3600 Barrier (April 2023), I wrote,

Getting back to the 'CCRL Blitz Rating List' there are currently 17 engines rated over 3600. Based on that same list, it looks like we're already good for a 'Breaking 3700' post seven months (or so) from now, which should have at least two engines.

Make that nine months -- because I'm slowing down -- and let's have a look at the latest CCRL rating list.

CCRL Blitz Rating List (Feb 2024)
(computerchess.org.uk/ccrl)

Now we have one engine over 3800 (Stockfish 16), 15 engines over 3700, and (not shown) 11 engines over 3600. It looks statistically strange to have more engines rated 3700 than 3600, but that's what the numbers say. See you a year from now for the '3800 Barrier' -- unless my slowdown grinds to a halt.

TCEC/CCC 2023-H1 Summary

Six months have passed since the previous summary of the fortnightly TCEC/CCC posts on this blog. That post was appropriately titled TCEC/CCC 2022-H2 Summary (January 2023). Here are the latest six months of posts that covered the first half of 2023.

TCEC/CCC	Off-week
2023-01-09: TCEC Cup 11 Underway; CCC19 Bullet in Semifinals	2023-01-16: Chessify Progress Report
2023-01-23: LCZero Wins TCEC Cup 11; Stockfish Wins CCC19 Bullet	2023-01-30: Talkchess Talks Top Engines
2023-02-06: TCEC Swiss 4 Underway; CCC Mystery Matches	2023-02-13: Chessify's Other Engines
2023-02-20: Stockfish Wins TCEC Swiss 4; TCEC S24 L2 & CCC20 Blitz Underway	2023-02-27: Chessify vs. Chessbase Engine Cloud
2023-03-06: TCEC S24 L1, CCC20 Blitz Semifinals: Final Week	2023-03-13: NextChessMove.com
2023-03-20: TCEC S24 DivP, CCC20 Blitz Final : Both Underway	2023-03-27: Chessify Resources
2023-04-03: TCEC S24 DivP Finishing; Stockfish Wins CCC20 Blitz Final	2023-04-10: OpenBench Interlude
2023-04-17: TCEC S24 Sufi, CCC20 Rapid - Both Underway	2023-04-24: The Komodo Baton
2023-05-15: Stockfish Wins TCEC S24; Engine Mysteries	2023-05-22: An 'Advice-taking' Chess Machine
2023-05-29: TCEC FRC6, CCC20 Rapid Finals; Both Underway	2023-06-05: The CCC's Mystery Engine
2023-06-12: TCEC FRC6, CCC20 Rapid; Stockfish Wins Both	2023-06-19: Chessify Coins and Combos
2023-06-26: Stockfish Wins TCEC DFRC2; CCC20 Bullet Underway

The last cell in the table represents this current post. In the previous post, 'TCEC DFRC2, CCC20 Bullet', I wrote,

What goes around doesn't always come around. Sometimes it stops. The TCEC is starting S25; the CCC is coming to the end of CCC20. I'm almost certain this current post will be the last in the series on this blog.

Make that the next-to-last post. This summary is the last post.

Stockfish Wins TCEC DFRC2; CCC20 Bullet Underway

In this continuing, fortnightly series on the world's top two engine vs. engine competitions, the previous post was TCEC FRC6, CCC20 Rapid; Stockfish Wins Both (June 2023). Expanding that title produces the following summary:-

TCEC: In the 'S24 FRC6 Final', Stockfish beat LCZero +15-12=23. The site then organized 'S24 DFRC2', which has reached 'League 1', the third of four stages. • CCC: In the 'CCC20 Rapid Finals', Stockfish crushed Lc0 143.0-107.0, then crushed the Mystery engine 67.0-33.0 in an exhibition match. The site then launched the 'CCC20 Bullet Entry League'.

Fast forward two weeks and the current situation looks like this...

TCEC: Stockfish beat LCZero +10-7=33 in the 'S24 DFRC2 Final'. This was the final event for TCEC season 24 (S24). The plans for S25 announce,

Now !Swiss5; then !1day, Cup 12, !1day, Leagues, !1day, !FRD1, !1day, !S26; (current plan, can change, also !r-mobility (armageddon) demo or bonus might be inserted somewhere); !1day per move bonus done in alternate and parallel with other tournaments (ongoing)

What's FRD? The !Command says,

Fischer Random Double event, combining FRC and DFRC in one event: FRC in early part, DFRC for later and final parts.

This is a new form of engine competition. It's unclear what issue it is trying to resolve.

CCC: In the 'CCC20 Bullet' tournment, the 'Entry League' was followed by 'Qualifier 1' and 'Qualifier 2'. Before launching the 'CCC20 Bullet Main' event, the site is conducting a match between Dragon and the Mystery engine.

Given that the sequence for both CCC19 and CCC20 was Blitz, Rapid, Bullet, we can expect that the next event will be 'CCC21 Blitz'. Bold prediction: Stockfish will win both 'CCC20 Bullet' and 'CCC21 Blitz'. Timid prediction: Stockfish will win both 'CCC20 Bullet' and 'CCC21 Blitz'.

...According to my first summary of engine vs. engine events, TCEC/CCC 2019 Q1-Q3 Summary (October 2019), I've been doing this series of posts since January 2019. Back then I was following TCEC S14 and CCC03.

What goes around doesn't always come around. Sometimes it stops. The TCEC is starting S25; the CCC is coming to the end of CCC20. I'm almost certain this current post will be the last in the series on this blog.

