shashki35 Vologda ru

Checker Programs Reviewed and Rated

The Largest Unbiased and Independent Checker Program Review Site on the Internet!

Please note that all programs reviewed are tested on the Windows XP operating system unless explicitly stated otherwise (there are a couple of old MS-DOS programs and a couple of Linux programs). I do not have a Mac and so I cannot test Mac programs; independent guest reviews are welcomed.

Purveyors of checker programs, please read this before sending me email about your latest creation!

• I will not advertise your program on my web site, link to your site, etc. This is an independent and unbiased review site!
  
• If it isn't 8x8 American/English checkers, I don't cover it, so don't tell me about your four-dimensional trapezoidal piece checker variant.
• If your program is a Class D or F entrant, thanks, but no thanks. As stated below, I no longer spend time reviewing bad programs.

---

Here are reviews, commentary, and ratings for a large and growing number of checker playing computer programs.

Quick advice on using the ratings: if you want the best, go with a Class A program. You might go to Class B to select entries smaller in size or with special visual appeal. You might consider Class C for certain needs, such as Linux programs. Avoid Classes D and F unless you're looking for a children's program or have a specific reason, and even then, be very selective.

There are many, many checker programs out there; every time I search the net I find new ones or ones not previously discovered. While a few are really worthy, most of them are pure junk. I review as many as I can, but I can't possibly buy and review them all, and in recent months (fall 2005 onward) I've not been trying to keep up with the all-too-numerous new releases of unworthy dreck.

Disclaimer: the ratings and comments given here are my own. Your own evaluation will vary. You must do your own analysis prior to spending money or making selections. I take no responsibility and assume no liability of any kind.

1. Scope and Definitions

This review page deliberately limits itself in a number of ways, and uses certain groupings and terminology.

• Only 8x8 American/English checkers/draughts is considered. There are many programs that play 10x10 international draughts and other variants, but I'm staying focused here. (Exception: if a program plays multiple variants, including checkers, the program will be evaluated but just for the checkers component.)

• I don't consider programs that you can play against on-line. I'm only looking at software that you can download, buy, or obtain somehow to run on your own computer, without connecting to the outside world.

• I only review programs with which I have personally spent sufficient time to get a basis for a fair and equitable evaluation. I do not go by ancedotes, claims made on web sites, or the reviews of others (with the exception of solicited guest reviews from reviewers I trust, who may have a program I cannot readily obtain; all guest reviews are so marked). This means that some world-class programs such as WCC Platinum, Colossus, Nemesis, and Wyllie are unfortunately not yet evaluated. But, I do plan to get the next release of Nemesis (if it is ever available) and review it in due course; and WCC Platinum is in my hands with a review to follow. Wyllie is no longer marketed and so unavailable to me; and the advanced version of Colossus would cost me about a week's paycheck so I don't plan to acquire it anytime in the predictable future.

I have revised the categorizations of the programs. I used to group them by a subjective measure of overall quality and seriousness, but this resulted in ever-increasing clutter. I now simply group them by playing strength, and do subevaluations of features and other factors; I also give an overall final impression in a short phrase or sentence. Here is how I now present the reviews; the class system is meant to resemble the American school grading system, where 'A' is best, 'D' is bad, and 'F' is failing (there is no 'E').

• Class A. A Class A entry is a serious, high-grade program with very high playing strength (one might say 'world class'). There are definite strength variations in this group, but these are the few and the proud. Class A programs give human grandmasters a run for their money.

• Class B. A Class B program is a strong player, and is one that is able to defeat Martin Fierz' Simple Checkers (q.v. below in the Class C listings) in a convincing enough manner, but clearly can't stand against the Class A entries. Again there is a lot of variation here in playing strength, but these are all quite good, and in some cases were once world class. Class B programs give human experts a good game, although fare less well against masters and certainly grandmasters. (As of February 2005 I have moved a substantial number of programs out of Class B and into Class C in the interests of better rating accuracy.)

