Databases, Engines and Over-the-Horizon Killer Sacs

The following game is just fun – any instructional value is accidental.

Most modern players have great faith in their analytical engine, but it’s worth recalling that even 3400–rated monsters are not all-seeing. For sharp opening lines, a good database is just as essential as a strong engine. While browsing through a recent TWIC I spotted a perfect piece of computer-aided prep all the way to the end of the game. My guess is that White found the winning idea in his database rather than had it suggested by an engine.

Laurent – Gulbas, Belgium 2014

1.e4 d6 2.d4 Nf6 3.Nc3 g6 4.f4 Bg7 5.Nf3 0–0 6.e5!? Nfd7

As played by Gulbas before. 6…dxe5 is less wild, but may also offer White chances of an edge after 7.fxe5.

7.h4 c5 8.h5 cxd4 9.Qxd4 dxe5 10.Qf2 e4 led to a win for Black in Philipowski – Gulbas, Belgium 2010.

7…Nb6 is much safer, but leaves White’s centre looking solid. A recent game continued: 8.Bb3 Nc6 9.0–0 Bg4 10.Be3 Na5 11.Qe2 Qd7 12.h3 Bxf3 13.Rxf3 Nxb3 14.axb3² Pruijssers – Cuijpers, Netherlands 2014.

8.e6 Nb6
As played 41 times on my database, but probably close to lost. 8…fxe6 is ugly and admits Black is worse, but at least he is probably not getting mated.

9.exf7+ Kh8
This is where a good database is more valuable than a strong engine. The engine will keep suggesting that White move the attacked bishop. Perhaps the horizon will stretch far enough if you leave the engine fixed here overnight, but that is impractical for most people to do with every position in their repertoire.

Gloriously crude. White sees a king on h8 and a rook on h1, and that’s about all. The usual move is 10.Be2 but it is painfully feeble in comparison.

10…Bg4 does not stop the advance: 11.h5!

11.h5 Bf5 12.hxg6
This is a wonder-novelty. Well it was when Ivanisevic played it in 2012. 12.g4 and 12.Ng5 were the old messy moves.

The engine can quickly take it from here: after just a few seconds its first choices over the next few moves lead to a win.

13.f5! Bxf5 14.Ng5 Qd7
14…Qc8 is perhaps a more challenging defence, but still losing. 15.Qh5 h6 16.Nd5!+- If you see the Ivanisevic game below, perhaps this line is what White was remembering from his prep. Naturally I at first had no idea why going to c8 is tougher than d7. The point is that if White plays as in the game, the queen can slide along to g8: 16.d5 Ne5 17.Ne6 Rxf7 18.Bxh6 Bxh6 19.Qxh6+ Bh7 20.Ne4 Qg8 Black is just hanging on.

15.Qh5 h6

The engine gives this at once as winning.
The original game was less convincing but all good fun: 16.Nd5 e5 (16…cxd4!?) 17.g4 (17.Nf6!) 17…Bxc2 18.0–0 Bg6 19.Qxg6 Qxg4+ 20.Kh2 hxg5 21.Bxg5 Nd7 22.Rg1 Qf3 23.Bf6 Qf2+ 24.Rg2 1–0 Ivanisevic – Dzhumaev, Al-Ain 2012. So how could Dzhumaev have avoided this? Probably he couldn’t. Just accept that this sort of thing will happen occasionally if you play sharp lines.

16…Ne5 17.Ne6 Rxf7 18.Bxh6 Bxh6 19.Qxh6+ Bh7 20.Ne4 Qa4 21.N4g5 Kg8 22.Nxf7 Qe4+ 23.Kf1
Black resigned as 23…Qf5+ 24.Kg1 Qxf7 25.Rf1 is mating.

I would not be surprised if White had the whole game on his computer before the game started. How to avoid this happening to you? If you have a sharp forcing position in your repertoire (like 6…Nfd7) then you need to keep up to date with the latest games. Even the best engine won’t save you from an over-the-horizon killer sac.