[For further information from the various stakeholders in the engine-to-engine events, see the tab 'TCEC/CCC Links' at the top of this page. • NB: Leela = LC0 = LCzero; Dragon = KomodoDragon]

Chessify Coins and Combos

The last time I posted about Chessify was Chessify Resources (March 2023; 'Here are a few resources to help explore the Chessify service.'). Since then I've learned much about the service and one of these days might document most my recent experiences on this blog.

An important aspect that has eluded me is the Chessify price structure -- plans, coins, and combos. What is the difference between them and how do they work together? Although I had watched the video embedded below, it introduced so many new concepts that I had trouble understanding its basic point.

Chessify Subscription Plans - Price, Features, and More (3:31) • '[Published on] Feb 12, 2023'

The video description says,

Learn everything you need to know about the Chessify subscription plans: price, features, differences, etc. The plans offer access to powerful cloud servers for chess engine analysis, a wide range of chess engines, such as Stockfish and Leela Chess Zero, a 9 Million+ game database updated weekly, opening references, cloud storage, and more.

The video uses a chart which I've reproduced at the bottom of the post. I'm currently on the least expensive of the subscription services, the 'Amateur' plan. Is it worth my while to upgrade to a higher plan? And what is the purpose of 'coins'? Fortunately, there's another video to answer these questions: What are the Chessify Coins and Combo Plans? (youtube.com; Chessify AI).

Here's everything you need to know about the Chessify coin packages and combo plans which offer both coins and standard subscription plan features. The coins give you access to up to 1 BN/s speed cloud servers for analyzing with the strongest chess engines, such as Stockfish and Leela Chess Zero, while the combo plans also offer a 9 Million+ game database updated weekly, opening references, cloud storage, and more.

The FAQ on the pricing page explains that subscription plans are for shared servers and coins are for dedicated servers. When I started using Chessify, I was given 500 coins to help debug a blocking problem I was having. I still have 300 unspent coins, because I didn't know what to do with them. Now I know.

***

Source: Chessify Subscription Plans and Coins
(chessify.me/pricing)

TCEC FRC6, CCC20 Rapid; Stockfish Wins Both

It's time for another fortnightly report on the world's top two engine vs. engine chess competitions. For the previous report, see TCEC FRC6, CCC20 Rapid Finals; Both Underway (May 2023). Following is a summary of that report.

TCEC: The site launched a chess960 event, 'S24 FRC6', which is currently in the 'First phase - League C', the third of four preliminary six-engine double round robins. • CCC: The engine Mystery played five 100-game matches, In the 'CCC20 Rapid Challenger Match' Lc0 beat Dragon 129.5-120.5, and is currently losing to Stockfish by seven points in the 'Finals' after a little more than a quarter of the 250 games have been played.

Since then the TCEC FRC6 event has morphed into the DFRC2, and the CCC Rapid has morphed into the Bullet. Same old, same old? With Stockfish continuing to win everything in sight, it sure seems that way.

TCEC: In the 'S24 FRC6 Final', Stockfish beat LCZero +15-12=23. The site then organized 'S24 DFRC2', which has reached 'League 1', the third of four stages. For more about the event, see TCEC DFRC 2 (wiki.chessdom.org), and TCEC DFRC rules (ditto). The rules state,

The TCEC DFRC is a progressive leagues knockout format played in TCEC, including all invited qualifying participants. The qualifying engines consist of the top engines able to play Double Fischer Random Chess natively under Linux.

For the previous TCEC iterations of FRC6 and DFRC2 on this blog, see Stockfish Wins TCEC FRC5, Leads CCC18 Rapid Semifinal (August 2022), and Stockfish Wins TCEC DFRC1, Leads CCC18 Rapid Final (August 2022). I'll have more to say about the more recent iterations on my chess960 blog, which is linked on the sidebar.

CCC: In the 'CCC20 Rapid Finals', Stockfish crushed Lc0 143.0-107.0. The Fish then crushed the Mystery engine 67.0-33.0, losing only three games along the way. The site then launched the 'CCC20 Bullet Entry League'. For the previous CCC iteration of its Bullet event, see LCZero Wins TCEC Cup 11; Stockfish Wins CCC19 Bullet (January 2023).

[For further information from the various stakeholders in the engine-to-engine events, see the tab 'TCEC/CCC Links' at the top of this page. • NB: Leela = LC0 = LCzero; Dragon = KomodoDragon]

The CCC's Mystery Engine

In last week's post on engine vs. engine events, TCEC FRC6, CCC20 Rapid Finals; Both Underway (May 2023), for the CCC side of the report I noted,

Mystery played five 100-game matches, beating Koivisto, Berserk, and Ethereal; then losing to Dragon and Lc0.

Mystery played its first CCC game in the CCC20 Rapid: Entry League (chess.com). On the CCC's Discord forum, the early discussion was brief:-

Q: Mystery engine? Or is the engine actually called Mystery? • A: Probably a new Chess.com bot.

That's also what I would have guessed, but there's more to the story. The engine, a last minute CCC20 replacement for 'Demolito : Filling out the 12th spot', qualified into the next two 'CCC20 Rapid' stages, Qualifier #1 (chess.com), and Qualifier #2 (ditto). A handful of Mystery's 'Entry League' games were recorded in Mystery chess engine enters in [CCC]?! (ditto).

Fast forward to mid-May and we find Unknown "Mystery" Engine (chess.stackexchange.com). The writer observed,

One of the strangest things is that at 'CCC20 Rapid: Qualifier #2' it performed poorly. But at the same time, in a one-on-one battle with the "Berserker", it won confidently. And now beats "Ethereal", one of the strongest engines.