• Class C. These are programs that are roughly as good as Simple Checkers. This is not at all a bad level of play, but it is not on a par with the better entries in the higher classes. Some older programs fall here as well; they were at one time in better standing but have either fallen behind or were not developed further. These give a good game to the typical checker enthusiast, but not to the expert or above.

• Class D. Class D programs lose to Simple Checkers consistently, but play along well enough to have given at least a credible performance. These programs are often non-serious 'toys' although there is an unfortunate number that pretend to be more than this. The programs in this category might play well enough against a "dub" or "scrub" but not against better competition.

• Class F. Class F programs are easily outclassed by Simple Checkers and don't even put up credible resistance. There are far too many of these; more than I can ever review and more than I can afford to purchase. With few exceptions, these are a complete waste of time and money; again, sadly, some make completely unjustified claims and pretenses. Even "dubs" and "scrubs" defeat these much of the time.

I have tried to review the latest versions of each game, but in some cases I may have reviewed an older version. Please let me know if you think this is the case.

I admit that I am often sarcastic and merciless. When someone is out there ready to take my money, I expect commensurate value. I will not be gentle with those who talk big and deliver little. If someone says they have a championship level program, they need to be able to back that up.

And anyways, who am I, a poor player on a good day, to be reviewing checker programs? Well, I've worked with software for decades, played board games even longer, and I have ideas which I am not reluctant to express. You are completely at liberty to ignore my opinions! However, if you feel an injustice has been done, please contact me. Those who have done so have found me to be very willing, and in fact anxious, to correct errors.

2. Rating Parameters. I rate programs according to several criteria. Again, I use the American school grading system, where A is the best grade and F is the worst (failing, in fact). The criteria are these:

• Playing strength: how strong a game the program plays against both human and computer opponents, with competition against Simple Checkers. often the deciding factor. (In both the top and bottom of these ratings, making distinctions is difficult. I don't try to decide which of two A programs is stronger.)

• The program's feature set: does it have a lot of very useful features, such as game or diagram export? Does it have essentials, such as move or game review? Will it do analysis?

• Interface: is the program easy to work with? Less importantly, does it have an attractive appearance?

• Overall: Combining all tangible and intangible factors, how does the program stack up? This is done not as a letter grade, but with a few words or a short phrase.

After some thought I decided not to separately rate program cost. I factor cost into the overall rating, and I comment on cost, but I go no further.

I also comment on opening and endgame play: does the program have an opening book or an endgame database, and of what type and size?

I do include screen shots for nearly all programs for which the review is complete. These screen shots are on separate pages so that this page won't have an enormously long loading time. I did prefer larger graphics to preserve detail and give a more accurate impression of the game's apprearance.

3. The Programs

CLASS A: WORLD CLASS PROGRAMS

KingsRow (updated 30 January 2005)

Now at version 1.15a, KingsRow runs as a game engine with the CheckerBoard interface. The competition version is apparently ever so slightly weaker than Cake Manchester (based on a recent match won narrowly by Cake Manchester); the download versions of KingsRow, however, may actually be stronger than the download version of Cake Manchester. (We are on a shaky footing here; as updated versions are made available for download, the situation will surely change. There is no doubt, though, that all versions of KingsRow and Cake Manchester are very strong indeed.)

KingsRow uses a very large calculated opening book of over 800,000 positions, and will handle the Chinook 8 piece endgame database and the Martin Fierz 8 piece database, making this engine formidable indeed. The author has recently calculated a 10 piece database which is something like 200GB, obviously not practical for download or even for usage by 99.9% of computer users (nor is it currently available to the public).

I feel pretty good about having gotten a draw against it a couple of times; let's not discuss how many hundreds of games I've lost.

It can be scary and formidable, but it's a fabulous playing partner for really intense and serious training (as I discuss in my companion article). It's also quite helpful in doing analysis of games you've played elsewhere.

It's freeware, too. You can't beat that.

