Et Mortuus Est Rex

Discussion of anything and everything relating to chess playing software and machines.

Moderators: Harvey Williamson, Dann Corbit, hgm

Forum rules
This textbox is used to restore diagrams posted with the [d] tag before the upgrade.
lkaufman
Posts: 4425
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Et Mortuus Est Rex

Post by lkaufman » Sat Jul 04, 2020 7:11 pm

Dr.Wael Deeb wrote:
Sat Jul 04, 2020 6:26 pm
lkaufman wrote:
Sat Jul 04, 2020 12:24 am
Dr.Wael Deeb wrote:
Fri Jul 03, 2020 10:34 pm
lkaufman wrote:
Fri Jul 03, 2020 9:43 pm
Dr.Wael Deeb wrote:
Fri Jul 03, 2020 8:15 pm
Hi,

The title is in Latin and it means The Dead King ....

I started a series of chess tournaments playing the chess engines rated in the CCRL from 2400 Elo and higher ....

Each tournament consists of 7 engines and myself playing 2 games against each other ....

The first engine I played is Simplex 0.9.8 x64 running on my i9 overclocked monster PC ....

Time control is 40 minutes + 20 seconds increment ....

The only advantage I use against the engines is that I am not entitled to the above mentioned time control meaning that my thinking time is infinite but usually I think around 2-5 minutes on a single move ....

Simplex 0.9.8 went down in the 2 games :mrgreen:

Note that I steer the chess engine to a specific opening line that is comfortable for me,in this case the Hedgehog of the Symmetrical English which I am pretty much familiar with also when I play the Sicilian Kan variation ....

Here I tricked Simplex to take the pawn on the b6 square with it's queen drafting it away from the white kingside ....

It was a risk and somehow I bluffed but it played along as I generally anticipated ....

All in all Simplex is a good playing chess program but it lacks tactical abilities and has a noticeable problems with positional understanding of the game ....

The game is annotated by Deep Rybka 4.1 SSE42 ....

Enjoy ....


It would be helpful to know your own Elo rating (over the board, if you have one, online if you don't) so that these games might shed some light on the accuracy of the CCRL ratings in human terms. If you are not close to 2400, and if you continue to score well against engines over 2400 then we need to figure out why the engines are rated that highly, when we know that the reference engines are under-rated. Also how many threads are they running on? This should offset the time odds factor.
Hi Larry,

Unfortunately I don't have OTB rating nor an online one ....
I started playing chess and became very interested back in Bulgaria when I was a student in the medical university ....

Due to my professional occupation I've never had the chance to become OTB chess player so I dedicated the little time I have to self education and playing against the chess engines ....

In human terms,you will never know the accuracy of a rating list being CCRL or any rating pool in that matter if you don't engage a human to play the engines ....

As a rough estimation I consider myself playing around the 2200 Elo so Simplex is definitely not a 2400 Elo chess engine and is quite possible highly overrated....

Note how I managed to hijack it's queen to take the pawn on the b6 square and it stayed there while the white king encountered a vicious attack ending the game in black favorite ....

I hope to play enough games climbing the CCRL rating list to gather at least a slight data regarding the chess engines versus the humans ....

Regarding the my hardware it's an Intel® Core™ i9-9900K Processor running a base frequency of 3.6 GHz and with a help of a friend of mine we managed to overclock it to 4.8 GHz so it's quite powerful ....

I let the chess engines use 8 threads when possible with 2048 ram memory ....

I have my own rating list which I started to rebuild recently but I didn't include myself in it,just chess engines :mrgreen:

Cheers,
Dr.D
The Simplex engine is only listed in the CCRL list with one thread. I don't know if it is an MP engine or not. Do you know whether it was actually using your 8 threads or perhaps just 1? If it was using 8, then it's quite fair with the time handicap, but if it was using just 1, then it was a time handicap game so it wouldn't be so strange for a 2200 player to beat a 2400, still a bit surprising. Is your 2200 estimate for yourself based on engine ratings you have played, or were you actually scoring around 50% with 2200 human players in the past, even if the games were not rated? If it is based on playing engines, then it would be a paradox if you perform at say 2400 vs engines.
Hi Larry

Simplex is one thread chess engine but again it run on a powerful i9 processor overclocked to 4.8 GHz with 2048 Mb hash table ...

The length of the game is around one hour for the engine which is already playing it's first move after a 12 moves opening line ....

Well,my 2200 estimation is purely based on playing the chess engines starting from a 1800 Elo and using the FIDE chess rating calculator ....

Note until the year 2016 I played a 980 games against various chess engines and studied a lot certain opening lines and middle game techniques when I had the time ....

Anyways,we'll see what happens next as I play the next rated chess engine in the CCRL rating : Tigran 2.4n x64 Which BTW I caught it in lovely Old Indian defense opening line which happens to be one of my best playing lines that I practice a lot ....