Someone commented,

Andrew Grant, the tournament director of CCC and author of Ethereal, seems to be the only one who knows Mystery's true identity. He did clarify that it was not a dumbed down version of Stockfish or Dragon as that would be rather lame. It seems to be an older engine (Maybe Laser?) with NNUE added to it but no tablebase support.

A few days later, the subject came up in one of the leading computer chess forums, Mystery engine at CCC (talkchess.com). The originator of the thread used the engine's UCI log to make an educated guess:-

The most obvious candidate would be a new engine developed by Chess.com as a successor of Komodo Dragon after their purchase of KomodoChess. The [UCI] options don't seem to match any known engine found on GitHub.

Some observers say that interest in chess engine events has been declining lately. At worst, the CCC is using the Mystery engine to rekindle interest. At best, it's a marketing strategy to introduce a new engine. I'm sure that we will soon learn much more.

TCEC FRC6, CCC20 Rapid Finals; Both Underway

The previous report on the world's top two engine vs. engine competitions was Stockfish Wins TCEC S24; Engine Mysteries (May 2023). To summarize that post:-

TCEC: Stockfish beat LCZero +20-16=64 to win the S24 Sufi. The site then conducted the following events: S24 VVLTC Bonus, S24 Kibitzer Bonus, and S24 4k 2, It is currently running the 'S24 4k 2 Final' • CCC: The site ran two 'CCC20 Rapid' events, a 12-engine 'Main' and a 6-engine 'Semifinal'. The 'Semifinal' was won by Stockfish, ahead of Lc0 and Dragon. Stockfish qualified for the 'Final', while the other two engines will play a 'Challenger Match' to determine the second finalist. Then followed an interlude of a series of 'Mystery' matches. Mystery beat Koivisto by a wide margin and is currently leading Berserk.

The report mentioned 'a couple of mysteries', which I still haven't resolved. Following is a roundup of the current situation.

TCEC: In the 'S24 4k 2 Final', ice4 beat 4ku +5-4=15. The site then launched a chess960 event, 'S24 FRC6'. For more about the event, see TCEC FRC 6 and TCEC FRC rules. The event is currently in the 'First phase - League C', the third of four preliminary six-engine double round robins.

CCC: Mystery played five 100-game matches, beating Koivisto, Berserk, and Ethereal; then losing to Dragon and Lc0. In the 250-game 'CCC20 Rapid Challenger Match' Lc0 beat Dragon 129.5-120.5, and is currently losing to Stockfish by seven points in the 'Finals' after a little more than a quarter of the 250 games have been played.

[For further information from the various stakeholders in the engine-to-engine events, see the tab 'TCEC/CCC Links' at the top of this page. • NB: Leela = LC0 = LCzero; Dragon = KomodoDragon]

An 'Advice-taking' Chess Machine

While working on the recent post, May 1973 & 1998 'On the Cover' (May 2023), I discovered a full-page ad for Scientific American in the May 1973 issue of Chess Life & Review. The ad is pictured below.

'Who has white - the man or the computer?'

The first paragraphs promoted an article on computer chess.

The game shown here is between senior master Charles Kalme and an IBM 370/155 computer. Can you tell which player has white?

The game is reprinted from the article, "An Advice-Taking Chess Computer," in the June issue of SCIENTIFIC AMERICAN -- now on sale at leading newsstands. In the article, Albert Zobrist and Frederic Carlson of the University of Southern California explain their development of a whole new dimension in programming computers to play chess. Under the Zobrist - Carlson program a chess master can transfer his knowledge of the game to the computer. Their "advice-taking" chess machine, moreover, is capable of generalizing from its own mistakes.

That 'a chess master can transfer his knowledge of the game to the computer' isn't as surprising in 2023 as it must have been in 1973. These days the problem is the opposite -- AI engines have trouble transfering their knowledge of the game to a human master. That a 'chess machine is capable of generalizing from its own mistakes' sounds like a stretch back in 1973.

The next paragraph was a summary of the Levy wager. It's worth remembering that the 1973 interest in chess computers was taking place during the chess boom following the 1972 Fischer - Spassky match.

THE £1000 WAGER • International master David Levy of England has placed a £1000 wager that a computer, given until 1978. to prepare, will not survive a 10-game match with him. Mikhail Botvinnik, the top-ranked player engaged in development of computer chess programs, told Levy, "I feel very sorry for your money." William Lombardy, whose analysis of the Kalme v. computer game also appears in the June SCIENTIFIC AMERICAN, writes, "My guess is that Levy will have his hands full in 1978."

I've already mentioned the SciAm article in a previous post: Scientific American's Computer Chess (January 2010):-

1973: A.L. Zobrist and F.R. Carlson, Jr.; 'An Advice-Taking Chess Computer'; Scientific American, Vol.228 no.6 (June 1973); p.92-105

Back to the question, 'Can you tell which player has white?' The game isn't available on The chess games of Charles Kalme (chessgames.com), so I entered it myself:-

[Event "Charles Kalme vs. IBM 370/155 computer; Who plays White?"]
[Site "Scientific American, June 1973"]
[Date "????.??.??"]
[Round "?"]
[White "?"]
[Black "?"]
[Result "0-1"]