Please see CheckerBoard for a screen shot.

Feature Set: A (based on CheckerBoard)

Interface: A (based on CheckerBoard)

Opening Book: Enormous.

Endgame Database: Chinook 8 piece and others.

Overall: Superhuman

Cake Manchester (updated 30 January 2005)

Nearly all of the comments made about KingsRow apply to Cake Manchester (previous versions were Cake Sans Souci, Cake++, and others). The opening book supplied with the download version is calculated, and is just shy of 100,000 positions. There is an option to calculate the 6-piece database, which you should do at once, and then you can download a small (4 vs 4) subset of an 8-piece database. Cake Manchester does not use the 8 piece Chinook database.

The author has separately computed an opening database with about 1,500,000 positions, and his own highly-efficient 8 piece endgame database. He is looking for a host to make these available for general downloading – any takers out there?

This is another world-class engine, and based on the competition version, at the moment likely the world's best, if by a narrow margin. I'm delighted that I was able to get about two draws in about a bazillion games. It too is freeware.

Please see CheckerBoard for a screen shot.

Interesting note: Martin Fierz was kind enough to write and tell me of the origin of the name 'Sans Souci' (a previous version) which of course means, 'without worry.' But the name comes from an apartment building at the Diamond Head end of Waikiki where Martin resided while attending grad school at UH Manoa. Certainly a beautiful and inspirational part of the world.

Feature Set: A (based on CheckerBoard)

Interface: A (based on CheckerBoard)

Opening Database: Large (in the distribution version; there is a much larger one which Martin plans to release.)

Endgame Database: 6-piece (in the distribution version; 8-piece database may become available, see above).

Overall: Superhuman.

CheckerBoard (updated 13 November 2004)

CheckerBoard is an interface, not a complete game-playing program. With it you use any of several engines. Cake Manchester, a world class competitor, comes with CheckerBoard; but you can also use KingsRow, another world class engine, or any of several others of lesser strength. I have just upgraded to version 1.611 and CheckerBoard's feature set continues to grow (and there is an even more recent release just out that improves the database search capabilities).

CheckerBoard has one of the most plain displays of the lot. The pieces are just colored circles on a square grid (the earlier slightly-off-square problem is fixed). Moving consists of clicking on a piece, to highlight it, and clicking on the destination square (you don't drag and drop, and in the case of multiple captures, you click on the very last square, which can be a bit disconcerting at first).

But all of this should not fool you; the plain exterior conceals a host of unexcelled features. Some of them are a bit concealed, in fact, but if you experiment enough and actually read all of the documentation (what a concept!) you'll find things like:

• Complete game analysis. Drop in a game and turn the engine loose and it will show you all possible moves in ranked order with a score for each, to compare with the move actually made. (In the latest versions, the analysis output is shown on a browser page which is well arranged, and clear and easy to understand. Tip: be sure to set 'display all scores' in the engine options, so you get the full list of moves and values.)

• PDN game export to the clipboard and game import from the clipboard.

• FEN (position notation) import and export to and from the clipboard.

• Graphical export of the game board.

• Get this: export to HTML to put on your website, with Javascript to allow interactive playback of a game or position! No other interface has anything remotely as good as this.

• Complete, flexible search and recall in extremely large databases of games in PDN format (thousands available on the internet).

There is a lot more in the way of options, features, and play aids. CheckerBoard may look plain but it's as good as you'll find anywhere at any price--- and it's free.

The only interface feature that I would like to see added is a separate window or panel showing the game moves. As it stands, you can get that in the analysis, HTML export, or clipboard, but not directly along with the game. There are also now some new drop-in graphic sets to alleviate the 'plain' look.

A screenshot is here (as of version 1.601).

Feature Set: A

Interface: A

Opening Book: Depends on engine chosen; will handle enormous sizes.

Endgame Database: Depends on engine chosen; will handle 8-piece databases.