Cheers,
Dr.D
So just to be clear, your 2200 estimate is based on your results vs. engines running on much older/inferior hardware to what you now have, and calculated based on the CCRL 40/15 rating list for those engines? Is that accurate? If so, presumably you were about 50-50 with engines rated near 2200 on that list. So if it turns out that you are now making a plus score against engines over 2400 on the same list, despite running them on a much better computer now, what would be your explanation? Have you improved by a class since 2016? Are you taking much more time than you took in the earlier games? Are you forcing the engines to play only the openings you have found to suit you the best? Are you thinking that the 2400 engines are just not much better than the 2200 rated engines vs. humans? Any other theory?
Komodo rules!

User avatar
Dr.Wael Deeb
Posts: 9772
Joined: Wed Mar 08, 2006 7:44 pm
Location: Amman,Jordan

Re: Et Mortuus Est Rex

Post by Dr.Wael Deeb » Sat Jul 04, 2020 7:38 pm

So just to be clear, your 2200 estimate is based on your results vs. engines running on much older/inferior hardware to what you now have, and calculated based on the CCRL 40/15 rating list for those engines? Is that accurate?

Correct ....

If so, presumably you were about 50-50 with engines rated near 2200 on that list. So if it turns out that you are now making a plus score against engines over 2400 on the same list, despite running them on a much better computer now, what would be your explanation? Have you improved by a class since 2016?

I definitely did ....
Actually I want to share something interesting that is not only related to the computer chess ....

Even in the gym,I am now performing much better than when I was younger,believe it or not ....
Call it good genes,the sport type,the antioxidant diet,etc.

When I set to play chess now,I see much clearer than before 10 years for example and I enjoy it and for that I don't have an explanation ....


Are you taking much more time than you took in the earlier games?

No,on average the same time if not shorter actually ....

Are you forcing the engines to play only the openings you have found to suit you the best?

Of course I do and this is the main factor ....

By forcing the engine to play certain opening lines I mean steering the game toward a well known opening line not dropping the opening line over the engine and forcing it to play from there if you understand what I mean ....

I will not ago against a chess engine playing the Sicilian Najdorf or the Dutch defense with black or any other sharp opening lines where I will get blown out of the water with tactical fireworks ....

Take a look at the second game with Simplex :



What do you think happened here :!: :?:


Are you thinking that the 2400 engines are just not much better than the 2200 rated engines vs. humans? Any other theory?

I am confused here I can't answer ....

Let me play another 20 games or so and maybe we'll get to a consensus ....

Cheers,
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….

lkaufman
Posts: 4425
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Et Mortuus Est Rex

Post by lkaufman » Sat Jul 04, 2020 8:40 pm

Dr.Wael Deeb wrote:
Sat Jul 04, 2020 7:38 pm
So just to be clear, your 2200 estimate is based on your results vs. engines running on much older/inferior hardware to what you now have, and calculated based on the CCRL 40/15 rating list for those engines? Is that accurate?

Correct ....

If so, presumably you were about 50-50 with engines rated near 2200 on that list. So if it turns out that you are now making a plus score against engines over 2400 on the same list, despite running them on a much better computer now, what would be your explanation? Have you improved by a class since 2016?

I definitely did ....
Actually I want to share something interesting that is not only related to the computer chess ....

Even in the gym,I am now performing much better than when I was younger,believe it or not ....
Call it good genes,the sport type,the antioxidant diet,etc.

When I set to play chess now,I see much clearer than before 10 years for example and I enjoy it and for that I don't have an explanation ....


Are you taking much more time than you took in the earlier games?

No,on average the same time if not shorter actually ....

Are you forcing the engines to play only the openings you have found to suit you the best?

Of course I do and this is the main factor ....

By forcing the engine to play certain opening lines I mean steering the game toward a well known opening line not dropping the opening line over the engine and forcing it to play from there if you understand what I mean ....

I will not ago against a chess engine playing the Sicilian Najdorf or the Dutch defense with black or any other sharp opening lines where I will get blown out of the water with tactical fireworks ....

Take a look at the second game with Simplex :



What do you think happened here :!: :?:


Are you thinking that the 2400 engines are just not much better than the 2200 rated engines vs. humans? Any other theory?

I am confused here I can't answer ....

Let me play another 20 games or so and maybe we'll get to a consensus ....