1.e4 e5 2.Nf3 Nc6 3.Bc4 Bc5 4.O-O d6 5.h3 Nf6 6.Nc3 O-O 7.d3 h6 8.Nd5 Nxd5 9.Bxd5 Ne7 10.Bb3 Ng6 11.c3 Be6 12.d4 Bxb3 13.Qxb3 Bb6 14.dxe5 dxe5 15.Re1 Qf6 16.Be3 Bxe3 17.Rxe3 b6 18.Rd1 Rad8 19.Red3 Rxd3 20.Rxd3 Nf4 21.Re3 Rd8 22.Qc4 Qg6 23.Ne1 Rd1 24.Qf1 Rd2 25.Qc4 c6 26.b4 b5 27.Qb3 Qg5 28.c4 Qh4 29.Rf3 Re2 30.Qd1 Rxe4 31.cxb5 cxb5 32.Nc2 Qg5 33.Qf1 Rc4 34.Rg3 Qh5 35.Ne1 Ne2+ 36.Kh2 Nxg3 37.Kxg3 Rc1 38.f3 Qg5+ 0-1

A comment on the Chessgames.com page informs,

Dec-01-14 Petrosianic: I submitted a very odd Kalme game to the database once, but they never added it. It was from an advert for chess computers from the early 70's. The ad had a full game score between Kalme and a computer, but they didn't tell you who was who, and asked you to guess. I figured obviously Kalme was on the winning side (because if the computer had beaten him, they'd REALLY have bragged about that), and submitted the game. But it's not here.

I imagine Chessgames.com has a policy against having computer games on their database. As for the Levy wager, see Levy's Bet | Mastering the Game (computerhistory.org).

Stockfish Wins TCEC S24; Engine Mysteries

For this post on the world's top two engine vs. engine competitions, four weeks have passed since the previous post: TCEC S24 Sufi, CCC20 Rapid - Both Underway (April 2023). Here's a summary of the engine action reported at that time:-

TCEC: Stockfish won the 'DivP Playoff' with an undefeated score; LCZero finished second to qualify into the S24 Superfinal (aka Sufi). After 40 games in the 100-game Sufi, LCZero has a slim one point lead. • CCC: The 15-engine OpenBench event, titled 'OpenBench Interlude #8 - All Hands', was won by Berserk. This was followed by the first stage of the 'CCC20 Rapid' event, renamed from 'Newcomers' to 'Entry League', which was followed by 'Qualifier #1' and 'Qualifier #2'.

A lot can happen in a month, but the TCEC and the CCC both leave plenty of trail markers to follow their evolutions. Following is a summary of the past four weeks, including a couple of mysteries.

TCEC: Stockfish beat LCZero +20-16=64 to win the S24 Sufi. How many wins in a row does this make? If we go back to the previous Sufi, TCEC S23, CCC19 Rapid : Stockfish Wins Both (November 2022; 'Stockfish beat LCZero +27-10=63 to win its sixth straight Superfinal'), and add one, we get seventh straight Superfinal.

After the S24 Superfinal, the site conducted the following events (the links are to posts on this blog for the the previous iteration of the same event):-

• S24 VVLTC Bonus, 'Very Very Long Time Control'; see TCEC More Interludes; CCC19 Bullet Underway (December 2022; 'Stockfish won the VVLTC, then drew the two-game SVVLTC with LCZero')
• S24 Kibitzer Bonus; LCZero dag-master-[...] beat Stockfish master-[...]. The purpose of this match is a mystery.
• S24 4k 2, TCEC: 'Engine maximum size is 4 kiB (4096 bytes)'; see TCEC Cup 11 Underway; CCC19 Bullet in Semifinals (January 2023; 'Engine ice4 won the S23 4k tournament ahead of 4ku and three other engines')

The site is currently running the 'S24 4k 2 Final' between 4ku and ice4, the engines that finished 1st and 2nd in the six-engine 'S24 4k 2'. With one game left to play, ice4 has a one point advantage. The next scheduled event is 'FRC 6'.

CCC: After the 12-engine 'CCC20 Rapid Qualifier #2', the site ran two more 'CCC20 Rapid' events, a 12-engine 'Main' and a 6-engine 'Semifinal'. The three engines promoting from 'Qualifier #2' all finished in the bottom half of 'Main', while the top half promoted to the 'Semifinal'. The 'Semifinal' was won by Stockfish, ahead of Lc0 and Dragon. Stockfish qualified for the 'Final', while the other two engines will play a 'Challenger Match' to determine the second finalist.

Before the last two 'CCC20 Rapid' matches, the site conducted a series of 'Mystery' matches. These were described as:-

Mystery vs ... Berserk? Ethereal? Dragon? Lc0....? Until losing.

Mystery beat Koivisto by a wide margin and is currently leading Berserk.

[For further information from the various stakeholders in the engine-to-engine events, see the tab 'TCEC/CCC Links' at the top of this page. • NB: Leela = LC0 = LCzero; Dragon = KomodoDragon]

The Komodo Baton

Whenever I need an idea for an off-week engine post, seen previously in OpenBench Interlude (April 2023), I turn to Talkchess.com. The last time this happened was Talkchess Talks Top Engines (January 2023). Nestled in the news about 'Top Engines' was a Talkchess discussion identified as:-

2022-12-19: Dragon 3.2 Released at KomodoChess.com [lkaufman] • Talkchess user 'lkaufman' is GM Larry Kaufman.

A brief survey of recent Talkchess threads for this current post revealed:-

• 2023-03-17: KomodoChess and Chess.com [lkaufman] • 'Five years ago, Chess.com purchased a stake in KomodoChess along with rights to use our software. Now the merger is complete; Chess.com has bought out KomodoChess entirely. Chess.com will now run the website, pay for further development of Komodo, and for the most part take over the responsibilities of both myself and Mark Lefler. Mark and I will remain as paid consultants thru 2025, but Chess.com will make the decisions. Dietrich Kappe remains onboard as the NN trainer.'

