Wednesday, February 28, 2018

To analyze using a computer part 3

December was a dark month for Stockfish. First Stockfih was beaten by Alpha Zero see chess.com. Next Stockfish didn't manage to qualify for the TCEC-super-final of season 10 the first since years. If you would only look at those facts then many would decide to stop working with Stockfish.

However if we are investigating more closely the data then we see a very different picture. The beating by Alpha Zero was eventually only a difference of 64-36 which corresponds to about a gap of 100 elo see the fide-handbook. Besides there was a lot of critics on how Stockfish was operated. In my previous article I already explained how easily you can get a difference of a couple of hundred ratingpoints by just manipulating some parameters. This is not even taking into account the missing opening-book and the very limited hash-table which definitely also impacted the playing-strength of the engine. In short a match where Stockfish would get better conditions, could well reverse the final result but this will never be allowed of course by deepmind of Google.

Also the early elimination of Stockfish after stage 2 in season 10 was no disgrace at all. Contrary to previous years after stage 2 immediately the super-final was played. Consequently a lot depended on how the best engines performed against the weaker ones. Stockfish was the only engine not losing a game but finally ended a half point from the leaders. So the 2 finalists (Komodo en Houdini) never proofed their superiority. Personally I believe Stockfish is the strongest engine today available. After the final it won the TCEC-rapid and the newest release 9  (available since beginning of this month) leads at ccrl with 39 elo.

Naturally Chessbase doesn't report anything about the success of Stockfish on their website. The engine destroys their market. Even when a new top-engine like Houdini 6 is introduced then mainly negative comments are received. You really need to be stupid to pay 100 euro while a stronger engine is available for free. An open forum creates extra visitors but can also cause damage to your business. Personally I think such comments are inappropriate. First you get more than just an engine for the 100 euro. Also you can't expect that everybody wants to work for free. There is nothing wrong with trying to make a living from creating new interesting things.

Still if we only concentrate at the engine then we can wonder how necessary it is for ourselves to get an extra engine beside Stockfish. Do other engines have an added value for us? Well honestly I doubt it for 99,9% of the players. The quality of Stockfish's analysis suffice for any player till at least 2600 fide and maybe even higher. Only for some theoreticians like correspondence-players and the world-top in otb it becomes doubtful to only rely at Stockfish. I call it doubtful as today it is really not clear if an extra engine will still bring something extra. To support this statement, I made during the 2 weeks of the last Christmas holidays a special research-project.

In the 6th round of the last Open Leuven again like last summer at Gent see evolution, I didn't stand a chance against the Belgian IM Stefan Docx. Again I was surprised in the opening and was trailing the whole game. However this time the problem of the opening was more serious than last time. Despite many hours of analysis I was not able to repair the system at home. In the end I had to admit the opening was not fully correct so I should look for something else instead. However I am rather reluctant today to learn something completely new from scratch. I play few games so the work should be proportional. Stefan recommended the classical Dutch as that is the closest related to my repertoire. Just recently our current world-champion Magnus demonstrated the viability of this opening in a secret online blitz-game.
Nevertheless Magnus can probably play anything especially at blitz and still win. Even in that Magnus' game black's position was several moves pretty dubious. So I wasn't convinced yet to take up the classical Dutch in my repertoire. I needed to know more about the quality of the opening. Of course the first thing you need to check are recent articles/ books about this opening. The e-book The Killer Dutch, published in 2015 can probably considered today as the current best up to date theoretical work about this opening. The author is the English grandmaster Simon Williams. He is the biggest expert in the opening. Unfortunately from theoretical point of view this book comes short. Simon doesn't write for theoreticians but he explains the opening from a pure practical use in tournament-play. That is understandable as the practical player is his main reading-audience. So some critical lines are a little too easy categorized as harmless.

If no good theoretical references exist then the only thing which remains is to start your own research. However that is easier said than done. Even an opening like the classical Dutch contains today a myriad of variations. Below you see a screenshot of my current personal opening-book only built with games of the Megadatase in which at least one color has + 2300 fide.
That is +1700 games and we still need to add the correspondence-games and the engine-games which also could influence the evaluation of the different lines. In my article studying openings part 2 I already explained that 100 games often take about a week to digest. So I realized in advance that I had to change my working-methods to process +1000 games within an acceptable time-frame. The first adaptation was to prioritize the lines played the most often in practice instead of the lines recommended by the engine. Especially at a very early stage of the opening which is here the case, we often see that the engine plays inferior moves compared to a strong opening-book (see also the earlier reference to the match between Alpha Zero and Stockfish). It is necessary to analyze side-lines to support the main-lines but analyzing side-lines to detect the main-lines is mainly a waste of time.