Overall: Superb (based on features and interface only)

Nemesis 2.0 review projected

Review to come when version 2 is released and I buy it, as planned. By every account Nemesis is world class in every way.

CLASS B: EXCELLENT PROGRAMS

Sage

NOTE: Sage is no longer available.

Sage 9.0 is a very nice shareware program with good playing strength, an attractive presentation, and numerous features. The 'full' version is only $19. The policy on upgrades is not clear but the author has been more than willing to send them to me without additional cost.

As one of the strongest programs in Class B, Sage plays a very strong game indeed. A feature that you might or might not consider an advantage is that the Sage opening book appears to be based on published play. (I say 'appears' because the documentation is not clear about this. But it looks to be this way.) I personally find this useful; I can compare my moves with published play later on and see howthey may differ. Sage will use the Chinook 6-piece end game data base as well (although it seems up to you to be sure it's installed).

Sage shows 3D images of pieces on a plain square board. The colors are far from official, but you can make them anything you wish. Don't. I got official green and buff colors set up only to find that the pieces can't be changed, and they don't go well with an official board. So, live with the off-color colors.

Sage displays numerous windows: the opening book (very nice), the lines of analysis, the game record, and the board itself.

Piece movement is a very appealing drag and drop, although if a piece has only one legal move, the piece runs out from under the mouse and makes the legal move, an effect that can be disconcerting.

The feature set is very good. There is the usual import and export of games in PDN format, and there are some very nice database features. The program comes with a large database of games, and you can add more from various sources on the internet. The database search feature is especially well done; you can search by player, by date, and notably, you can search for positions similar to one you're interested in. The database features seem to be among the most complete and extensive of any program I've reviewed here. You can also look at positions and groups of positions, although this is a feature I don't fully understand (and it's not in the shareware version).

Analysis is a little non-intuitive (it took me a bit of time to figure it all out) but you can drop in games and get an analysis of the moves; this is a useful feature.

Sage, in my subjective judgement, seems to have a less 'intense' overall feel than KingsRow or Cake. (Sage allows substitution of engines, though, at the loss of some features. Cake will run with Sage, for instance.) This does not mean that Sage is not a good playing partner; it is an excellent one (I've managed two draws with Sage in quite a number of games). In my own practice play, I often alternate between Sage and KingsRow. Sometimes I just want that 'softer' feel and a look at Sage's opening book. If you want a shareware program, this is the best $19 you will ever spend.

Here is a screenshot.

Feature Set: A

Interface: A

Openings Database: Medium (at least).

Endgame Database: Can use Chinook 6-piece.

Overall: Excellent value and probably top of Class B in strength; was at one time world-class.

Nexus 99 (full version) Review in progress.� NO LONGER AVAILABLE.

A strong program with excellent features, including identification by name of the opening being played. This is the predecessor of Nemesis; you could think of it now as the little brother, but it is a real competitor. It is available as shareware on a trial basis and in a full version, on CD, for $25, which includes large ending databases. If you don't want to go all the way and buy Nemesis, certainly this is an inexpensive and most viable option.

The program is loaded with play features, from good opening databases (which appear to be published play rather than calculated, but I don't know this for sure) to game analysis, and fine-grained control over all important game settings.

A screen shot is here

Feature Set: A

Interface: A

Opening Book: Medium (at least)

Endgame Database: 6 piece

Overall: Excellent value in a program that was once "world class" and shows it in every way.

WCC Gold Plus

WCC Gold Plus is a very strong program with reasonable features and a unique end-game database. They call it the 'perfect play' database and I do not pretend to understand the details, except that the authors have written papers to show why it is better than the Chinook databases, and they may be right. (Ed Trice was kind enough to send me this very clear explanation.)

The Gold Plus freeware version comes with a six piece database; the full Platinum version has a seven piece database. (WCC comes in several versions, with Gold Plus being the entry level and Platinum being the top end.)

The display is very nice and unique among high-end programs; it is a 3D perspective of a marble board (much like the faux marble one on my coffee table), and pieces that remind you of the stacking plastic ones that you buy in toy stores (but these at least are not red and black). There are a few different display options, but no way to get a top-down view.

