Milenium Genius News???

klute · Post by **klute** » Thu Sep 10, 2015 1:42 pm

Monsieur Plastique wrote:I'm talking about chess play here at around the level of an amateur expert - say the very high 1900s to around 2100 per the FIDE rating scale.

I was around that level in my competitive playing days a long time ago. My two best memories were smashing a soon-to-be IM as Black in a French and also a win against a WFM although I was dissatisfied with my play in the latter game, which was a bit scrappy. Both games were in official slow-play tournaments.

I never quite cracked a 2000 FIDE rating though - last time I looked my FIDE rating was 1967 from 13 years ago or something like that.

In my view the main difference between an Expert / Candidate Master and fully-fledged masters is consistency, constant practice and a methodical approach to technique.

Perhaps I might have got there if I had really studied harder, but...

...computer chess won out Regards

klute · Post by **klute** » Thu Sep 10, 2015 3:44 pm

Monsieur Plastique wrote:of all the machines I could choose to train against, it would be a 32 bit Lang machine with hash tables.

I also highly recommend the top 8-bit Schröders for those wanting a close to human-like experience at this level.

I know everyone raves about high-end Novags and yes they're exciting, but they can be a bit too coffee-house.

Post by **Fernando** » Thu Sep 10, 2015 4:01 pm

Monsieur Plastique wrote:
spacious_mind wrote:Let's just say that Lang's can be very frustrating as they get into positions where they begin moving aimlessly. Their saving grace is that a lot of their opponents at that time had similar problems.
But Nick, I'm talking about chess play here at around the level of an amateur expert - say the very high 1900s to around 2100 per the FIDE rating scale. You are talking, however, as if I am using a FIDE Master paradigm to describe a Lang program. I'm not - a FIDE master is a lot stronger than any Lang program in any dedicated machine.

Strong amateur players around the 2000 mark are obviously very good, however the fundamental reason they are not FIDE Masters is because they make more weak and unjustified moves more of the time than a true master does. They also make more blunders more of the time. In terms of their "best" chess, many players around the 2000 mark will play like an IM on a really good day - that is where they play their "best" game and without any of the weaker moves and blunders. The occasionally stupid and dumb moves (whether it be a tactical blunder, a dubious long-term strategic decision or weak positional move) is exactly why these players are rated at 2000 instead of 200 points higher. So I don't really think it is fair to say a Lang machine is completely unlike them even if the Lang machines make stupid moves themselves as well - they are merely in commensurate company with similarly-string humans in terms of their chess playing ability.

As I did actually go onto say in my original post that you quoted, sometimes the games of the strong Langs can get indeed quite dowdy and drawish which is effectively the same (or I meant it to mean the same) as the criticisms that you have levelled at them. However it takes two to tango. I myself saw this behaviour too in the test games I have been playing using my Genius phone at 3 seconds per move (about the same computational speed as the new Millennium at tournament level) - there was some pointless queen meandering in the test game for example that reminded me of machines from beginning of the 1980s. I even managed myself to draw against the Lang Genius program running at around 2500 ELO just out of the opening precisely because this dubious stubbornness that can result in weak moves. I actually think a lot of the reason for the strange moves we do see in the Lang programs is because of the positional evaluations they use coupled with an overly cautious contempt factor - the programs will occasionally rather make a silly looking move that does not improve it's position than perhaps make another move that might possibly be valued slightly lower but down the track proves to be better and would result in a much more coherent looking game.

That said, all programs that have ever made their way into a dedicated machine have had shortcomings in one or more respects. I'm not yet aware of any dedicated machine that you could put in front of me and I wouldn't be able to point out dubious, annoying and even unfathomable behaviour at least some of the time.

Nevertheless, you are of course entitled to your opinion but my point stands as I made it - all things considered these Lang programs in my experience emulate better than most - more often than not - the "feeling" of playing a human being who is around the high 1900s to around the 2100 level. Meaning it emulates better than other programs the inherent strengths and weaknesses of player at that standard. The strong Kittinger, Morsch and Shroeder programs are of course a lot of fun and a great challenge to play against but quite honestly I don't really know of any humans at a corresponding ELO level who play even remotely the way they do. Their play is more akin to a much stronger player than their actual rating playing a (very) rapid game - at least in terms of everything bar outright tactical ability.[/i][/u][/i]

If I were preparing to play in an under 2000 tournament, of all the machines I could choose to train against, it would be a 32 bit Lang machine with hash tables. That would give me the very best match practice and would hone my strengths in order to successfully compete against the human players I would come up against.