A second important time-improvement without loss of quality was expected by not checking everything anymore by a 2nd engine (see my old article to analyze using a computer part 1). It is very time-consuming to switch between engines even if 2 computers are used. That is the link to my introduction in which I announced a special research-project. Instead of checking everything twice, I only did for some positions which I considered critical for the evaluation of the opening. This means only the positions where best play of both sides could/ would still give an edge to white by the first engine.

In the end only 18 positions remained to be checked by a second engine. The result was stunning. Only for 3 positions there was a conflict due to the smallest possible difference of only 1 hundredth of a pawn. I discussed this absurd phenomenon in my article annotations. I like to use strict boundaries to achieve a very objective method of analyzing but in some exceptional cases this can create some weird evaluations. In other words the only conclusion of the project is that the extra analysis with the 2nd engine was not adding anything substantial.

It again demonstrates how drastic the engines have evolved in the last years. 10 years ago it was rare that 2 top-engines were so often agreeing. You could easily find positions in which one engine would tell you that white is winning while the other one would state black is winning. Extra analysis were necessary to find who was right or wrong. In some cases the truth was somewhere in middle. Today not only we see top-engines prefer the same move but also sometimes show the same main-line. In fact it is not so surprising as engines are getting closer to perfection. Besides Stockfich applies an open source strategy. Everybody is allowed to see their code and learn from it. Of course other top-engines copy stuff which again diminishes the differences between them. Before the old top-engines were more closed.

So this means we can skip the 2nd engine forever. No, that conclusion we can't make yet. It is not because we don't see differences for the classical Dutch that there exist none between the engines. The qualities of an engine are not depending on just one opening. It is also the most important reason why tests of engines happen with a wide range of openings. Recently I detected a serious difference of evaluation in a position popping up from a Spanish Breyer-opening. See below screenshot in which we see both engines calculating in parallel.


While Stockfish claims a big advantage, Komodo states it is approximately equal. So for some special positions some extra analysis remains necessary to know which engine to trust. Anyway the number of special position quickly diminishes.

Theoreticians will still have to use a second engine. For players not interested in maximum quality, it makes no sense anymore to buy a 2nd engine. I will use also less a 2nd engine in the future. Only for detecting small differences (0,3 pawn) I still see an added value. Today a 2nd engine has become redundant in positions with an evaluation higher than +2 or lower than -2.

Finally maybe some reader wonders what I concluded about the classical Dutch. A very concise summary of the weeks of analysis can be found below.

Brabo

Addendum 21 March 2018
At http://www.chesspub.com/cgi-bin/chess/YaBB.pl?num=1369191586/75 an important improvement was shown for black by the German FM Stefan Buecker in the classical Dutch. The novelty 13...a5 instead of the played 13...Bb7 revitalizes this line.

Tuesday, February 20, 2018

To analyze using a computer part 2

More than 2 decades ago I started to play chess. At that time a lot of scepticism existed about engines. Many avoided them as those things were considered not to play real chess and surely won't be able to teach you anything. This feeling was very widespread as even till a couple of years ago some older players still had never worked with any engine despite playing chess sometimes already for decades.

Such thoughts seem unimaginable for today's young generation used to work with many different chess-programs. Few will still deny that a computer can be very useful to learn chess. Besides we see everyday new programs popping up extending the current applications. Obviously this changed world has created the need for guidance. However as was already valid about 6 years ago see part 1, today still very few or no good literature can be found about this subject. Therefore I expect we will see an increasing demand for special info-sessions about how and which chess-software to use. Recently I gave presentations for LSV and Mechelen. The +60 years old Belgian FM Johan Goormachtigh has offered 5 courses in KGSRL (Gent) which will also include how to work with Chessbase. I guess his content will be limited to the basics considering the target-audience and his rather sporadic use of the tools.