28 thoughts on “Databases, Engines and Over-the-Horizon Killer Sacs”

  1. I suppose the reason that your engine doesn’t agree with h4 as first choice is, that 11.h5 can be answered with g5 – giving back the piece and probably holding the position. My engine finds the given line (starting at h4) within 2 minutes, but it’s not it’s first choice due to 11. … g5 instead of Bf5 (following line : 11. … g5 12.Qd3 Qd7 13.Nxg5 Qf5 14.Qxc4

  2. I may be mistaken, but the line was discussed in one of the last New in Chess Yearbooks or Informator, maybe this is equal to a database .

    Looking forward to the next books 🙂

  3. @SugarLips
    Interesting, which engine is preferring 11…g5 to 11…Bf5 ? The ones I tested all have 11…Bf5 as top choice.

    Also, just based on a few seconds of thought so not a solid analysis, perhaps 13.Qxc4 is better than 13.Ng5 in your line. I would guess White is a little better.

  4. “If you have a sharp forcing position in your repertoire (like 6…Nfd7) then you need to keep up to date with the latest games.”
    Sure but an amateur having a full time day job does in most cases not have the time to go each week through twics and other relevant analysis to maybe find something relevant for their repertoire. I remember that Kramnik said that going through the materials is a daily job just to keep up.
    Personally as an amateur you need to make a choice. Or you don’t play sharp stuff or you just accept that such hick-ups can happen.

  5. Talking about databases, do you have access to the latest correspondence games? Today this is only available to members. It is a goldmine for each repertoire. I wrote a blogarticle about it:
    On chesspub grandmaster Tony Kosten was not aware about correspondence developments in the Panovvariation of the Spanish Chigorin. Just to point out that even professional players are struggling to keep up with theory or worse have access to all available important information.

  6. @brabo

    Sure, amateurs have limited time to study chess. I know this as I basically am an amateur player who plays very occasionally. Playing chess is very definitely not my day job.

    These days, by using a computer wisely, spotting relavant new games in your repertoire can almost be an automatic process. At Kramnik’s level, with an elite repertoire with both colours, I can imagine it is an endless task. But keeping up to date with the Pirc, for example, is not such a tough task. Also, perfection is not required, just be better than your opponent. Tough for Kramnik, given who he plays, but not so scary for the rest of us.

    Databases: curiously, this subject was in our minds last week as one of our number put some days into improving our databases and ensuring we have as many good sources as possible. That certainly includes correspendence games.

  7. “These days, by using a computer wisely, spotting relevant new games in your repertoire can almost be an automatic process.”

    This sounds too simplistic to me unless I miss some easy tools.

    You first need to create a database with keypositions of your repertoire. Next you need to download the latest twic and finally you do a search through the twic with each of the keypositions. (I am not talking yet about magazines and books which are in most cases not yet digitized).

    This semi-automatic process must be repeated every week just not to be vulnerable to the kind of defeats mentioned in above article.

    How many keypositions are in a standard repertoire? 20, 30, 100 ? And this must be done every week?

    Sound not only very boring but still very time-consuming. I don’t think this can work for me as an amateur (+2300 rated).

  8. @brabo
    I think you’re being too perfectionistic here. You seem to forget that the person on the otehr side of the table is also an amateur and not Kramnik. In my own experience, just following e.g. or a similar source like the newsletter of Chessvibes is mroe than enough for the average amateur. This really doesn’t take that much time, unless you play the Najord of course. But amateurs who play the Najdorf should not complain about the time it takes to keep up to date :-).

  9. Fact is that Cemil Gulbas, an international master did not check the database for 1 year on this keyposition of the Pirc otherwise he would not have lost in this way.

    Bruno Laurent also an international master (former Belgian champion) is well known in Belgium for having excellent gamepreparations. Last year he defeated my team-capitain with another piece of nice preparation: I remember that my (nevertheless very experienced) team-capitain was upset about finding afterwards most of it back in the databases.