There are two main features lacking in the Gold Plus version. First, and least important, game import and export is pretty limited. Second, and quite important, is that there is no analysis mode as such. There is autoplay, but no facility for dropping in and analyzing a game you've played yourself.

Platinum is no longer marketed and in the not too distant future may be available for free, or nearly free, distribution. (It is available for free download on the Web at this time.)

Here is a screenshot.

Feature Set: B

Interface: A

Opening Book: Appears medium to large.

Endgame Database: 6 piece "perfect play"

Overall: An excellent entry in all respects.

Blitz

NOTE: Blitz is no longer available.

Blitz is an older DOS game that was a predecessor to Sage. In its time it was a 'very serious' program and is downgraded only due to extreme age. I list it here because I actually use it, on a couple of my ancient machines, like my Honolulu computer. It has the advantage of being small (about 30k!) and very portable; I sometimes run it from a bootable floppy disk (see my web page on bootables for an explanation of this concept).

Blitz plays a rather strong game. It is certainly stronger than Simple Checkers. It has a small built-in opening book which is certainly adequate for a few moves or so, and there is an undocumented option for adding an external opening book, which I have yet to figure out. Like many older, smaller programs, it plays a good tactical game. It won't miss, or allow, shots and combinations. It has a bit less of an understanding of positional play, of course, but it does well even in that regard.

The author has now released the full version as freeware, making it available to all; this is a very generous move on his part and he deserves our thanks and support. There are a few nice features: you can play back games, and scroll back and forth within a game; you can save and restore games; and you can set up positions for study or for computer autoplay.

To move pieces, you drag and drop with the mouse, or use the keyboard to move around a giant cursor, and select pieces and destinations (use the mouse, trust me). When you select a piece, legal moves for that piece are shown in a very irritating manner, with the legal destination squares flashing on and off, and if the speaker is on, an annoying continual beep.

The board is plain and done in odd colors, but it's practical enough. And you can run in CGA mode if you wish, making this program compatible with the oldest of DOS computers.

As I said, I use Blitz at times. I run it from a floppy in environments where I don't want to install software to the hard drive. I also run it on some ancient hardware just for fun or when that's all that is available. And an important feature is that it runs perfectly under Linux using the DOS emulator.

It is not my playing partner of choice when I'm able to play Sage or KingsRow, and I even won a game against it once, but it is still an interesting piece of history that plays well and credibly. It is very probably best-in-class for DOS programs (and I plan a 'playoff' to find this out definitively).

I managed to get this screenshot, which has a small white artifact which I invite you to ignore.

Feature Set: C

Interface: B

Opening Book: Small

Endgame Database: None

Overall: Very good for what it is and plays a strong game.

Cake++ review forthcoming

The only real reason to review Cake++ (having been long superceded by the rest of the Cake series) is because it is an engine option for Xcheckers and Capers. This is likely to be a strong Class B entrant. Author Martin Fierz, by the way, has an update in progress (February 2005) supplying Cake++ with a very large opening book, which will make this a formidible Linux engine.

Capers Review to come

Capers is a relatively new engine interface for Linux systems running the Gnome desktop. It runs both Simple Checkers and Cake++ and represents a substantial improvement over the XCheckers interface, though if you are not running the Gnome desktop it won't be an option for you.

GUI Checkers

Some years ago, Jon Kreuzer started work on a checker playing program as an experiment in developing game-tree searching code. The experiment was refined over time, and when a graphical-user interface (GUI) was added, Jon gave his program a logical name: GUI Checkers.

The nascent GUI Checkers got the attention of world-class checker programmer Martin Fierz (creator of CheckerBoard and the Cake series of engines), and Jon was inspired to invest additional effort to improve GUI's playing strength and features.