So every ambitious player will today submit his games to the critical evaluations of an engine. Nowadays probably the DroidFish Chess app is the most used engine to get a quick verdict of a game. Almost 100% of the young players possess a smartphone so after their games they will use it to quickly get an assessment of the moves. Unfortunately most of them won't look at the game anymore at home.  2 months ago there was a discussion at chesspub about how useful such homework could still be. Some poster claimed that I am just losing my time by analyzing deeper my games and it would be better to spend that time to other chess-activities like reading chess-books, solving exercises,...

He is definitely not alone with this view. In the past more than one international master told me they were spending much less time at analyzing their played games. So maybe indeed there are better methods to improve at chess. On the other hand it is not a hard proof. Maybe those international masters would've been today (much) stronger if they would have spent more time at analyzing their games. Another argument is that the current top-engines are so strong that it makes no sense to  give an engine a lot of time to analyze the moves. I made a quick experiment to test this statement by running a full analysis using the Fritz 15 interface with below hyper-fast configuration and using the engine Komodo 11.
This means Komodo gets only 1 second per move. Next I use a threshold of 30 so if my move deviates 0,3 pawns from the best move then the engine will comment. Finally for the opening I use as reference-database a correspondence database. For the test I use my game against the Dutch IM Xander Wemmers which was already covered in my article secret as that game contains a number of mistakes. The amazing output is produced in just a couple of minutes by the machine.

I and Xander belong to the 1% best players but still at the rate of 1 second per move Komodo 11 could discover all the important mistakes while only running on a very moderate portable. It just shows how much stronger our best engines are today. Even with some very serious handicaps the devices still play much better than us. I made a small study to illustrate how much better this exactly is.
  • Estimated fide-rating of the current top-engine at the rate of 1 minute per move: 3200 (CCRL and SSDF talk about 3400 elo but I think the actual fide-elo can be easily a couple of hundreds lower.)
  • 1 year old top-engine: -52 elo (see my article rise of the machines part 2)
  • Extra ply : 66 elo (see citeseerx.ist.psu.edu)
  • Extra engine running in parallel: - 1 ply (some tests on my laptop)
  • Extra line running in parallel: - 1 ply (some tests on my laptop)
  • Halving the time: - 1 ply (see wikispaces.com/Depth)
Example: 3 year old engine, 2 engines running in parallel, 15 seconds per move, 3 lines running in parallel
  • Start-base : 3200 elo
  • 3 year old engine: - 156 elo
  • 2 engines running in parallel: - 66 elo
  • 15 seconds per move: - 132 elo
  • 3 lines running in parallel:  -132 elo
Analysis are happening at the strength of 2714 elo.

So even if you use an outdated engine in a crippled way then the output is still at the level of a super-grandmaster. Shouldn't that be sufficient for us non professionals? Well I think we should be careful. It does make sense to strive for the highest quality. An opening-analysis is something you want to reuse later so to avoid any rework it is better to do it properly. Besides even killer-novelties can occur on amateur-level.

Also we should admit that the output of an engine is often very difficult to interpret correctly (see above dump). It is one thing to know where the mistakes are made but to understand and correct them is sometimes very hard. Often a lot of extra analyzing is necessary to get a full diagnosis. Check my article the butterfly-effect in which I demonstrated how I discovered by making some deep analysis why a small change of the position creates a big change of the evaluation. A technique which I often use for this kind of work is to let the engine play against itself.

Finally I also believe that working daily with the top-engines will improve your own understanding of chess. You need a tool to detect the small positional errors as it is not enough to just correct the big errors to become a master. Also just looking continuously to strong moves will be very good for your own development. Any trainer will recommend their students to watch and study grandmastergames and engines play even better than any grandmaster. The same comment I make about the lomonosov tablebases. No human is capable to copy the accuracy of the tablebases. Still I do experience that just consulting regularly the tablebases has improved my intuition in many endgames. I can much better estimate which endgames have or do not have serious winning chances.

When I started to analyze my games in 1990 using the assistance of my very first table-computer Mephisto Europa A (playing-strength 1700 elo) I was obliged to spend a lot of time just to achieve a minimum level of accuracy. This need doesn't exist anymore today but I learned in the meanwhile there are other reasons to still analyze deeply your own games. In any case it is much worse for your own development to analyze too little than to analyze too much.