Cheers,
Dr.D
OK, that was bizarre, on move 17 the engine just blundered a pawn and opened the diagonal for your bishop. Even a 1 ply search on an forty year old engine wouldn't do anything that obviously bad. I can't imagine that an engine that plays such a blunder would get a 2400 CCRL rating, unless it's a bug that just happens very rarely and just happened in this game. Very strange.
Regarding openings, I wouldn't expect you to take on an engine in a Najdorf Sicilian or other such sharp opening, but if you are going to avoid such things by opening 1.b3, why not let the engine play whatever it likes (either with or without an opening book, maybe without would be the most interesting) right from move 1 rather than choose some arbitrary set of opening moves for it? That seems like a pretty pure test of strength; the engine has no book, and you have thrown out most book theory by an opening like 1.b3 (not all, but enough).
Komodo rules!

User avatar
Dr.Wael Deeb
Posts: 9772
Joined: Wed Mar 08, 2006 7:44 pm
Location: Amman,Jordan

Re: Et Mortuus Est Rex

Post by Dr.Wael Deeb » Sat Jul 04, 2020 9:13 pm

OK, that was bizarre, on move 17 the engine just blundered a pawn and opened the diagonal for your bishop. Even a 1 ply search on an forty year old engine wouldn't do anything that obviously bad. I can't imagine that an engine that plays such a blunder would get a 2400 CCRL rating, unless it's a bug that just happens very rarely and just happened in this game. Very strange.
Regarding openings, I wouldn't expect you to take on an engine in a Najdorf Sicilian or other such sharp opening, but if you are going to avoid such things by opening 1.b3, why not let the engine play whatever it likes (either with or without an opening book, maybe without would be the most interesting) right from move 1 rather than choose some arbitrary set of opening moves for it? That seems like a pretty pure test of strength; the engine has no book, and you have thrown out most book theory by an opening like 1.b3 (not all, but enough).


Yes it was definitely a glitch and I am not very proud of this game but again running on a powerful nowadays hardware should at least prevent the engine from making a ?? chess move ....

Exactly my thoughts,this engine is definitely overrated and I think the programs for calculating the Elo are the reason for this ....
That's why in my rating list I use the FIDE chess rating calculator manually ....

As I said before,I don't chose some arbitrary set of opening for the engine ....
What I do is this:
I let the engine use the Fritz 17 opening book for example with the tournament book feature turned on so when I play 1.b3 it chooses the best move in the book and I respond according to my opening lines which I usually play until the 12th move and that's it,game on ....

I started a game against Tigran 2.4n x64 with the black pieces and when I pressed the interval to let the engines choose it's first move,it played 1.d4 after which I steered the game to A54 Old Indian Defense which I am pretty much familiar with it's nature as an opening line ....
I know the major patterns of the opening,I am familiar with the main strategic ideas,possible white tactical shots,black counter attacks,etc.
Why playing random opening I know nothing about and get slaughtered :!: :?:

Now what you're assuming is called a theme chess match where I force the engine to play the Sicilian Kann variation for example or the King's Gambit or whatever ....
Maybe in the future if I have time I will perform a theme tournament and see how it fares :D

For now I want to climb the CCRL rating list and see where I can get ....

Finally,even though it's only two games I think that Simplex is a 2100 2000 chess engine at best maybe less ....

Cheers,
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….

carldaman
Posts: 2027
Joined: Sat Jun 02, 2012 12:13 am

Re: Et Mortuus Est Rex

Post by carldaman » Sat Jul 04, 2020 10:16 pm

Dr. Deeb, you're clearly an old-school anti-engine specialist, who can steer the opening into favorable closed, strategic opening lines where Simplex is clearly not of 2400 strength. I hope someday you could share your many instructive wins (over the years) against engine opponents. It's become a lost art in the Stockfish era. :)

lkaufman
Posts: 4425
Joined: Sun Jan 10, 2010 5:15 am
Location: Maryland USA
Contact:

Re: Et Mortuus Est Rex

Post by lkaufman » Sat Jul 04, 2020 11:22 pm

Dr.Wael Deeb wrote:
Sat Jul 04, 2020 9:13 pm
OK, that was bizarre, on move 17 the engine just blundered a pawn and opened the diagonal for your bishop. Even a 1 ply search on an forty year old engine wouldn't do anything that obviously bad. I can't imagine that an engine that plays such a blunder would get a 2400 CCRL rating, unless it's a bug that just happens very rarely and just happened in this game. Very strange.
Regarding openings, I wouldn't expect you to take on an engine in a Najdorf Sicilian or other such sharp opening, but if you are going to avoid such things by opening 1.b3, why not let the engine play whatever it likes (either with or without an opening book, maybe without would be the most interesting) right from move 1 rather than choose some arbitrary set of opening moves for it? That seems like a pretty pure test of strength; the engine has no book, and you have thrown out most book theory by an opening like 1.b3 (not all, but enough).


Yes it was definitely a glitch and I am not very proud of this game but again running on a powerful nowadays hardware should at least prevent the engine from making a ?? chess move ....

