"Had a Lot of Fun, Could Have Been a Lot Better" or, Cheap at Twice the Price by Taylor Kingston Master Chess 98, CD-ROM computer software, Chess For Less, Riviera Beach, Florida, USA, 1998, $49.00. Over the last 8 or 10 years, the use of chess databases has become increasingly common, for a variety of purposes. For the aspiring student, they provide probably the fastest way to study a great number of master games. For writers, historians and scholars, they provide quick access to thousands of games, many obscure, which one would otherwise have to search many books to find. For correspondence players and game annotators, they are useful as openings references and in finding positions similar to those being analyzed. Professional players use them to study the games of prospective opponents in upcoming tournaments, either leisurely on a computer at home, or sometimes hurriedly on laptops just after the pairings are announced at a big Swiss-system event. However, until very recently, databases have tended to be on the expensive side, some costing over $500 and providing relatively few games. As storage devices became cheaper and more efficient, though, and as the PC-related chess market has expanded, it was inevitable that lower-priced databases would appear. It is this niche that Chess For Less has attempted to fill, first with the Essentia database, and now with Master Chess 98, incorporating their Ultimate Game Collection III. Though their product has some notable flaws, and falls somewhat short of their claims, it provides more games than any other database to date, and (with one major exception) gives adequate performance at a surprisingly low price. MC98 is basically a combination of three main features: a database program called Super Essentia, the Ultimate Game Collection III (1.1 million games ranging from May 1998 back to 1485), and the Zarkov 4.3 chess-playing program, which, it is claimed, has an Elo rating of 2500 at normal (viz. 40/2) time levels, and 2810 at blitz. I will consider each of these components in turn. The database program is a much improved version of its predecessor Essentia, which had several serious limitations. Essentia returned search results in groups of only 250 games, requiring tiresome repetitions if a search found many matches. Also multiple search results by different criteria (e.g. games won by Kasparov against Karpov as both White and Black) could not be combined into one list. It also claimed a very small search depth of 8 moves (which in practice was only 7.5). Super Essentia, in contrast, creates a full list of all matching games in one pass. If one changes search criteria, the list can be cleared, or the new matches added to the existing list. One can define the search by name, result, ECO code(s), year(s), and/or move order. A 64-move search depth is claimed, but I found it refused to go past 50. Still, that should be adequate for almost any purpose. One merely commands "set move order" and makes the moves by clicking and dragging with the mouse on the display board to the desired position. When using an exact order search, matching games are returned very promptly, as the program continuously narrows its search as one enters moves (transpositional searches are another matter; more on that below). The search list can then be scanned sequentially, or one can move back and forth through it at will. Having chosen a game, one can print the score, bring up its position at any point, click through it manually, or have it self-animate. This last feature has a bug: animation goes at a too-quick pace of 1 move per second no matter what setting one chooses. A free correction update is promised, available for download from C4L's website www.bookup.com. I also experienced a problem with printout. On one of my two PCs printout was quite legible, but on the other Super Essentia latched on to some unreadable font that looked like a combination of Informant symbols and Norse runes. C4L's online support had no cure for this. UGC III's increase of 100,000 games over Essentia appears due mainly to having more games by currently active players. For example Essentia had 1629 games by world champion Kasparov, while UGC III has 1701. Other top GMs show similar increases: Anand from 1094 to 1180, Kramnik from 785 to 865, Karpov from 2357 to 2518, Shirov from 1022 to 1117. However, I was disappointed not to see any increase among players of past eras. While of course they aren't playing any new games, in many cases recent research has rediscovered games that merit inclusion. For example Alekhine, Keres, Charousek, and Anderssen, to name only four, have all had major new collections recently published, but their UGC III numbers are the same as before (1526, 1622, 155, and 800, respectively). While GMs boast the lion's share of the games on UGC, one needn't be rated 2600 to appear. Many players at or even below national master level appear, and I would say there's an excellent chance that you or someone you know personally are on it. I found many of my past postal chess opponents, several players from my own chessic backwater of Vermont such as Steven Winer and Matt Noble, old friends from 20 years ago in San Diego, and there were even five games by a guy named Hanon Russell (from Connecticut, I believe). It is clear the compilers are drawing from small publications, as they added several recent games by Winer which have appeared only in my own Vermont newsletter and Massachusetts' Chess Horizons. I felt a bit miffed at being left out myself, especially considering that I have notched wins against many players on UGC III. On second thought, though, by that standard Chess For Less would be obliged to include games from everyone who has ever beaten me, which would probably require a whole 'nother disk. Minor players of the past are not ignored on UGC III either. Ever hear of Cyril Vansittart? He was an also-ran in the Vizayanagaram Tournament, a minor event collateral with the major London 1883 tournament won by Zukertort. Vansittart's was not exactly an achievement to guarantee chessic immortality, yet now his games live again on compact disc. While there are some gaps in UGC's coverage (several Vizayanagaram