GM Kaufman listed a number of reasons for his personal decision. One of them was:-

Komodo Dragon and Stockfish have effectively "solved" standard chess in the sense that with long time limits and large thread counts, they are both probably invincible if allowed to choose their own openings or at least get to choose from near-optimal openings.

Don't expect the former World Senior Champion to ride off into the sunset. He promises,

I hope that I am still able to make some significant contributions to the further development of Komodo Dragon. I remain interested in and involved with computer chess.

His Talkchess signature says, 'Komodo rules!'. The same could be said for him.

The first response to Kaufman's announcement informed, 'You have Andrew Grant on the team now -- he's employed by Chess.com, and I suspect he'll be able to increase the standard of play.' Is this the passing of the baton to the new generation of engine developers? By a not-so-curious coincidence, Andrew Grant also figured in the previous 'OpenBench Interlude' post.

TCEC S24 Sufi, CCC20 Rapid - Both Underway

The previous post in our fortnightly look at the two most important engine vs. engine competitions was TCEC S24 DivP Finishing; Stockfish Wins CCC20 Blitz Final (April 2023). Here's a summary of that post:-

TCEC: DivP is a few games away from its finish. Stockfish is a point ahead of LCZero, which has a big lead over KomodoDragon. For this season the organizers have organized the new 'DivP Playoff': 'The top four of DivP face each other to determine who goes to SuFi 24.' • CCC: Stockfish beat Lc0 by 293.0-207.0, making Stockfish the winner of the 'CCC20 Blitz Final'. The site is currently conducting another OpenBench event. Next: 'CCC20 Rapid Newcomers'.

TCEC: The previous post also noted some drama in determining the 'top four of DivP':-

Ethereal is currently in fourth place, a point ahead of Berserk. The two engines are scheduled to meet in one of the last games of DivP, so anything is possible.

Berserk won the game with the White pieces, achieving the same score as Ethereal and advancing to the 'DivP Playoff' on superior tiebreak. Stockfish won the playoff with an undefeated score; LCZero finished second to qualify into the S24 Superfinal (aka Sufi).

After 40 games in the 100-game Sufi, LCZero has a slim one point lead. At the same point in the S23 Sufi, with the same engines playing, Stockfish had a five point lead. Has there been a fundamental change in the relative strength of the two engines or is this a fluke? The event has another ten days to run.

CCC: The 15-engine OpenBench event, titled 'OpenBench Interlude #8 - All Hands', was won by Berserk, ahead of Ethereal and Rubi. The first stage of the 'CCC20 Rapid' event, renamed from 'Newcomers' to 'Entry League', was followed by 'Qualifier #1' and 'Qualifier #2'. All three stages had 12 engines, with the first two stages qualifying six engines into the following stage. Three engines will qualify from 'Qualifier #2' into the 12-engine 'Main' stage. 'Qualifier #2' is nearing the end of the second of its four rounds.

[For further information from the various stakeholders in the engine-to-engine events, see the tab 'TCEC/CCC Links' at the top of this page. • NB: Leela = LC0 = LCzero; Dragon = KomodoDragon]

OpenBench Interlude

In last week's TCEC/CCC post, TCEC S24 DivP Finishing; Stockfish Wins CCC20 Blitz Final (April 2023), I ended saying,

In previous posts, I've mentioned OpenBench several times in passing. It's high time I devoted an off-week post to the concept. Maybe next week...

This being next week and if not today, maybe never, let's first look at those passing references:-

• 2022-09-05: Chess324 Is a Thing • 'OpenBench was featured in an August CCC event called, 'OpenBench Interlude #1'.'
• 2022-10-31: TCEC S23 Sufi, CCC19 Rapid : Both Underway • 'After a few exhibition events, including two in an 'OpenBench Interlude' series, the [CCC] launched a 'CCC19 Rapid' multi-stage event.'
• 2023-02-06: TCEC Swiss 4 Underway; CCC Mystery Matches • 'As I was writing this post, the [CCC] had just finished 'OpenBench Interlude #7 - Top 6'.'

What can we learn from the TCEC/CCC !commands? The TCEC's '!openbench', points to chess.grantnet.us, while the CCC says nothing.

The Grantnet.us page is largely incomprehensible, although it points to Discord.com and Github.com resources. Those are also incomprehensible. I assume that chess engine developers have access to a Rosetta stone that de-mystifies all of the above, which is all that really matters to the rest of us. The Wikipedia page Chess engine (wikipedia.org), says,

In 2019, Ethereal author Andrew Grant started the distributed computing testing framework OpenBench, based upon Stockfish's testing framework, and it is now the most widely-used testing framework for chess engines.

Wikipedia adds '[citation needed]' to the 'most widely-used' statement, but Wikipedia says that about nearly everything that doesn't have a footnote. The wiki at OpenBench (chessprogramming.org; 'last edited on 15 April 2021'), explains,

An open source chess engine testing framework authored by Andrew Grant, written in Python and licensed under the GNU GPL. It is inspired [by] Fishtest and makes use of distributed computing, allowing anyone to contribute CPU time to further improve the development of open source chess engines.

We're nearly two years down the road from that 'last edited' date. Maybe the next update will explain the purpose of the CCC competitions that are limited to OpenBench engines. The 'OpenBench Interlude #8' event mentioned in last week's TCEC/CCC post was won by no.1 Berserk, ahead of no.2 Ethereal, no.3 Rubi, and 12 other engines.