We put GUI through its paces here, first running our standard test against Simple Checkers. GUI won easily, and we wondered just how far GUI might go. We next tested against Marujito 1.08a, which itself can defeat Simple Checkers. GUI won that encounter as well. So we kept pushing. We then ran GUI against the solid Class B engine Damas 99. The result was an interesting draw (click here for an animation), played without error or missed opportunity on either side.

But we had a hunch, and so we played GUI against Nexus, the strong forerunner of the world-class Nemesis program. Amazingly, GUI won the encounter! Of course, one game doesn't tell the whole story, but the win was most impressive, and you can click here to see an animation.

GUI Checkers 1.00 is thus established as an excellent Class B entry. It doesn't beat the Class A programs, but it certainly plays a very worthy game. GUI features a small opening book and a small endgame database, a really nice look and feel, serious-minded features such as FEN and PDN import and export, and basic features such as move review and the like. GUI has a small opening book and a small 2x2 endgame database.

A large screen shot is here.

Feature Set: A

Interface: A

Opening Book: Small

Endgame Database: Small

Overall: A top notch entry in Class B.

Damas 99

The author calls his program 'very strong' and that is somewhat understated. While it isn't world class, it certainly plays very well indeed. In my standard test run, it handily defeats Simple Checkers, showing that the authors claim of strength is in fact substantiated. (It draws against the strong Class B program GUI Checkers.) This is a very compact program packed with playing ability and everyone should get a copy. Of course, the weaknesses are evident: there are no opening or endgame databases, and the features are basic (save, restore, set-up, undo moves). There is no move list, analysis, etc., but then again, this is a small and unpretentious item that just happens to play a fine game of checkers. It is no longer in development and one wonders what it could have become had it been pursued further.

Feature Set: C

Interface: B

Opening Book: None

Endgame Database: None

Overall: A nice entry and worth playing for variety and more than sufficient challenge for most of us.

Interface: C

Opening Book: None

Endgame Database: None

Overall: A great choice for quality casual play.

CLASS C: ACCEPTABLE STRENGTH

Wincheck

Wincheck is an interesting sort of a thing. It is part of a series, a rather large series, of similar looking checkers and draughts games done by a Franco-Spanish programming team. They make the claim that these are the most graphically and visually spectacular of all checker programs, and there is some truth to that claim, along with some problems.

Wincheck comes in 3 versions: shareware, registered, and 'analysis.' The analysis version is not cheap; it's about $50 US. And, if you want more than one of the progams in the series, you'll pay another $50 for each and every one. The whole series would be about $1,000 US or so. (There is international, Canadian, Brasilian, Sri Lankan, you name it -- every rule set I've ever heard of, and more, is available.)

The authors claim that 8x8 American/English checkers is not their game, and that the strength of their program could be higher. Perhaps that is true, but the program plays a reasonable Class C game notwithstanding. The opening book is adequate (I imagine it is based on limited published play; it has that feel), the game's tactics are strong, the move searches are good, and the positional play is fair enough. This isn't Nexus or Sage, but it isn't third-rate.

Features are severely crippled in the shareware version (no move review, limited play levels, limited choice of graphics). The registered version gets you move review, more play levels and more graphics. But to use the full potential of the program you need the $50 analysis version.

Like CheckerBoard, the analysis version lets you 'drop in' a game and have the program suggest alternative moves based on its analysis. The actual move made in the game is also scored so you can get a relative idea of how the actual and recommended moves stack up in quality, a very valuable feature. However, I would am not sure I would trust the analysis capabilities of a Class C program; you're better off with a Class A alternative.

You can export and import games and positions, but I didn't see a feature for graphic exports, nor did I see anything for database searches; overall this makes the feature set substantially weaker than CheckerBoard.

Piece movement is mouse drag-and-drop. The program has a very nice and unobtrusive way of highlighting the set of legal moves once you've picked up a piece. This isn't really necessary in 8x8 American/English checkers, but I'm sure it's a holdover from the other games in this series.

