MATH, MATH, MATH

by Steve Lopez

A couple of useful items popped up this week on the Usenet chess newsgroups, so I thought I'd pass them along.

I'd often wondered why the Doctor? module seemed to analyze so quickly. When you run it in ChessBase, the dang thing revs up to a double-digit search depth in nothing flat. I hypothesized that this was due to a drastic pruning of the search tree; I thought that Doctor? only looked at four or five initial moves instead of all of the possible moves in the position.

A posted reply in the newsgroup corrected my misconception. It seems that the ply display in Doctor? is given in thirds of a ply. In other words, when Doctor? displays "18" as a search depth, it has actually searched six plies ahead (18 thirds). The idea is to give the user a more exact idea about how deeply the engine has searched the position.

In a related post, Bob Pawlak (he of the million dollars worth of software) asked how strong Doctor? is compared to the other Fritz engines. Bob, it might just be me and my peculiar style of play, but Doctor? gives me fits. Even on Friend and Sparring modes, the engine whups the tar out of me. So it's strong enough, and it has a different style of play than either Fritz5 or Hiarcs6.

I think the big question here is whether or not it's worthwhile for an individual user to invest in an additional engine. One of the best features of Fritz4 and 5 is the ability to use multiple playing engines within the same graphic interface. Each of the available engines (several for Fritz if you include tweaks within a release number, two for Hiarcs, and one for Doctor?, with more engines on the way) plays in a somewhat different style. In my opinion, it's great to be able to play several different computerized opponents without having to learn menus/commands/procedures for an entirely different interface each time.

But your mileage may vary, so it's up to you to decide if it's worth the extra ducats to buy another engine. For those considering additional engines for Fritz, I would consider buying the Hiarcs6 engine before buying the one for Doctor?. The reason for this is that Hiarcs6 plays a very positional style of chess, compared to Fritz's "hell-for-leather" tactics, creating a nice contrast. But Doctor? is tough and I'm sure you'll end up swearing at it before many games have passed.

Someone inquired about how Fritz derives it's handicap in Friend mode. Fritz programmer Matthias Wuellenweber (who has STILL not sent me a photo of his infant son Johannes as he once promised to do) explains the procedure in terms that only a mathematician could love. Here's my question, followed by Matthias' answer:

"...how does it derive the handicap? What basis does it use to assign a handicap? Does it go something like: for every game you lose, the handicap goes up by some set amount (say +100)? Or even some random amount (+39 to +111)?"

"It is an empirical, roughly linear function depending on the difference between the current score for Fritz and a 50% score."

My translation of what Matthias said: "Something like that, but we're not telling."

Dang! I knew I should have stayed awake in math class!

What he's actually saying is that if you should (by some quirk) draw all your games against Fritz, you'll have a zero handicap -- no advantage or disadvantage. The more you lose, the more material Fritz is willing to toss to play as badly as you (or else it will simply select moves from its analysis tree with less-than-optimum evaluation scores). The more you win, the better Fritz will play (by choosing moves with higher evaluations). The exact formula (the "empirical linear function"), however, understandably remains a trade secret.

Finally, a few weeks ago I asked for your input on changing the background color for this Web page. The vote is in (yes, friends, one vote). As it stands, both of my Web pages will remain the lovely shade of "Fruit of the Loom athletic briefs gray" that you're seeing now.



BOOKS, BOOKS, BOOKS

by Steve Lopez

I get a fair amount of e-mail about that other Web page: Battle Royale. By far, the question I'm asked most frequently is whether or not I can recommend other sources (Internet or otherwise) of games in which every move is annotated.

Unfortunately, there aren't many books written this way. The reason: it's a veritable ton of work. Annotating a forty move game in this manner is a minimum of fours hours worth of work. That's a minimum, mind you. Some of the games you've already seen in Battle Royale took 16 hours or more to annotate. This is also one reason why, starting sometime in late winter or early spring, Battle Royale will revert to a more conventional annotation style: only the highlights will be explained (the other being that it's likely the typical reader who has already plowed through the first 30 games will no longer need the move-by-move guidance).

There is other chess material available presented in move-by-move style. In perusing my own chess library, I was able to find a number of examples. Unfortunately, some of these books are out of print. However, I have seen all of the out of print volumes available in used book shops, remaindered book outlets, and chess book catalogs within the last year, so you should still be able to find them (and at a savings over the cover price, too!).

The granddaddy of these volumes is Logical Chess: Move by Move by Irving Chernev (Simon and Schuster, ISBN 0-671-21135-8). This one just went out of print in 1996, but I saw five or six copies for sale at a book outlet in West Virginia about a year ago, so copies shouldn't be too hard to locate.

Chernev dissects 33 chess games in intricate detail. The book is written with the beginning player in mind, so Chernev passes up explaining some deep strategic concepts in order to keep it simple. The book is divided into three sections: "The Kingside Attack", "The Queen's Pawn Opening", and "The Chess Master Explains His Ideas". Chernev sticks well to the topic and is adept at explaining what's happening.

You won't find too many epic confrontations in this book. Most of the games are between a recognized legendary master and some contemporary lesser light (but there are a few battles between titans in the book). All of the games are interesting, enjoyable, and highly instructive.

One caveat: the book was written in 1957, so the games are in English descriptive notation. Even so, the volume is well worth having and certainly worth the search.