Breaking the 3600 Barrier

The post from a couple of days ago, Baffling eBay (April 2023), was post no.3600 on this blog. That means it's time for another 'Breaking the Barrier' post, last seen in Breaking the 3500 Barrier (October 2022). In that '3500' post I wrote

If I ever do a 'Breaking 3600' post, I now have two CCRL rating lists to consult, but I'll cross that bridge when I come to it.

I am now standing in front of that bridge -- OK, sitting in front of it -- and I see that there are four CCRL rating lists. Shown below are the top five engines from three of the lists. Note that the third list is over three years old.

www.computerchess.org.uk/ccrl

Stockfish is the top engine on each of the lists, with a 3749 'Blitz' rating, and a 3536 '40/15', i.e. 'Rapid', rating. In the '3500' post I commented on the 200 point difference between the 'Blitz' and 'Rapid' ratings and am just as clueless now as I was then. In the same post I commented on the claim 'Carlsen breaks 2900 barrier':-

Given that Chess24.com is a defacto Magnus Carlsen fan club and that the record involved a '2900 Tour Rating' -- whatever that is -- the whole thing looks suspicious. TBD: Review for 'Breaking the 3600 Barrier'.

GM Carlsen's FIDE rating page Carlsen, Magnus (ratings.fide.com), shows him going over 2900 for both blitz and rapid ratings, but not for the standard rating. His current FIDE standard rating is 2853, making him the only player currently over 2800. GMs Nepomniachtchi and Ding Liren, the two players who will soon be playing for the title of World Champion, are rated no.2 and no.3 on the FIDE list at just under 2800.

Getting back to the 'CCRL Blitz Rating List' there are currently 17 engines rated over 3600. Based on that same list, it looks like we're already good for a 'Breaking 3700' post seven months (or so) from now, which should have at least two engines.

TCEC S24 DivP Finishing; Stockfish Wins CCC20 Blitz Final

In the previous fortnightly post, TCEC S24 DivP, CCC20 Blitz Final : Both Underway (March 2023), the intrepid sponsors of the chess world's leading engine vs. engine competitions, the TCEC and the CCC, were both conducting important contests. Here's a summary of that post:-

TCEC: With two rounds completed in the eight round Premier Division (DivP), KomodoDragon, LCZero, and Stockfish are significantly ahead of the other five engines. • CCC: Lc0 beat Dragon by 32 points in the 500-game 'Challenger Match', and is currently trailing Stockfish by 16 points after a sixth of the 500-game 'Final Match' has been played.

The TCEC is still running S24 DivP, but the CCC is moving on. Here's the current status of both platforms:-

TCEC: DivP is a few games away from its finish. Stockfish is a point ahead of LCZero, which has a big lead over KomodoDragon. That means Stockfish will meet LCZero in the final match (aka Superfinal, aka Sufi), right? No, wrong.

That's the way TCEC organized its flagship league event through S23; see TCEC S23 Sufi, CCC19 Rapid : Nearing the Finish (November 2022), for the previous season. For this season the organizers have organized the new 'DivP Playoff':-

!divpplayoff • The top 4 of DivP face each other to determine who goes to SuFi 24.

Ethereal is currently in fourth place, a point ahead of Berserk. The two engines are scheduled to meet in one of the last games of DivP, so anything is possible.

CCC: In the previous post, 'S24 DivP, CCC20 Blitz Final', I also calculated,

Lc0 is currently trailing Stockfish by 16 points after a sixth of the 500-game 'Final Match' has been played. This extrapolates to a difference of nearly 100 points when the match will have finished.

The final score was Stockfish over Lc0 by 293.0-207.0, making Stockfish the winner of the 'CCC20 Blitz Final'. The site is currently conducting another OpenBench event...

!now • OpenBench Interlude #8 -- All engines except Laser [Crashing :( ] and 4ku [Too weak :( ]

...then plans...

!next • CCC20 Rapid: Newcomers

In previous posts, I've mentioned OpenBench several times in passing. It's high time I devoted an off-week post to the concept. Maybe next week...

[For further information from the various stakeholders in the engine-to-engine events, see the tab 'TCEC/CCC Links' at the top of this page. • NB: Leela = LC0 = LCzero; Dragon = KomodoDragon]

Chessify Resources

A couple of months ago, in Chessify Progress Report (January 2023), I ended the post saying,

I'll continue to evaluate the [Chessify] service and perhaps learn why it's worth a subscription. If I discover anything important that I've overlooked to date, I'll come back with another post.

A couple of weeks later, the Chessify people started handing out free trial memberships. After a few weeks of the trial, I decided that a subscription was indeed worth the money, if for no other reason than to understand better the technology behind the service.

In that same post, I noted that there was 'No chess960 support'. Unfortunately, my test using the traditional start position encountered a snag when I received no moves from my opponents for over a week. I decided to analyze chess960 positions.

The main problem with chess960 in a traditional chess environment stems from the castling rules. Since chess960 games tend to become extremely tactical after a few moves have been played, there is nevertheless some value in trying to confirm the tactics with a traditional, non-chess960 engine. In one game, my chess960 engine indicated that White would have a big advantage after a certain move. The Chessify engines didn't like the move at all. It turned out that the difference hinged on a later move where White castled giving check, which is generally a strong move.

I'll continue using Chessify to look at chess960 positions. In the meantime, here are a few resources to help explore the Chessify service.

How to Analyze with Cloud Chess Engines on Chessify (2:10) • '[Published on] Feb 12, 2023'

The description for the corresponding Youtube video page says,

Learn how to analyze your chess games with Stockfish and other chess engines on the Chessify Analysis Dashboard. Choose from the variety of engine options, select the speed at which you want to analyze, and get the best move suggestions from the strongest chess engines in the world.