Brabo

Tuesday, February 13, 2018

UltraCorr-X

The buzzword Internet of things sounds terrible but at the same time describes very well a very fast growing new phenomenon. More and more (new) applications are monitored and operated via the internet instead of humans. Also chess has changed. Online databases exist since the beginning of the internet but the pioneer chess.db was the first in 2012 to update at daily basis a database of played games. Suddenly it wasn't necessary anymore to buy expensive databases and update them regularly. By then end of 2013 already 40% of all players and coaches at the world youth-championship in Al Ain were using this online database see statistics chess.db.

Of course Chessbase didn't wait to counterattack and launched its own online database beginning of 2014. They update also daily their database with new games but were able to connect this new tool to their mainproduct Chessbase 14 (still release 12 in the year 2014)  This allowed them to gain back a lot of customers as it is much easier to prepare games using just one program. Later that year Chessbase also launched their cloud applications see part 1 and part 2. Sharing databases between people or just different devices (smartphone, tablet, computer,...) becomes very easy via the cloud.

Personally I am not yet adapting myself to this new internet-revolution. Today the internet-coverage is not yet 100% everywhere. A week ago there was an article at hln that still 39 Walloon cities have no 4G and even making a mobile call is difficult see here. Besides there are often extra costs connected to the internet. You need an internet-provider and the owner of the database can charge you something. Also big online databases have extra limitations as the functions are often restricted to avoid overload of the servers.

So I prefer databases stored directly on the hard-disk of my computer. Of course you lose extra time to regularly update those databases with the newest games or to synchronize your devices. A reaction at my LSV-session about chess-programs was that this seems a very cum-cumbersome task. If you are just an amateur then you don't want to spend daily some time at it. I fully understood this comment as I neither want to do that. To get a maximum return I decided to only twice per year update my databases. I do that just before the only 2 open tournaments (Gent and Leuven) I play each year. For the Belgium interclubs I fill the gaps by quickly checking chess.db to find out if there is something else I should analyze.

The updating process was described in my article using databases. A bit more than 1 hour this process takes and till recently I was satisfied about the results. Until recently as 2 months ago I detected by coincidence via the online chessdatabase that I had missed 2 important reference-games in my database. The first one was played at the Deutsche E-mail-Schachclub.
A second game was played at the Lechenicher Chess-server. Black was the Belg Aime Truyens.
So the killer-novelty which I discussed in my last article, was not so new and was already played twice. That is just 1 specific position in which I found out a gap in my nevertheless freshly updated databases but likely there are many more of them. I am not collecting those type of correspondence-games not played at iccf. I realized that I should do so I started to look around willing to eventually pay a small sum for it.

My first address is of course Chessbase. They offer Corr Database 2018. 1,4 million games but not cheap at all as they ask 189,9 euro. Openingmaster offers us Om Corr. 1,7 million games against a very reasonable 39 euro/ year. However the cheapest offer is not always the best offer. A recent review at chesspub says the service is pretty bad. In the past I've detected openingmaster was often for long periods of time inactive. Finally there is UltraCorr-X made by the Irish Senior International Master Tim Harding. You get 1,7 million games for 52,5 euro including the above games I was missing.

As I had already once bought a correspondence-database from Tim of good quality, I didn't doubt to choose his product again. I didn't regret as again the quality is excellent. One little remark for new users is that you need to remove the encryption after downloading the database to get access. The encryption-key you get from Tim but you need Chessbase to activate it. The Fritz 15 interface can not do this job but I got a hint from one of my students. You can download for free Chessbase reader which does have the feature to activate the key. I strongly recommend UltraCorr-X for any ambitious +2200 fide rated player.

Brabo

Tuesday, February 6, 2018

The (non-) sense of blitz part 2

Teaching children keeps yourself young of mind but I rather get the opposite feeling. Some of them consider me a dinosaur from a distinct past. Anyone of my generation knows what a rotary telephone is but for today's youth this device is very confusing see a funny article at hln. Also other topics like technology, music or chess are all creating a big gap between myself and my students. Recently I was again surprised that even a group of +20 years old players never heard before of twic or ssdf. Twic already exists since 1994 and is one of the very first if not the first newssite about chess. It is still active today. SSDF is the still running computer-ranking created in 1984!