Now, as to those graphics: the board is shown in a very realistic perspective view, with 3D pieces. The overall effect is very good indeed, and would be fabulous if the colors weren't terrible. The dark pieces tend to blend in too much with the dark squares, and the red and white pieces are only available in the registered version. Also, the default graphics set (and language) doesn't seem to be saved between program runs. I've found myself in the French interface a number of times because I was too lazy to reselect English. (The French interface is actually best because some of the English translations aren't so great.)

I do play this one for variety at times. I win once in a while, and draw a little.

In my now-standard test against Simple Checkers, Wincheck draws, but seems in control most of the time, so I'd give Wincheck the slight edge.

You can view a screenshot here.

Feature Set: B

Interface: A

Opening Book: Small (as near as I can tell).

Endgame Database: None.

Overall: Visually stunning and plays a satisfactory game.

Maverick Checkers

(Guest review contributed by Peter Billings)

Maverick Checkers is an interesting game. It can be both played using a web browser or as a regular standalone application. This game can also be played on several operating systems. It plays a nice game of checkers. I'm not the best of players but I do pretty well in various online checkers sites. I have yet to defeat Maverick Checkers on the higher levels which is a bit frustrating a times but atleast you can change the difficulty levels. You can play either by time or ply. Although Maverick Checkers will easily defeat the average player, I doubt it will do the same for those at the grandmaster level.

Of all the checkers programs I've played, Maverick checkers has some of the best features. It has all the basic features such as loading, saving, and setting up games. It also auto plays games in PDN format, which is nice. Sometimes I watch my previous games without having to continuously click the next button. The game also has a feature to export to HTML similar to that of CheckerBoard but even better! It also contains analyzing capabilities but I won't put much trust in it. If you want to analyze your games, I would go with Cake or another one of the stronger programs.

The game has several nice pieces and board designs to choose from. There are a lot of other features available which are listed on the game's website.

All in all, I would recommend everyone to try this game. It is Java-based, and you can play it online in your web browser.

Strength Comparison

In an attempt to create a fair comparison with some of the other programs rated here, I simply let the programs play it out. The games I chose for comparison were: 1) Damas,
2) Gui Checkers, 3) ICheckers, 4) Simple Checkers and 5) Diamond Checkers. As it turns out, the games didn't turn out to be as fair as I initially thought it would for the programs that offered only time settings. Maverick checkers was the only program that stayed within the time limit most of the time. The others substantially exceeded the time limit quite a bit.

Maverick Checkers has several available engines but I only used the default engine for the comparisons (Mav v0.70).

1. Versus Damas: Damas handily defeated Maverick Checkers in the matches. All the matches were time based.
2. Versus Gui Checkers: This match was a little more interesting that that with Damas. In the game based on 2 second think times, Maverick Checkers put up a good fight until towards the end. Gui Checkers however tended to exceed the time limit a lot. In the game based on equal level/ply, Maverick Checkers won some games and drew the rest. Gui Checkers however seems to perform better towards the end of the game which may be a result of the endgame database.
3. Versus ICheckers: This was not much of a competition at all. Maverick easily defeated ICheckers in all of its matches based on both time and ply.
4. Versus Simple Checkers: The games where close but Simple Checkers won most of them and drew the rest. Once again, Simple Checkers used more time per move that Maverick Checkers which probably affected the game outcomes. Unfortunately, I couldn't put them in a match together based on ply.
5. Diamond Checkers: Finally, I compared Maverick Checkers to Diamond Checkers. This was particularly interesting because both programs are written in Java and can be played through a web browser. Maverick Checkers defeated Diamond in most of the games and drew a few.
6. Other Checkers Java Applets: Maverick Checkers easily defeated all the other checkers applets I matched it against. As a matter of fact, it also defeated these programs.

• Shallow Yellow
• Darkfish
• Jeu de Checkers
  

Feature Set: A

Interface: A

Opening Book: None

Endgame Database: None

Overall: This is a nice program that will challenge the average player. Not as strong as some of the other Class C programs but far stronger than Class D in terms of playing
strength. This is a well done game.