Other videos are available at Chessify AI (youtube.com). Although they are intended to overcome the lack of online documentation, they don't go into enough detail to do that successfully.

More info is available through Discord | #website-support | Chessify (discord.com). The videos point to the service's Twitter, Facebook, and Instagram pages, but I haven't looked at them yet. Other resources that I should follow are Latest News About Chessify (chessify.me) and Blogs about Chess Engines, Puzzles, & More (ditto).

TCEC S24 DivP, CCC20 Blitz Final : Both Underway

In the previous fortnightly report on the world's two leading engine vs. engine competitions, TCEC S24 L1, CCC20 Blitz Semifinals: Final Week (March 2023), both the TCEC and CCC were well into their respective seasons. Following is a summary of that report.

TCEC: S24 L1 is underway with Koivisto leading by a significant margin. Berserk is in a tie for second place. • CCC: Of the 12 engines that competed in 'CCC20 Blitz Main', six qualified into the 'CCC20 Blitz Semifinals', which is underway.

After another two weeks, both competitions are seeing the usual three engines vying for the top places. Following is the current status.

TCEC: Koivisto won S24 L1 three points ahead of Berserk, which finished a point ahead of the next three engines. Both Koivisto and Berserk promoted into the Premier Division (DivP). With two rounds completed in the eight engine, eight round tournament, they are in fourth and fifth place place, significantly behind KomodoDragon, LCZero, and Stockfish. LCZero beat Stockfish in an early game of the third round.

CCC: In the 'CCC20 Blitz Semifinals', Stockfish, Lc0, and Dragon finished in that order. Lc0 beat Dragon by 32 points in the 500-game 'Challenger Match', and is currently trailing Stockfish by 16 points after a sixth of the 500-game 'Final Match' has been played. This extrapolates to a difference of nearly 100 points when the match will have finished.

[For further information from the various stakeholders in the engine-to-engine events, see the tab 'TCEC/CCC Links' at the top of this page. • NB: Leela = LC0 = LCzero; Dragon = KomodoDragon]

NextChessMove.com

In the previous off-week engine post, Chessify vs. Chessbase Engine Cloud (February 2023), I noted,

Since the beginning of the year, the off-week engine posts -- meaning those weeks that I'm not tracking TCEC/CCC -- have been looking at the Chessify engine cloud.

While the Chessbase service deserves a deeper look, I'm going to switch my attention to another, similar service that I can use immediately without installing expensive software that I don't really need. The following screen capture shows the home page for 'Next Chess Move [NCM]: The strongest online chess calculator'.

NextChessMove.com

The intoduction at the top of that page explains,

Drag pieces to configure the board and press 'Calculate Next Move'. I'll tell you what the computer player does.

Last year we briefly saw the same site in the post Stockfish Breaks All the Barriers (April 2022; 'NCM Stockfish Dev Builds). At that time I neglected to delve into the real purpose of the site.

The function to 'Calculate Next Move' begs to be used on something other than the traditional start position that appears there by default. I went back to one of the games that I'm using to try Chessify, as described in the recent post Chessify Progress Report (January 2023).

I chose a game that has progressed to a Queen and Pawn (Q+P) endgame where I have an extra Pawn. It's a difficult endgame and during the Chessify trial I've learned a lot from offline resources about Q+P endgames in general. I've steadily improved the position of my King and Queen to their maximum potential, but am now facing a decision on how to proceed. There are several possible plans, but the Chessify engines insist that all variations lead to a draw.

Back to NCM, I used the FEN function, shown below the chessboard, to enter the FEN string from the current position of my Q+P endgame. I then unleashed both Stockfish 15.1 and LCZero 0.29.0 -- separately, of course -- on that position.

My first impression was that NCM is somewhat tedious to operate. 'Calculate Next Move' thinks for about five seconds, then returns a single move that (1) needs an action to apply it to the current position, followed by (2) another action to change the 'Active Color', i.e. the side to move, before it's ready to calculate the following move. None of the remaining moves in the 'Principal Variation' (PV) are shown.

Despite that repetitive procedure, the short test worked well and I'm ready for a more extensive test on a real game (where engines are permitted). The top of the right column on the NCM home page informs,

NCM Pro $19/year • Get stronger moves from NCM's 16 CPU core and RTX 2080 GPU dedicated servers. Free trial includes ten minutes of calculation time. Paid members get unlimited calculations.

The link for 'See Details' points to Next Chess Move: About NCM Pro, where more is explained. As soon as I finish one of my Chessify test games, I'll start the NCM free trial on a new game.

The major downside to the service is obvious. It's a magnet for online chess cheaters.

Cheating for all Ages

For the last few weeks I've been running a weekly series on Chess.com's Game Review Tools (February 2023). I'll continue with Chess.com, but change the focus. That post mentioned,

[Chess.com] has a good reputation for vigorously enforcing its no-engine policy, even if it leads to controversial decisions.

I'll start with a summary of past posts on this blog that dealt with various aspects of cheating. It's a real mixed bag. Posts marked '(*)' featured Chess.com.

Recent months have seen heavy media interest in the so-called Niemann affair. The entire discussion gets a '(*)' because Chess.com played a leading role.

