Chess Engines: A personal look (part 1)

Chess Engines: A personal look (part 1)

BrendanJNorman
BrendanJNorman
Jun 21, 2015, 1:32 AM |
3

So guys, It's been ages since I've written some chess related content and I've kind of missed it. 


In all my busyness and travel (I'm now living in China again) I've missed putting my chess thoughts, ideas and experiences down in writing as well as making videos for my YouTube subscribers and 4000 odd Facebook fans, plus the guys who I teach privately and whomever stumbles across my content on the so-called "inna-webz" ™.

 

(Above: Where I'm now living again... Jinan in China's Shandong Province)


After ongoing prompting from some of the above people, I've decided to start to create chess content again, but in future I'll try to focus more on what interests ME in chess (otherwise its boring right? Plus I'm sure a lot of you share my chess interests) and I'll be doing my best to avoid the boring, dry, over-diplomatic and wordy waffling that a lot of online writers (especially chess writers) suffer from.


 If I think a chess opening sucks I'll say it openly…If I think a certain GM (despite being an absolute genius over the board) is an arrogant pig, you'll hear it here first haha.


 Don't get me wrong though…I'm not going to go out of my way to create hype and controversy, but at the same time I'm not going to steer clear of my true opinions (often quite strong) as I did before, to avoid hurting some ultra-diplomatic, hyper sensitive keyboard jockey's feelings…I love chess and I love helping less experienced players to improve their understanding of the game.


 So let's get started shall we?


 One thing that has really fascinated me recently is chess engines.


 Thoughts on Chess Engines:


 My interest in chess engines is not so much in the strength of them at all..


 Sure, Komodo, Stockfish (plus all the other fishes) are super strong now and could eat Wonderboy Carlsen for breakfast with probably a score of 9.5-0.5, but doesn’t that get old after a while?


I, like a lot of decent players also get tired of arm-chair generals on sites like Chessdom (a site where live broadcasts of top tournaments also include move-by-move analysis from top engines) who, when Vishy Anand and co. make a single move which make the engine's heart jump, angrily type "OMG!! Anand sucks! He was good before, but now he sucks! The engine said 23.Rc1 is better".


Firstly,  of course even the top guys make regular mistakes by the standard of a 3400 rated engine, and if we ran the games of almost any top player in history through the engine, we'd get the same result.


Secondly, there are still positions that engines (even top ones) need some time to understand and even a patzer like me can sometimes see the engine's evaluation and say "Bull****! That’s not better for white! What if I do this?" *enters move*, only to see the engine say I'm completely wrong, then see my idea and have a heart attack in its evaluation before finally agreeing with me.


This DOES happen, but you have to be strong enough to use it as a TOOL, and not just consult it like some all-knowing, divine oracle.


Anyway, I went off on a tangent there…


What I was saying is that I am not really interested in the STRONGEST engines at all…


 What Interests me about Chess Engines:


 The thing which has been taking up a lot of my leisure time recently has been the search for an engine (no matter the strength) which most accurately simulates a TRUE human opponent.


What originally prompted this search was the fact that I decided to return to China to live and do some business (including a chess school for kids ,which I'll tell you about another time!!) and I really miss playing real chess.


Real chess and real time controls.


Sure, I could play on the internet, but this is not a good solution for two reasons:


 1. Due to having a mix of paranoid, corrupt politicians and an education system that teaches its 1.6 billion people to basically believe lies and renounce free thinking, China's internet needs to be strongly controlled/censored and so speeds are very very slow (read: frequent dropouts and speeds lower than dial-up a lot of the time).

There are good days and bad days, but my rating at blitz can drop from 2200+ to mid 2000s just due to disconnections and losses on time If I play a lot, so often I choose not to.

2. If I choose to play long time-control games, at least 50% of players in my rating range are cheating with engines anyway, so why not at lease choose which one I face?


So I've been looking for an engine who plays positionally well, but still misevaluates positions on occasion, an engine who is tactically solid (or even ambitious), but doesn’t blow me away with 13 move combinations and an engine who makes material sacrifices which are speculative in nature and gain some at least visual compensation or initiative, but not the obvious forced mate or regaining of material…they can be objectively unsound even.


I'm happy to report that I have found several that fit the bill and are fantastic training partners, even for players much weaker than me, now let me give some examples of their play.


The first I will look at is WChess 1.06...


WChess 1.06 is a very old engine, but one that plays with a rating of 2337 and has been decribed in the following way:


"Plays like a human grandmaster. Games are virtually indistinguishable"


Obviously this caught my attention, so lets see what stuck about about WChess.

1.Positionally well, but not perfect:

What follows is a game played at a 30 45 time control against WChess who plays the opening well and gets an advantage with the novelty 11.e4! after which white accepts an isolated d-pawn in exchange forvery active pieces.


The engine then makes a positionally suspect decision with 13.Bxf6?! (Simply 13.Rfe1+= is better) after which I manage to equalize.

After 15.Rfe1 it seems white is still quite active but with some careful exchanges I managed to hold the balance with some exchanges.

Aimless play in the ending had the engine skating on thin ice, but my inaccuracy with 28...e5 (28...Kf7-e8-d8 was the winning plan) made things difficult and eventually the game was drawn.

 

So, what do you think? Human-like or not?

Next post, I'll introduce the knowledge-based chess engine which has a very attractive (even unbelievable) tactical style which is attractive, but not overwhelmingly strong... A pleasure to play against!

Stay tuned.