I am one of those expert levl player -2100 apro- and as such I can say that the most adequate engine for a player from 1900 to this rating is, I think, something as Chess System Tal.
It is LOT more fun than Genius and strong enough.
Same with Hiarcs, though they are too strong...
Kittinger stuff is wonderful too.

Fern

Monsieur Plastique · Post by **Monsieur Plastique** » Thu Sep 10, 2015 4:11 pm

klute wrote:I was around that level in my competitive playing days a long time ago. My two best memories were smashing a soon-to-be IM as Black in a French and also a win against a WFM although I was dissatisfied with my play in the latter game, which was a bit scrappy. Both games were in official slow-play tournaments.

And my claim to fame is beating the Queensland State Chess champion and a genuine FIDE Master. Pity they were both effectively about 1400 ELO players at the time I met them over the board though!

Of course I know better than many of what member Klute is capable of over a chess board. In fact when I was mentioning a decent 2000-ish player playing like an IM on a good day, some of his stunning victorious skirmishes were the ones I was thinking about.

I also remember back in late 1980s after he had just bought the Mephisto Exclusive MMIV which amongst other things he was using for human tournament preparation. I played through the score of his first formal game against it at 40 moves in 2 hours. From memory the machine lost a minor piece and then went on to get crushed into a little corner typical of a Kasparov victory. So yes, human players around 2000 ELO are formidable on their best days. And last time I checked, even Kasparov will refuse to play a simul against players rated at or over 1900 unless special approval and conditions are met first.

Even when Klute was long out of formal tournament practice, I remember giving him my latest beta version of Fruit running on the 64 Mhz Nintendo console to muck around with, only for him to humiliate the program in a rapid game like it was some pathetic, drunken, coffee-house amateur.

Of course the good thing about our Klute is that unlike his Sci-Fi TV namesake, this one does not electrocute you to a crisp if you happen to lose.

One thing I have not mentioned specifically I don't think - whenever I talk about how machines play, my comments pertain almost exclusively to their behaviour at the classical tournament time controls - these are the only levels I use when I play machines myself and they are the only levels I use when I play machine versus machine. So far as the Lang programs are concerned, it is pretty obvious that they scale well - after all, Kasparov did lose a rapid match to what was effectively little more than an evolved Lang dedicated machine ported to a PC running on steroids. And Larry Kaufmann also sang the praises of the Lang programs even as early as the Amsterdam model (whilst acknowledging of course the shortcomings).

Monsieur Plastique · Post by **Monsieur Plastique** » Thu Sep 10, 2015 4:17 pm

Fernando wrote:I can say that the most adequate engine for a player from 1900 to this rating is, I think, something as Chess System Tal.
It is LOT more fun than Genius and strong enough.

I agree that there are many machines and programs out there that are more fun to play against than Genius. However I am not sure that serious humans in a serious 2000 ELO tournament play like Chess system Tal, so I am not sure I agree that it makes the best training tool. But it is a training tool.

What I definitely wouldn't be doing for tournament preparation is to play against any modern engines. It is impossible to hobble them properly - the play becomes far more inconsistent than a human at a comparable rating and generally the "target" rating you set ends up being quite a lot higher than it's actual play. And trying to play a 3000 ELO engine isn't terribly good training when you are a piece down by move 40. You don't get the right type of end game practice!

Post by **Fernando** » Thu Sep 10, 2015 4:18 pm

Monsieur Plastique wrote:
klute wrote:I was around that level in my competitive playing days a long time ago. My two best memories were smashing a soon-to-be IM as Black in a French and also a win against a WFM although I was dissatisfied with my play in the latter game, which was a bit scrappy. Both games were in official slow-play tournaments.
And my claim to fame is beating the Queensland State Chess champion and a genuine FIDE Master. Pity they were both effectively about 1400 ELO players at the time I met them over the board though!