• 2022-09-08: Chess Players Behaving Very, Very Badly
• 2022-09-19: Talkchess Talks Current Topics
• 2022-09-29: Cheating Mania
• 2022-10-27: Chess960 Mania
• 2022-11-24: Steamin' Niemann
• 2022-11-29: Disappearing Yahoos
• 2022-12-06: How Many for Carlsen? For Niemann?
• 2023-01-31: First Yahoos of 2023
• 2023-02-12: 'The Von Niemann Probe'

Another recent, recurring topic has been the controversy surrounding the 1997 match Kasparov vs. IBM's Deep Blue (m-w.com). It provided some precedent for the Niemann affair.

• 2019-07-28: The World Was Watching • 'Garry insisted that IBM cheated.'
• 2022-05-16: 25 Years Ago in Chess History
• 2022-09-25: 'The Root of All Evil'?
• 2022-10-17: 'One Move In, One Move Out'

The rest of the posts are listed in unfiltered, reverse chronological order.

• 2022-01-24: Correspondence Forums in the 2000s
• 2021-03-30: The Yahoos' Database Flags Cheating (*)
• 2020-08-28: Update on Online Chess Events (*) • 'Activity for the three main [online play] sites has declined, reflecting perhaps the gradual lifting of coronavirus lockdowns or perhaps the frustration of interminable cheating scandals.'
• 2020-05-10: An Influential Organization (*)
• 2020-02-28: FIDE's Media and Fair Play Commissions 2019
• 2020-02-07: Spectating the 90th FIDE Congress?
• 2019-07-30: A Cheating Yahoo • Igors Rausis
• 2019-01-24: FIDE's Ethics Commission 2018
• 2019-01-17: FIDE's 'Anti-Cheating' / 'Fair Play' Commission 2018
• 2018-01-18: FIDE's Ethics Commission 2017
• 2018-01-04: FIDE's Anti-Cheating Commission 2017
• 2017-03-31: PRO Chess League (*)
• 2016-12-27: FIDE's Anti-Cheating Commission 2016
• 2016-12-20: FIDE's Journalist Commission 2016
• 2016-04-14: ACP Survey 2016 - Results
• 2015-12-10: FIDE Anti-Cheating Guidelines
• 2015-12-03: FIDE Anti-Cheating Proposal
• 2015-11-26: Ethics in Chess Politics - Stories
• 2015-11-24: Ethics in Chess Politics - Cases
• 2015-11-10: Millionaire Open II - Chess Life
• 2015-08-18: 2015 CJA Awards
• 2014-10-09: Winter's 'Unsolved Chess Mysteries'
• 2014-06-10: The 'No Cheating' Jigsaw Puzzle • 'Two essays on one of the hottest subjects in chess have appeared this month'
• 2013-12-12: Ethics and Cheating
• 2013-04-26: Kramnik on the Unification Match
• 2012-10-25: Feller, Rybka, Arbiters, and More
• 2012-10-08: Most 'HTTP 404' Are External Errors
• 2011-01-18: 'We Have to Take Him as He Is'
• 2010-01-12: Scientific American's Computer Chess
• 2007-01-25: An Amazing Coincidence
• 2006-12-08: The Machine Sees Further Ahead

The list needs a lot more work. I'll fill in some of the blanks as I continue with the Chess.com series.

TCEC S24 L1, CCC20 Blitz Semifinals: Final Week

For the previous fortnightly post on the two most important, ongoing engine vs. engine competitions the keyword was 'Underway': Stockfish Wins TCEC Swiss 4; TCEC S24 L2 & CCC20 Blitz Underway (February 2023). Here's a summary of that post:-

TCEC: Stockfish won 'Swiss 4', a point ahead of KomodoDragon and LCZero. The site then started S24 League 2 (see diagram for S24 stages). • CCC: The 'CCC20 Blitz Newcomers' and 'Qualifier #1' events have finished and 'Qualifier #2' is underway. The next stage, 'CCC20 Blitz Main', will have 12 engines with three coming from 'Qualifier #2'.

The keyword is still 'Underway', although both competitions have advanced considerably. Both are reaching the stages that determine the three best engines to close their respective seasons.

TCEC: S24 League 1 is underway with Koivisto leading by a significant margin. Berserk is in a tie for second place. Both engines featured in the recent post Chessify's Other Engines (February 2023). L1 will run for another week.

Between L2 &nd L1, the TCEC conducted the 'L2 Igel Gauntlet', with Igel playing against 11 other engines. The TCEC S24 Wiki page explained in two parts the reason for the event. The first part said:-

Tournament Director's decision regarding Booot: It came to the attention of the author of Igel that Booot was finally ready, after many setbacks its author had experienced due to the war in his country [Ukraine]. Igel author in a gentleman gesture then offered Booot to take its place in League 2, and this offer was accepted by TD-decision.

And for the second part:-

Tournament Director's decision regarding Igel: The many Booot crashes distort League 2 results. To alleviate this problem, Igel will be placed back after this League's play and play a 2 rounds gauntlet with reverse games against the other 11 League 2 engines. After removing Booot's results this will give the final table and promotions for League 2.

That was a great show of sportsmanship by all concerned with good calls by the TCEC TD.

CCC: Of the 12 engines that competed in 'CCC20 Blitz Main', six qualified into the 'CCC20 Blitz Semifinals', which is underway. With less than 10% of the games completed, Stockfish already leads by a comfortable margin ahead of Dragon and Lc0, which are ahead of the three other engines. The event will finish within a week, when the second and third engines will compete in a 'Challenger Match' to determine who will face the winner of the Semifinals in the Final match.

[For further information from the various stakeholders in the engine-to-engine events, see the tab 'TCEC/CCC Links' at the top of this page. • NB: Leela = LC0 = LCzero; Dragon = KomodoDragon]