[chessmind] Dennis Monokroussos: All rook endings are drawn...not!

[Email subscription to blog articles]

Posted by Dennis Monokroussos:
All rook endings are drawn...not!
http://chessmind.powerblogs.com/posts/1188447259.shtml


   If I had a quarter for every time I'd heard or read the phrase "all
   rook endings are drawn" (henceforth AREAD), I wouldn't be rich, but I
   could enjoy a nice vacation from the proceeds. This saying of the
   great player [1]Siegbert Tarrasch, much maligned for his dogmatic
   formulations of useful rules of thumb, is obviously false if taken
   literally. Even if we take it, as suggested in the previous sentence,
   as a rule of thumb, it still doesn't fare so well. There is a kernel
   of truth buried under the false and misleading aspects of the cliche,
   however, and we'll try to separate the wheat from the chaff in this
   post.

   ([2]show)

   Against AREAD, not only is it literally false (obviously), it's not
   even true that most rook endings are drawn. Worse still, according to
   the database-derived statistics given in Andrei Volokitin and Vladimir
   Grabinsky's [3]Perfect Your Chess (p. 38), bishop endings, queen
   endings, and knight endings are all more drawish than rook endings
   (the percentages are 47%, 43%, 40% and 38%, respectively); only pure
   king and pawn endings are less drawish, with a figure of 27%. Given
   these numbers, AREAD is not only an exaggeration; it's a risible
   falsehood. Is there any way to salvage Herr Doktor Tarrasch's
   reputation, at least with respect to this adage?
   I believe there is. What makes rook endings seem more drawish than
   they really are is the wide range of cases where a one- or even a
   two-pawn advantage will not suffice for a win. Let's consider some
   examples.
   First, of course, there are the simplest cases of rook and pawn vs.
   rook. These are almost always drawn if the defending king is in front
   of the pawn (the exceptions are well-known, so I won't bother
   detailing them here). Of course this is typical of the other endgames
   already mentioned, so this isn't too illuminating. There are more
   interesting draws, though - sometimes even if the king is cut off from
   the queening square or even the queening file the defender can often
   draw if his rook can give checks from the long side.
   There are plenty of cases with multiple pawns, too. If the pawns are
   all on the same side, being a pawn down is often tolerable - many such
   rook endings are not just drawn, they're easily drawn. This is not the
   case for knight endings, where 4 vs. 3 on the same side is possibly a
   forced win, even when the weaker side has an ideal pawn formation
   (e.g. Black with pawns on f7, g6 and h5). It goes almost without
   saying that a 4 vs. 3 advantage in a pure pawn ending is an incredibly
   easy win in almost every instance.
   Furthermore, in rook endings with pawns on both sides (i.e. the
   kingside and the queenside, not (just) Black and White), if the
   defender's rook is very active (e.g. on its seventh rank), a draw can
   often be achieved, as one of two scenarios will often arise: (1) both
   sides will go on a pawn-eating spree, typically leading to one of the
   draws mentioned above, or (2) the side with the extra pawn will be
   tied down to its protection, and unable to make further progress.
   There are cases when two extra pawns are not enough, many of which are
   quite normal and not something out of an endgame study competition.
   The best-known case is that of rook plus f- and h-pawn (or a- and
   c-pawn) against rook. As long as the defender's king is in front of
   the pawns and not cut off on the back rank, this ending generally
   should be drawn.
   Draws with two extra pawns is a common occurrence when one of the
   pawns is an a- or h-pawn. Here's a typical example: White: Kb5, Ra8,
   pa7, pg6; Black: Kg7, Ra1. White can make no progress, because the
   g-pawn is stuck and the rook can't move without losing his only trump,
   the a-pawn. White can try 1.Kb6, of course, hoping to move his rook,
   but the Black rook checks the king away from the pawn and then returns
   to the a-file (1...Rb1+ 2.Ka6 Ra1+ 3.Kb6 Rb1+ 4.Kc6 Ra1 etc.).
   Another, perhaps more surprising way to draw despite a two pawn
   disadvantage arises when the pawns are connected but blockaded. This
   pattern is not always a draw, but it is (a draw) often enough for the
   strong side to be wary of allowing such a blockade. Here's a sample
   draw: White: Kf4, Rb4, pg5, ph6; Black: Kg6, Ra6 (Kling & Horwitz,
   1851). It's a draw no matter whose move it is, e.g. 1.Rd4 Rb6 2.Rd8
   Rb4+ 3.Ke5 Rb7 4.Rg8+ Kh7 5.Rd8 Kg6! 6.Kf4 Rb4+ and White isn't making
   any progress.
   In addition to the drawing ideas given above, there are other, more
   particular cases less susceptible to useful generalization - the
   materially weaker side has a dangerous passed pawn, or enjoys
   attacking prospects that might result in perpetual check, or can
   utilize stalemate tricks, etc. The big point is that there are many
   typical situations where a material advantage in a rook ending is
   insufficient to win, and it is in this sense (and this sense alone)
   that we can extract some genuine wisdom from the poorly-phrased AREAD.
   (A challenge for the would-be Tarrasches - or better, [4]Tartakowers
   (who may actually be the culprit behind AREAD) - out there: can you
   think of a new slogan in place of AREAD that's short, catchy, and gets
   the main point right?)
   For further research, I've included some of the material above - both
   directly and as examples of the principles discussed here - at this
   [5]link. Readers might also want to check out this typically fine
   [6]article by Karsten Müller on the Chess Cafe website (this
   duplicates and extends several of the examples from my game list, as I
   noticed a little after the fact), this [7]Wikipedia article on rook
   endings, and on a very practical level, this old [8]post from my
   previous blog.
   ([9]hide)

References

   1. http://en.wikipedia.org/wiki/Siegbert_Tarrasch
   2. file://localhost/var/www/powerblogs/chessmind/posts/1188447259.html
   3. http://www.amazon.com/Perfect-Your-Chess-Andrei-Volokitin/dp/1904600824/ref=pd_bbs_sr_1/002-2873027-9598428?ie=UTF8&s=books&qid=1188467729&sr=8-1
   4. http://en.wikipedia.org/wiki/Savielly_Tartakower#Quotations
   5. http://chessmind.powerblogs.com/files/all_r_endings_are_drawn.htm
   6. http://www.chesscafe.com/text/mueller01.pdf
   7. http://en.wikipedia.org/wiki/Rook_and_pawn_versus_rook_endgame
   8. http://chessstuff.blogspot.com/2005/03/not-all-rook-endings-are-drawn.html
   9. file://localhost/var/www/powerblogs/chessmind/posts/1188447259.html