Of course I know better than many of what member Klute is capable of over a chess board. In fact when I was mentioning a decent 2000-ish player playing like an IM on a good day, some of his stunning victorious skirmishes were the ones I was thinking about.

I also remember back in late 1980s after he had just bought the Mephisto Exclusive MMIV which amongst other things he was using for human tournament preparation. I played through the score of his first formal game against it at 40 moves in 2 hours. From memory the machine lost a minor piece and then went on to get crushed into a little corner typical of a Kasparov victory. So yes, human players around 2000 ELO are formidable on their best days. And last time I checked, even Kasparov will refuse to play a simul against players rated at or over 1900 unless special approval and conditions are met first.

Even when Klute was long out of formal tournament practice, I remember giving him my latest beta version of Fruit running on the 64 Mhz Nintendo console to muck around with, only for him to humiliate the program in a rapid game like it was some pathetic, drunken, coffee-house amateur.

Of course the good thing about our Klute is that unlike his Sci-Fi TV namesake, this one does not electrocute you to a crisp if you happen to lose.

One thing I have not mentioned specifically I don't think - whenever I talk about how machines play, my comments pertain almost exclusively to their behaviour at the classical tournament time controls - these are the only levels I use when I play machines myself and they are the only levels I use when I play machine versus machine. So far as the Lang programs are concerned, it is pretty obvious that they scale well - after all, Kasparov did lose a rapid match to what was effectively little more than an evolved Lang dedicated machine ported to a PC running on steroids. And Larry Kaufmann also sang the praises of the Lang programs even as early as the Amsterdam model (whilst acknowledging of course the shortcomings).

Well, well, I see then that with Klute I could face a challenger of probably my strength....when I was 30 years old...
Perhaps my best performance in these last years was a draw with Vishy, here in Santiago, a fast game that we played as we talked too. Nice guy, you know...

Paradise Lost regards
Fern

Monsieur Plastique · Post by **Monsieur Plastique** » Thu Sep 10, 2015 4:32 pm

Fernando wrote:Perhaps my best performance in these last years was a draw with Vishy, here in Santiago, a fast game that we played as we talked too. Nice guy, you know...

Impressive

I have actually gotten better as I get older and I feel that at nearly 50 I am probably playing better in serious slow games than ever before, though I certainly could never be fit to enter any tournament. Still, that isn't really saying very much. I think for me to have achieved 2000 ELO would have required a monumental commitment to prolonged and serious study and thus a very serious time commitment. I am not even sure I would ever have had the ability to get to that level and sustain it no matter how hard I tried. I remember once talking to Peter Parr about this and he opined that an existing, competent adult amateur who decided to commit themselves to very serious, formalised study would likely gain on average around 200 points through doing so. So based on that I think my ceiling might have been in the 1900s at best.

I just play for fun and have never had any serious aspirations to be anything more than a very good hobby strength player. I've had a few moments such as a relatively recent crushing victory over a Saitek Cosmos in a serious 40 in 2 game but I think for the most part I am best classified as an enthusiast who plays a competent game amongst the middling amateur ranks.

Mind you, for me chess stops being fun when you get too serious about it anyway. The study of the game has become so clinical since the rise of PC programs and databases that almost all the fun has gone out of playing at the cutting edge of theory.

If chess theory had never outgrown the game as it was played prior to WWI, I for one would be very happy! What fun is there in chess these days when the first 40 moves are played out of book and then two moves later someone makes a mistake? I purposely only have two opening books at home - Nun's Chess Openings from the 1990s and Understanding the Chess Openings by Sam Collins! I am happy with just those two!

