Saturday 4 September 2010

"Castle because you have to, not because you can"

Richard Reti
Words of wisdom from Richard Reti.
This is one of the first pieces of advice that beginners are given by their more experienced colleagues. If the beginner has the temerity to ask “Why?” then a well versed answer is recited in the manner of a priest quoting scripture.

“Get your King out the centre and away from the open files. Develop a Rook by bringing it in from the cold and towards the centre (where the open files are!)”

On this occasion I have quoted fellow blogger Greenpawn34 but a paraphrased version of this answer is essentially the one that any experienced player will pass on (often quite condescendingly) to his or her young sensei.

There are exceptions to every rule of course. I can certainly think of some examples from Grandmaster play where one player has chosen the perfect moment not to castle and gone on to win brilliantly (perhaps I’ll share one or two in a future post). However, the explanation for these exceptions is simply that, in most cases, it would have been a very bad idea to castle. This is almost always due to tactical considerations. In other words, if they’d castled there would have been a high chance of being mated!

If an experienced player is doing his or her job properly they ought to be able to produce some sort of example with which to demonstrate the vulnerability of a king that has been left in the centre of the board for too long. This usually involves the bloody and spectacular demolition of a hapless beginner at the hands of a more experienced opponent. I have my own example of course!


What most experts don’t tend to do is give the beginner a broader statistical overview of why you should castle. This perspective is even more compelling than providing an isolated example as Green Pawn demonstrates ably in a recent blog post on Redhotpawn.

"Got some cold stats for you. On the 1400 DB (a database of games played on Redhotpawn) there have been 19,250 checkmates between 6 and 20 moves. I chose cold checkmates because it is King safety we are talking about and it also cuts out all the resignations due to a pieces lost and time outs.

Greenpawn34
  • White mates = 11,637
  • Black mates = 7,613

  • White 0-0 and got mated = 3,444
  • White 0-0-0 and got mated = 527
  • White uncastled and mated = 3,642

  • Black 0-0 and got mated = 4,073
  • Black 0-0-0 and got mated = 618
  • Black uncastled and mated= 6,946

Not much of a difference between White 0-0 and not castling is there? Think this highlights the inability of a weaker player to defend himself against attacks to his King. Safest advice to give to a less experienced player based on these figures is to Castle Queenside.

Ok. So, maybe we wouldn’t always advise a less experienced player to castle Queen’s side but the statistics tell an interesting story. In well over half of the games where black got mated in under 20 moves he had failed to castle. When white has been mated in under 20 moves he has been uncastled in about one third of those games. These figures are really interesting and should convince even the most sceptical of beginners to adhere to Richard Reti's commandment!

1 comment:

Unknown said...

I think greenpawn34 has grossly misused statistics! Without knowing in how many games overall 0-0 and 0-0-0 occurred we cannot know how dangerous castling is. For example, there may have been only 527 games played on redhotpawn in which white castled queenside and 344,400 games in which white castled kingside, giving the chance of mate if castled queenside as 100% and the chance of mate if castled kingside as 1%!