    I just give this little anecdote to put things in the right context.

    I strongly believe that most amateurs are not keeping up with the latest developments in their repertoire (for whatever reason) to avoid defeats like in above article. Bruno Laurent is probably an exception (also not sure as he maybe just detected a weak spot in their repertoire and exploited it successfully).

    The article let us believe that the exception of keeping up to date the repertoire should be the rule (for amateurs) and it is simple to follow the rule. My experience tells me the opposite so I disagree.

  10. @brabo
    Where did you read your conclusion in the article? It says that if you play sharp, forcing lines you’d better keep up to date with the latest games rather than relying too much on your engine. You can of course choose not to, but then you run the risk of running into home preparation. Seems sound advice to me! I didn’t read that keeping up to date the reportoire should be the rule – that’s your own generalisation.

  11. @Ray
    Rule or recommendation, I am no English expert but the 2 are for me closely related.
    I don’t believe IM Cemil Gulbas only relied on the engine for his analysis and didn’t consult the latest games when he studied this position. He is a way too strong player for that. No very likely it is more than a year ago that he studied this position. There is plenty of other stuff to study.
    Therefore I don’t believe this piece of advice, recommendation, rule or whatever you want or don’t want to name it, will help him.
    Players of this strength know that they need also to consult databases. Differently said, I don’t find this article particular valuable except of a funny anecdote that somebody run into some preparation.

  12. @Ray
    “It says that if you play sharp, forcing lines you’d better keep up to date with the latest games rather than relying too much on your engine.”
    Here I believe you make a mistake. In the article white not rather relies on the latest games but improves with an engine, see move 16.d5 ! and the remaining moves which are all the top choice of the engine for white. So it is not a story of rather but a story of doing both: database and engine. In fact the engine part will always be much more timeconsuming than the database check as you need to let the engine run in parallel for each move (at least a few seconds).
    Again adding it all up then this is not feasible for most amateurs to do such job in such way that a repertoire is up to date.

    Another small but important remark. I believe black didn’t look at this line in his preparation. In the first division of the Belgium interclubs (i play first board in a different team) it is very difficult to know in advance whom you have to play against as there is no limit on the number of players in a team. Besides if you had a look to my previous link then you will notice that Bruno Laurent likes to vary his openings (1.e4 in this example but 1.d4 in my example) so very difficult to forecast what exactly he will play.

  13. @brabo
    “For sharp opening lines, a good database is just as essential as a strong engine”

    “It says that if you play sharp, forcing lines you’d better keep up to date with the latest games rather than relying TOO much on your engine.”
    From what I can see, according to John, engines and databases go hand in hand. You need both to work well enough. So, what is your argument?
    It isn’t too difficult to update your Opening files every week. I follow the top Super GM games regularly, and I compare what is in my repertoire with those games. For example, the recent game Caruana – Aronian at the Zuerich Chess Challenge. I play the Marshall and I found a solution for black.

    Of course, a game played by an IM in this line is not easily noticed by us, agreed. But, as Ray said, chesspublishing or chessvibes should be sufficient.

    And also, your opponents who play this line with the opposite color also might have not noticed this improvement. The game must be hidden in one of millions in their databases.

    We can’t be perfect, there might be holes in the repertoire of top Super GM’s too, for example, check Fressinet – Bacrot 2013 FIDE GP at Paris. Demolished in a King’s Indian in 24 moves. With the amount of theory in the KID, 24 moves is really less. (And yes, I remember that game because I always look at the important events like these)

    So won’t daily knowledge of chesspub and such sites be sufficient for us amateurs?
    Do you really need to be that perfect?

  14. @KIA Fan
    The article describes a situation between 2 International masters (important information which the article doesn’t tell us). That are very strong amateurs which spend a lot of time in preparing and studying openings. Quality chess has a tendency to publish work for the very ambitious players (see e.g. grandmaster repertoire series). Indeed sticking to chesspub and such sites is more than sufficient for the average amateur but here I do believe John is looking for more.