Post by **Steve B** » Thu Sep 10, 2015 5:35 pm

Monsieur Plastique wrote: I have actually gotten better as I get older and I feel that at nearly 50 I am probably playing better in serious slow games than ever before

and at nearly 60 I can say I have gotten worse as I grow older
I would mark the start of my decline at just around 50
went from playing third board for my University team(in my early 20's) ...to becoming bored with the game today..
well not bored exactly .. more like.. impatient
I don't even play long games against my computers anymore
nothing slower then game/30..usually game in 5 or 10
I also play at the "player adaptive" time level setting where the computer takes about the same amount of time that the player takes..but not many computers offer that level option

I remember the last time I played a game at 40/2 against a computer I would stand up and rap my knuckles repeatedly on the board in an act of impatience waiting for the computer to make its move
as the game went on I eventually wiped all of the pieces off the board..unplugged the adaptor (which I then spiked into the living room carpet )and declared myself the winner by time forfeit (while rushing off in a huff)

legend has it that Alehkine employed the "rapping knuckles" maneuver when playing in Simuls if an opponent did not move as soon as he got to his board
Well At Least the great Alehkine and I

Had That In Common Regards
Steve

klute · Post by **klute** » Thu Sep 10, 2015 7:00 pm

Fernando wrote:I see then that with Klute I could face a challenger of probably my strength

All of us would fry playing the "real" one:

https://www.youtube.com/watch?v=7yN_MJMLMeA

Post by **Fernando** » Thu Sep 10, 2015 7:05 pm

klute wrote:
Fernando wrote:I see then that with Klute I could face a challenger of probably my strength
All of us would fry playing the "real" one:

https://www.youtube.com/watch?v=7yN_MJMLMeA

I AM THE REAL
I AM A CHESS GOD
I AM IVAN THE TERRIBLE I AM THE BEST
I AM UNBEATABLE

from the madhouse regards
{Fern

spacious_mind · Post by **spacious_mind** » Fri Sep 11, 2015 12:50 am

Wow, catching up on reading and seeing all the posts generated regarding playstyle preferences!

Monsieur Plastique wrote:
But Nick, I'm talking about chess play here at around the level of an amateur expert - say the very high 1900s to around 2100 per the FIDE rating scale. You are talking, however, as if I am using a FIDE Master paradigm to describe a Lang program. I'm not - a FIDE master is a lot stronger than any Lang program in any dedicated machine.

As I stated I just think that Lang especially in closed passive situations begins to dither about and probably because of the highly selective search misses moves that could allow it to proceed favorably, but instead continues to dither around.

Maybe it is a preference thing since I like to play very active and tactical and I just find Lang's style not too difficult to be successful against and really kind of boring. Also I acknowledge that when I am making this judgment I am comparing to programs that came out just a tad later, who I just find better all round and more balanced in their play. ie. King for which the weakest computer that I have is Risc 2500 and then move forward to R30 and Chessmachines. Same with Schroeder. Both programmers I find much tougher nuts crack on these computers.

If I were preparing to play in an under 2000 tournament, of all the machines I could choose to train against, it would be a 32 bit Lang machine with hash tables. That would give me the very best match practice and would hone my strengths in order to successfully compete against the human players I would come up against.

Well from other posts that I am reading here, I guess it is a preference as a response to a persons style of play and enjoyment preference. Since I like to play aggressive and tactical, I tend to respect more opponents that I really have to concentrate much harder on otherwise as they will punish me mercilessly for any extravaganzas that I may play, and these are these King's, Morsch's and Schroeder computers. If I slip, I get nailed.

Give me swashbuckling 19th century play any day and any time!

A time of sacrifice and of masters of magic and gambits...

Adolf Harry Houdini Anderssen sends his sincere regards

Post by **Fernando** » Fri Sep 11, 2015 1:04 am

spacious_mind wrote:Wow, catching up on reading and seeing all the posts generated regarding playstyle preferences!

Monsieur Plastique wrote:
But Nick, I'm talking about chess play here at around the level of an amateur expert - say the very high 1900s to around 2100 per the FIDE rating scale. You are talking, however, as if I am using a FIDE Master paradigm to describe a Lang program. I'm not - a FIDE master is a lot stronger than any Lang program in any dedicated machine.
As I stated I just think that Lang especially in closed passive situations begins to dither about and probably because of the highly selective search misses moves that could allow it to proceed favorably, but instead continues to dither around.

Maybe it is a preference thing since I like to play very active and tactical and I just find Lang's style not too difficult to be successful against and really kind of boring. Also I acknowledge that when I am making this judgment I am comparing to programs that came out just a tad later, who I just find better all round and more balanced in their play. ie. King for which the weakest computer that I have is Risc 2500 and then move forward to R30 and Chessmachines. Same with Schroeder. Both programmers I find much tougher nuts crack on these computers.

