All Rook endings are drawn

 

   …except those that aren’t.

     What about double Rook endings?  Are they more likely, less likely or likely to be drawn with the same frequency?  The 'all Rook endings are drawn' rule of thumb is probably a good one, but I know it's wrong.
     Mark Dvoretsky wrote, "...the number of positions that must be known exactly is relatively small. It is only in Rook endgames that it is essential to memorize thirty or forty concrete positions; in other types of endgames there are even fewer." For me, that’s way too many…beyond my ability actually, but it’s a moot point because Dvoretsky didn’t say what those 30-40 endings are.
     Karsten Müller in ChessBase Endings 13 did a CD on how to play double rook endings, but who has the time or gumption to study it? Fine’s Basic Chess Endings as been my ‘go to’ ending book for years, but there are only a few pages on double Rook endings in it. About all I was able to glean from it was superior side must take care to reduce to an ending which is favorable for him. That doesn’t tell me much though because I’m not always sure of 1) which side is superior and 2) which endings might be favorable.
     The endgame book I learned the most from was one I purchased back in the 1970’s by Peter Griffiths titled The Endings in Modern Theory and Practice. This is an excellent book, but you should probably be at least 1800 before tackling it and you’ll also need to be serious about studying it. Anyway, all Griffith had to say about double Rook endings was, “Sometimes they are nothing more than complicated versions of single Rook endings, but in many cases they have their own special characteristics…” Not much help and he only gave one gave example.
     The only thing I know for sure is that the main characteristic of such endings is the chance of doubling Rooks on the 2nd or 7th rank and this can often be so great an advantage that it usually overrides all other considerations, even a material advantage.
     In the following G15 we reached a double Rook ending early in the game and it appears that neither of us handled it well which is not surprising considering the complexity of the game.  In fact, anybody familiar with Pal Benko's long running endgame column in Chess Life will be aware that sometimes even Grandmasters don't always play Rook and Pawn endings well.  Also, given all the nuances of these endings, I'm not sure one can rely totally on engine analysis either.