players are missing!), in the main I was pleased at the level of inclusion of obscure players. MC98's blurb claims that its playing program Zarkov "is extremely strong tactically and has excellent positional judgment". This appears to be only half-true. Its tactical strength is hard to dispute: it routinely beat me easily in speed chess, even giving 10-minutes- to-1 odds (though at blitz I resemble Bobby Fischer mainly in the sense that I too am an adult male Caucasian). However its positional judgment, like that of many programs, occasionally appears suspect, for example in this position (See Diagram) White: Kg1, Qe2, Ra1, Rc1, Bc5; pawns - a4, d4, e3, f2, g3, h2 Black: Kg8, Qg5, Rb8, Rc6, Nd6; pawns - a6, c7, d5, f7, g7, h7 a good human player would see the advantage of 1. ... Nc4, putting the knight on a strong outpost and masking the weakness of the backward pawn on the c-file. Zarkov instead played 1. ... Qg6, and after 2. Bxd6 Qxd6 I was able to draw the game. Also like many computers, Zarkov can be extremely materialistic, to the detriment of positional factors. However, that often does not matter, its tactical skill being so high. As an example, in analyzing this complex position (See Diagram), White: Ke1, Qd6, Ra1, Nc3, Nh5,Bc4,Be3; pawns - a2, b2, c2, d7, f2, g4 Black: Kf8, Qd8,Ra8, Rh8 Nb8, Ne7, Bh7; pawns - a7, b7, c6, e4, e5, f7, g7, h6 from Bent-Parham, Cincinnati Open 1993, it found the winning move 1. Qf6! in about one second (something neither player did in the actual game, nor several prize judges in later analysis). For another example, I gave it this situation from a postal game of mine, Applebee-Kingston 1982 (See Diagram). White: Kg1, Qb4, Rd6,Rf1, Nh5, Bc6; pawns - a4, e3, f2, g2, h2 Black: Kg8, Qe7,Rc8, Rf7, Nc3, Bb3, Bh6; pawns - f5, h7 I recall spending hours, over the space of three days, before settling on 30. ... Be6!, 0-1. I was gratified to see Zarkov come to the same verdict, but chagrined and humbled that it did so in 10 seconds. This was on a 120-mHz Pentium machine, relatively clunky compared to PCs available now. Zarkov would only improve on today's 450-mHz machines. Zarkov comes with 19 different opening repertoires, some designed after those of famous GMs. One can choose among them to face almost any style of opening as Black or White. I did note a couple of unsound lines. The "B12.ZDB" book consistently chose a bad variation of the Caro-Kann: 1. e4 c6 2. d4 d5 3. ed5 cd5 4. c4 Nc6? (See Diagram), from which I could transpose to a refuted defense against the Smith-Morra Gambit Sicilian by 5. cd5 Qxd5 6. Nf3 Bg4 7. Nc3! (a sample continuation being 7. ... Qd8 8. d5 Ne5?? 9. Nxe5! Bxd1 10. Bb5+ etc.). Against the Center Attack line of the Ruy Lopez two books (KRAMNIKW.ZDB and POLGARW.ZDB) would consistently play 1. e4 e5 2. Nf3 Nc6 3. Bb5 a6 4. Ba4 Nf6 5. d4 ed4 6. 0-0 b5 7. Bb3 Nxe4? (See Diagram), which after 8. Re1 d5 transposes into the refuted Riga variation of the Open Defense. In both cases Zarkov would often end up a piece down in the first dozen moves. I was able to beat it in speed chess only when it played these two lines. Overall though I liked Zarkov's opening repertoire, especially its variety. In addition to standard lines such as the Sicilian, Petroff, and Indian defenses, it would sometimes choose offbeat lines such as the Budapest, Owen's Defense (1. ... b6), or Black Knights' Tango (1. d4 Nf6 2. c4 Nc6). A nice extra to Zarkov is the option of a Fischer clock, which adds a specified increment of time after each move, avoiding mad 10- moves-in-5-seconds scrambles. Also Zarkov will display a running analysis for both sides if desired, however unlike ChessBase's Fritz it will not take a game and analyze it in full with just one command; one must step it through each move. MC98's main shortcoming is transpositional searches. When ordered to find matching games with varying move orders, Super Essentia is decidedly less than super when more than a few moves are involved. Short searches worked well; given the (admittedly absurd) moves 1. e4 a6 2. Bb5 Nf6 3. Nf3 e5 4. 0-0 Nc6 it quickly found many Spanish Games, but with a 13-move King's Indian line it took 90 minutes. If I wanted to end an over-long search, the "STOP" button was of no use; the only remedy was manual override by "Control-Alt-Delete" to shut down the program. Worse, the program often failed completely to find matching games in situations where a match existed. Several times I entered moves which by a slightly different order reached the same point as a game I knew already to be on the database, but no match was returned. According to MC98's documentation, the move order search is based on "a positional tree of about 20,000 recent grandmaster games." Apparently if one enters moves outside the relatively narrow confines of that game set, the search program is helpless. Thus the utility of the move order search function appears disappointingly limited. I should mention that if one already has a chess database program such as ChessBase, and wants only the game collection, the 1.1 million games on UGC III are available separately for a lower price. Its performance under such conditions is outside the scope of this review. I found MC98's online support helpful. Several e-mail queries to bookup@coil.com received prompt and helpful replies (and I did not identify myself as a reviewer of their product). Corrections to several of the above-mentioned bugs are promised in future upgrades and will be made available for free down-load from their web-site. On the whole, then, my verdict on Master Chess 98 is a transposition of the Talking Heads lyric at the top of this column: could have been better, but I had a lot of fun. Especially in view of its very low price, about one-tenth of some other databases, it gives adequate performance for the money. Those desiring something better can probably find it in ChessBase, Chess Assistant or similar products, but must also expect to pay much more (the latest ChessBase release lists at about $600!). For those seeking an all- around home-computer chess tool Master Chess 98 could be better but still represents very good value, and a considerable improvement over its predecessor.