If I were preparing to play in an under 2000 tournament, of all the machines I could choose to train against, it would be a 32 bit Lang machine with hash tables. That would give me the very best match practice and would hone my strengths in order to successfully compete against the human players I would come up against.
Well from other posts that I am reading here, I guess it is a preference as a response to a persons style of play and enjoyment preference. Since I like to play aggressive and tactical, I tend to respect more opponents that I really have to concentrate much harder on otherwise as they will punish me mercilessly for any extravaganzas that I may play, and these are these King's, Morsch's and Schroeder computers. If I slip, I get nailed.

Give me swashbuckling 19th century play any day and any time!

A time of sacrifice and of masters of magic and gambits...

Adolf Harry Houdini Anderssen sends his sincere regards

Then you must get a chess system tal and play it.
There you will find the measure of your tastes.
Yes, it is a software, but....

Fern

Reinfeld · Post by **Reinfeld** » Fri Sep 11, 2015 6:04 am

Well, does it have to be said? I'm getting in line for this machine. How often do we get a new one?

- R.

Monsieur Plastique · Post by **Monsieur Plastique** » Fri Sep 11, 2015 10:02 am

spacious_mind wrote:As I stated I just think that Lang especially in closed passive situations begins to dither about and probably because of the highly selective search misses moves that could allow it to proceed favorably, but instead continues to dither around.

Yes. Totally agreed. This is definitely a weakness with the Genius program on my Windows 8 phone so I suspect it will be a weakness in the upcoming unit as indeed it is a weakness in the PC program if you give it a split second a move!!

This characteristic tends to fade on the PC program at long time controls but obviously the Mephisto machines and the upcoming Millennium simply will not have the CPU power needed to fully overcome this.

Like you I seem to have an easier time with the Lang programs as well. Mind you they are still much stronger than I am but I remember defeating a 32 bit Portorose Munchen at blitz and being quite surprised as I did not really need to think that hard at all. It was quite passive and made the win quite easy. And I am saying this as someone who has never, ever defeated a Mephisto MM IV at blitz. And in the games I have been playing against my Genius phone, it is pretty even until I eventually get ground down in the endgame. Infact, If I were to play the game in "parts" so I was totally fresh every 20 moves I think I could beat it. It's not really like my Nintendo Fritz where fireworks fly and I'm down a whole piece before the endgame no matter how hard I try.

So I think we actually probably agree more than you think.

I think in the end I would not want to own just a Genius program / dedicated. But owning that along with a hash table Novag / top of the line Schroder or Morsch and you have completely contrasting flavours.

I guess I think the way I do because no one has ever wiped me off the board in a swashbuckling display of aggressive fireworks (well, except for that darned Klute). I tend to either win or die slowly. So I guess this is why for me the Langs emulate more my experiences against human competition. Of course, if people played against me like my Nintendo Fritz I'd likely start practising against that Chess system Tal at full strength and stuff the fact that I'd lose each and every time.

It would be interesting to see if anyone has actually done a formalised and statistically valid study on how chess styles might vary according to rating. Personally I tend to notice that up to around 1800 -1900 and especially around the 1400 - 1700 mark, the players will let it all hang out in the pursuit of glory, imagined or otherwise. Then between about 1900 and 2100, I think the players tend to be more solid and cautious, not always wanting to commit to something that may be dubious. Possibly because they are better at anticipating the strongest responses from their opponents. Then when you get into the master ranks, you tend to get the best of both worlds (fireworks with caution). I know this is very general and there are regular exceptions, but I honestly feel that is a fairly good summation.

spacious_mind · Post by **spacious_mind** » Fri Sep 11, 2015 11:45 am

Monsieur Plastique wrote: So I think we actually probably agree more than you think.

I know we agree more than we disagree. It was your suggestion as training partner prior to going into a tournament where we differed, and I am sure it is based on what I think I would need as partners to be most successful. And, that is computers that punish me for any foolish or inaccurate extravaganzas. In other words force me to consider their devious plans as I rush into mine so that I can smash, crush and destroy them and not the other way around

Best regards