Exactly my thoughts,this engine is definitely overrated and I think the programs for calculating the Elo are the reason for this ....
That's why in my rating list I use the FIDE chess rating calculator manually ....

As I said before,I don't chose some arbitrary set of opening for the engine ....
What I do is this:
I let the engine use the Fritz 17 opening book for example with the tournament book feature turned on so when I play 1.b3 it chooses the best move in the book and I respond according to my opening lines which I usually play until the 12th move and that's it,game on ....

I started a game against Tigran 2.4n x64 with the black pieces and when I pressed the interval to let the engines choose it's first move,it played 1.d4 after which I steered the game to A54 Old Indian Defense which I am pretty much familiar with it's nature as an opening line ....
I know the major patterns of the opening,I am familiar with the main strategic ideas,possible white tactical shots,black counter attacks,etc.
Why playing random opening I know nothing about and get slaughtered :!: :?:

Now what you're assuming is called a theme chess match where I force the engine to play the Sicilian Kann variation for example or the King's Gambit or whatever ....
Maybe in the future if I have time I will perform a theme tournament and see how it fares :D

For now I want to climb the CCRL rating list and see where I can get ....

Finally,even though it's only two games I think that Simplex is a 2100 2000 chess engine at best maybe less ....

Cheers,
Dr.D
We can agree that an engine that makes one ply blunders should not be rated 2400, but I checked your theory as to the reason and it appears to be wrong. The win percentage for Simplex in CCRL was 47.3%. The formula they use, BayesElo, has the property of compressing ratings, so as compared to normal Elo it will overrate engine with poor scores. But this is close to 50%, so it would only overrate it by a few elo, and we are talking about hundreds of elo. I can only guess that some positions it plays 2800 level to offset the ones where it plays 2000 or worse.
It sounds like what you are doing is the same as just playing the engine a normal game from the start, with the engine set to use a specific opening book in tournament mode, which is quite normal. Is that right? That is perfectly fine of course.
I wasn't proposing theme tournaments. I was suggesting you just play against the machine using your normal openings, such as 1.b3, but with the opening book turned off. It's just a question of whether you want to play against the engine from the start or from engine + book from the start. Both are interesting. If you play sharp openings then playing vs. pure engine isn't very informative, but if you play lines like 1.b3 it is quite ok.
Good luck and perhaps we'll learn something from this experiment (other than that there is something wrong with Simplex!).
Komodo rules!

User avatar
Ovyron
Posts: 4405
Joined: Tue Jul 03, 2007 2:30 am

Re: Et Mortuus Est Rex

Post by Ovyron » Sun Jul 05, 2020 12:21 am

I agree with Larry, it'd be interesting to see what the engine plays by itself without book.

carldaman
Posts: 2027
Joined: Sat Jun 02, 2012 12:13 am

Re: Et Mortuus Est Rex

Post by carldaman » Sun Jul 05, 2020 3:29 am

It's always interesting to see what the engine will come up with on its own, but positionally-deficient 2400-level programs usually tend to play crummy, exploitable openings on their own. Even stronger A-B engines will struggle in that phase. Dr Deeb's job probably won't get any harder that way.

User avatar
Ovyron
Posts: 4405
Joined: Tue Jul 03, 2007 2:30 am

Re: Et Mortuus Est Rex

Post by Ovyron » Sun Jul 05, 2020 4:46 am

carldaman wrote:
Sun Jul 05, 2020 3:29 am
It's always interesting to see what the engine will come up with on its own, but positionally-deficient 2400-level programs usually tend to play crummy, exploitable openings on their own.
That will only make Dr.Wael Deeb get higher on the engine list. Imagine if he can get to Stockfish, that's the dream!

User avatar
Dr.Wael Deeb
Posts: 9772
Joined: Wed Mar 08, 2006 7:44 pm
Location: Amman,Jordan

Re: Et Mortuus Est Rex

Post by Dr.Wael Deeb » Sun Jul 05, 2020 7:59 am

carldaman wrote:
Sat Jul 04, 2020 10:16 pm
Dr. Deeb, you're clearly an old-school anti-engine specialist, who can steer the opening into favorable closed, strategic opening lines where Simplex is clearly not of 2400 strength. I hope someday you could share your many instructive wins (over the years) against engine opponents. It's become a lost art in the Stockfish era. :)
Thanks for your interest ....

Not all of my opening lines lead to closed positions ....

Take a look for example of my game with Simplex,it,s an English opening A30 which is anything but a closed position ....

Unfortunately I lost almost all of my old games due to a hardware crash but even though I prefer to share my recent games as I play them on a powerful hardware which makes it more interesting IMHO ....

Cheers,
Dr.D
_No one can hit as hard as life.But it ain’t about how hard you can hit.It’s about how hard you can get hit and keep moving forward.How much you can take and keep moving forward….

Post Reply