Briefly many things has changed since I started to play competitive chess. Many players have quit the game party because the technological revolution took away their pleasure. Others like myself tried to adapt themselves and even took advantage of the new possibilities. Personally I made enormous changes during the last 2 decades in my methods of how to prepare for a game. Not everything became more complicated as nobody wants to return to the era when we had to deal with suitcases full of heavy (paper) chess-books. Below I summarize the different methods which during those years were introduced in my personal game-preparation-evolution. Some exact dates couldn't be found back so were estimated.
  • 1996: I bought my first PC and Fritz-program. I made my first game-preparation based on my knowledge of previous games against the players. At that time I used my reasonable up to date (paper) opening-books to prepare and checked briefly some lines with an engine.
  • 1997: I started to play against stronger players. Some of them had some games in the very first databases which I looked up to check their repertoire.
  • 2003: I started to use the engine openingbook which was offered together with an engine. This way I got a better view of the different possibilities in an opening.
  • 2005: I created a database to collect and store my game-preparations per color. I realized that often there is very little time to prepare especially at the French interclub so you need to organize yourself.
  • 2007: I created a database of games played by players of Deurne. After the birth of my first child I stopped playing abroad and started to play the local clubchampionship. As commercial databases often don't include games of lower rated players, I felt the urge to collect and store myself some of those games which could be relevant for myself.
  • 2010: I started to check correspondence-games and engine-games to find ideas which can be used for a game-preparation.
  • 2012: I created a detail-database to deepen my opening-analysis of played tournament games so I can use this in my game-preparation.
  • 2013: I created my first engine-openingbook based upon standard tournament games played by +2300 players as the commercial engine-openingbooks are very quickly outdated.
  • 2013: I started to download regularly twic, iccf- and engine games to consult them during the game-preparations. I also started to check online databases like chess.db to get better updated of the very latest developments.
  • 2015: I introduced monte carlo system in my game-preparations to get a quick idea in an opening with very few or no reference-games. Fast analysis is becoming very interesting due to the ever stronger engines.
  • 2017: I bought Chess position trainer which I use today to practice some very specific opening-analysis in a game-preparation.
  • 2017: I started to use in a more structured method my online played blitz-games in a game-preparation.
So my most recent preparation-tool is a more systematic use of my online played blitzgames. In this article I will demonstrate how this can be done efficiently if 2 conditions are fulfilled. First you need to have an easy access to your online played games. That is the main-reason why I today prefer Playchess. Very few or no other interfaces take care of an automatic storage of the played games on the computer see the database myinternetgames.cbh which I mentioned in my previous article. The second obvious condition is that you play the same stuff in blitz as in standard chess. Some blitz-players have a special repertoire only fitting in blitz but they won't be able to use that knowledge to prepare games for standard chess.

A successful example of using blitz-experience in a game-preparation was implemented in the final opening-position which was closing the previous article. After whites 8th move there exist almost no games in the commercial databases. You could activate an engine to get an idea but that doesn't make much sense in such type of non forcible position. Black has too many options to just look at the evaluation of an engine. Much more useful is to check my online played games to see what in practice people like to play in that particular position. As blitz is chess of a rather low quality, I use a filter just looking to my lost games. Below we see a screenshot of my lost games.


Next I will check one by one with an engine which mistakes I made and especially if something could be improved in the opening. One of them was a game I lost after 8...c6 (see above marked in yellow). When I fed the position at the engine, it played a very remarkable gambit see the screenshot below.


Of course I spent some time to understand the consequences of this gambit. I repeated this for my other lost games (only 11 so not that much) hereby discovering some more nice novelties. Once this was finished I stopped the game-preparation. As my opponent has only a very limited amount of games in the database, any other preparation sounded useless. A couple of hours later our game started.
I was of course excited during the game when I found out that Edouard felt for one of my prepared traps. It was even difficult to hide my emotions not to spoil the surprise. In the end I won the game comfortably and only gave away 1 opportunity to escape in an already very complicated position. My opponent applauded me afterwards for my ingenuity but I admitted that I borrowed the fantastic idea from a +3000 rated engine.

Coincidence some readers will probably think. Maybe but these things happened to me too often. I regularly see that moves played in blitz will also occur in standard chess. Especially in positions with few tactics experienced players are mainly relying on instincts. My personal database of online played games contains already more than 60.000 games. That is a lot of material which I can use in a game-preparation.

Brabo