    I do want to make a clear difference between professional players and (very) strong amateurs. You can’t expect that a remedy working for professional players can also work for the strong amateurs. Cemil Gulbas belongs to the top 1% of the strongest amateurs and missed the reference game. I am sure he knows the remedy but it is not that simple to execute as is told in this article.

    So I find it very normal that hick-ups like in this article do happen . I find it even a bit inappropriate to make fun of the black player which is by the way very sympathetic.

    Look I would’ve liked the article if it was written from white’s point of view. How to prepare properly on an opponent and to achieve an easy win. That would’ve given something any slightly ambitious amateur could do and actually would’ve been informative.

  15. @KIA Fan
    “It isn’t too difficult to update your Opening files every week. I follow the top Super GM games regularly, and I compare what is in my repertoire with those games.”
    Do you have a full time job, wife and 2 kids ,…? It sounds not.

    “So won’t daily knowledge of chesspub and such sites be sufficient for us amateurs?
    Do you really need to be that perfect?”
    I’ve last months almost exclusively been playing against IMs and GMs so are you competing on the same level? Are we talking about the same kind of preparations and openings needed?

    Again don’t forget that the article is giving an example about 2 IM’s (very close to my rating) which are both in their 30ties so in a very similar situation as myself.

  16. @brabo
    “I find it even a bit inappropriate to make fun of the black player”

    Who is making fun of the black player? Not me. Calling it a fun game is not the same as making fun of the black player.

    Brabo, I know English is not your first language, but I don’t think that is the problem here. The meaning which seems to appear in your brain is nothing like the words I have written. You are arguing with straw men.

  17. @John Shaw
    Hear hear :-). And maybe I’m wrong, but I was under the impression that the general audience for this blog are amateurs (or to be more precise I should say ‘untitled players’), who – one might assume – as a rule don’t play almost exclusively against IM’s and GM’s. If you play almost exclusively against IM’s and GM’s there seem all the more reasons for proper opening preparation (that is, if you have a sharp, forcing opening reportoire, since that was what we were talking about in the first place).

  18. @John Shaw
    I was making the following logic in my brain.

    1) You claim that being up to date in a Pirc is not a tough task and these days, by using a computer wisely, spotting relevant new games in your repertoire can almost be an automatic process.
    2) So if an IM (like black) didn’t spot a more than 1 year old relevant game in his favorite Pirc then he made a stupid easily to avoid mistake.
    3) Showing stupid easily avoidable mistakes of somebody is not nice.

    Ok sometimes mistakes must be shown to learn from it but this contradicts a bit the intro which states that any instructional value is accidental.

    Anyway i feel a bit sorry that I have to clarify this as I am not the black player and only know him from the Belgium circuit. I would much more prefer that we discuss instead how to optimize the process of keeping up to date the repertoire. How do you keep track and how much time do you spend averagely? That is where the real value is for any player (amateur/ semi-prof or prof). E.g. yesterday I learned via chesspub how I could download all 1004 twics in one go in an easy readable format (cbv so better than pgn). This let me win a lot of time but still the real work remains.

  19. @Ray
    There is a huge difference between having kids or not. It is no coincidence that many (most?) active strong (almost the same as ambitious) amateurs have no kids.

  20. Gilchrist is a Legend

    What about players who play sometimes against IMs and GMs? For example, 2200 to 2400 that are neither IM/GM but not exactly pure amateur? Saying that one is an amateur to a non-chess person would think that one only knows how to move the pieces or something, but what does it really mean in the chess world? I have played in amateur sections, but that was when I was 1600 and lower.

    Also there is a saying, “Schaken is een levensstijl”. Perhaps in reverse–one has time for chess, herefore refrain from marriage and having children to focus more thereon.

Leave a Comment

Your email address will not be published. Required fields are marked *

Scroll to Top