A really rare bird is the book (or rather booklet) Move by Move by "Soviet Master Abramov" (translated by Roy DeVault; Chess Digest, no ISBN). This slim booklet was done with a typewriter and photocopier -- one of Ken Smith's "quickie" jobs from the old days. This 38-page booklet contains just eight games. None of them are annotated particularly well for beginners; much is left for the reader to figure out. It's more a book for intermediate players. I read my copy quickly one night while waiting for a game, and was neither challenged nor enlightened by it. Even so, it's an OK book -- get it if you can find it (I think Chess Digest still sells it) but don't kill yourself trying to find a copy.

Vastly more worthwhile are a couple of books by Bruce Pandolfini. I've been a fan of his work ever since I took up chess as a serious pursuit in 1989. One of the best things that came out of my full-time association with ChessBase from 1993 to 1996 was becoming friends with Bruce (so caveat emptor: my impartiality may be in question).

Bruce has written two books in which every move is explained. The first of these is Principles of the New Chess (Simon and Schuster, ISBN 0-671-60719-7). In this book, Pandolfini takes the concept to extremes: the entire 140 page book is a single annotated game. He takes a 23-move Scotch Opening and breaks it down into component parts. There are plenty of side trips along the way, with sections on opening tips, king safety, types of analysis, basic endgame tips, and other vital topics. It also contains plenty of illustrations so that a board is not required while reading the book. Overall, it's an excellent book for beginning and low intermediate players.

Bruce followed this up with Russian Chess (Simon and Schuster, ISBN 0-671-61984-5). This is one of my all-time favorite chess books, and I credit it for adding 50 points to my rating overnight. The book contains six games, with every move explained in a "question and answer" format. Whenever an important point is covered, it's denoted by the use of the word CONCEPT in bold capital letters. It also contains interesting and relevant quotes from chessplayers of Russian/Soviet extraction. I've read this book five or six times and I still return to it regularly; in fact, my original copy was becoming pretty ratty, so I recently purchased a second one. Russian Chess is, sadly, out of print, but you should be able to round up a copy at a used book store. By all means, move heaven and earth to get a copy of this book (which, I must confess, was a major inspiration for Battle Royale). Russian Chess is everything that an instructional chess book should be. I wish there were several dozen more books like it.

There are a couple of other books that deserve a mention as well, despite the fact that not every move is annotated.

The first of these is another Pandolfini effort: Kasparov and Deep Blue. It's an account of the 1997 match between the World Champion and IBM's silicon terror. I have mixed feelings about this book. First, if you're looking for background on the match or insights on how computers play chess, look elsewhere. Bruce is a chess guy, not a computer guy, and he's written a pure chess book here. Despite the fact that he frequently mentions that one of the combatants is a machine, the book reads pretty much like a standard match book in which both players are human. The focus of this volume is to explain from a chess standpoint why the moves were played. He occasionally makes reference to Garry playing an "anti-computer" move or Deep Blue playing something very "computer-like", but for the most part the games might as well have been played by two humans as far as Bruce is concerned.

The vast majority of the moves in this book are annotated; very few moves miss Pandolfini's scrutiny. I didn't do an official count, but I'd guess that 85% to 95% of the moves are explained.

One complaint you may have with this book is that it can be a bit of a tough read for a general-interest chess book. Most writers don't like to use the same word twice in a single paragraph; Bruce doesn't like to use the same on twice on the same page. This leads to more than a few rhetorical gyrations. Younger readers might want to keep a dictionary or thesaurus close at hand as they read this book. Bruce, in his quest to keep from repeating himself, goes a bit overboard in spots with the literary references, "ten dollar words", and somewhat flowery prose.

The games themselves are explained very well, and the book is aimed at the adult beginner to intermediate player. If the analysis you've read on the 'Net confused you and left you cold, check out this book. You'll understand the games completely after finishing it. If you want to better understand Deep Blue itself, however, you'll need to find another source; this is strictly a chess book, not a computer book.

The final book on my list is by "the Boss", Don Maddox, along with Bjarke Kristensen: Learn Chess from the World Champions (Dearborn Chess Library, ISBN 1-888320-00-1). The book began as a match book of the 1995 PCA World Championship but grew into something more: a book about chess strategy, aimed at late beginning to early intermediate players. It provides biographical information on all of the World Chess Champions from Steinitz to Kasparov, as well as valuable chess lessons drawn from their games. All games in which Kasparov faced Anand prior to and including the 1995 World Championship match are presented as well. Almost all of the moves are annotated (my guess is in the 80% to 90% range). The book is marred by a few historical inaccuracies (for example, Emanuel Lasker is named as the player who brought Capablanca's undefeated streak to an end at New York 1924; in fact, it was Richard Reti who handed Capa that loss), but don't let that deter you from reading this book. The annotations are excellent and the ideas are clearly and enjoyably explained.

I'm sure there are other books that could go on this list, but these are the ones with which I'm personally familiar. Happy reading!



A FINAL NOTE

Your questions, comments, and submissions are strongly encouraged. I'm particularly interested in receiving games for the Chess Embitterment column, as I flatly refuse to make fun of my own games week in and week out. We all make mistakes. Trust me -- you'll feel better if you talk about it.

Come on, make me laugh!