Webmaster's Note: We thank Mr. Billings for his review. We are in quite close agreement with it. Maverick was a pleasant surprise, combining some well-thought-out play features with an engine that plays a reasonable, if not great, game of checkers.� Click here for a screenshot.

Deep Brew

Deep Brew appears to have started out as a very ambitious project, and changed directions somewhere along the line. The author says he is going to restart, and I certainly hope he does. Deep Brew as found (or as left) is rather a nice piece of work, and as it is free and open source, it provides a lot of value.

The author has recently (as of January 2004) released an upgrade to what he called 'The Ancient Legacy Version' for Windows. This upgrade provides some essential new features and some bug fixes and cleanups, and significantly increases the appeal of the program.

Despite having no opening book or endgame database, Deep Brew plays quite a decent game. I set it to one second think time on my 2GHz Pentium, and I have a lot of trouble keeping up with it. It is a good tactical engine; I imagine it interprets position play in terms of material, but the search tree is good enough to make this come out right. Deep Brew is a worthy opponent indeed, and I find myself drawn back to this program when I want a change from the high end superhuman machines.

The feature set is somewhat developed, even if much less comprehensive than some of the more-developed world-class programs. You can set up positions, take back moves, and work from a well-formatted move list. You can also do position analysis (although it appears to be a bit of an effort to get Deep Brew to play both sides, and I wouldn't trust the analysis all that far).

The graphics are very nice with a clean click-on-piece, click-on-square play interface. The squares are official green and buff, and numbered with discrete little numbers that are not at all intrusive or bothersome. The pieces are official red and white.

Deep Brew drew against Simple Checkers; Deepbrew was a man down but Simple Checkers allowed draw by repetition. Tsk, tsk. Deep Brew played credibly and is likely short of Simple Checkers in strength; I don't consider it strong enough to go in Class B, but it clearly doesn't belong in Class D; hence, Class C!

Given additional development work, such as an opening book and a much needed endgame database, Deep Brew stands every chance of being a contender some day. I wish the author luck and success.

You can view a screenshot here.

Feature Set: B

Interface: B

Opening Book: None

Endgame Database: None

Overall: A nice entry with a lot of potential.

XCheckers (important 25 January 2005 update)

XCheckers, like CheckerBoard, is an interface and a display system, and uses other game engines. XCheckers works with the Simple Checkers game engine (and recently also with Cake++), which is reviewed below. XCheckers has few features, but at least you can scroll back and forth among moves. The display is nice enough, with very nice 3D checker pieces. XCheckers is a Unix-only freeware product. It is also intended to be a client for an International Checkers and Draughts Server (ICDS) which, to my knowledge, no one runs any longer.

Recently XCheckers has been made to work with the Cake++ engine, an ancestor of Cake Manchester and earlier versions in the Martin Fierz Cake series (e.g. Cake Las Vegas, Cake San Souci, etc.). Cake++ (to be separately reviewed in the future) is open-source and utilizes the Chinook 4-piece endgame database. It is a rather strong program. This now gives us a very usable Unix option for checkers. In addition, a new interface is in development with many more features (a beta is available; more on this in a future review).

I finally have a screenshot.

Feature Set: D

Interface: B

Opening Book: None

Endgame Database: None with Simple Checkers; Chinook 4-piece with Cake++.

Overall: A good option for Linux systems.

Blondie24

NOTE: No longer available.

The second neural network program to come into my possession, this one is a commercial offering selling for about $20 plus shipping, and is delivered in a hefty two-CD package. There is a great deal of hype surrounding this program, and the marketing.... well, you decide. Consider that the subtitle of the package is 'Checkers With An Attitude.'

The whole thing stems from serious work in the field of artificial intelligence. The author took the early work done by IBM in applying neural network techniques (called here 'Darwinian') and extended the concept. The result is a program that the marketers claim plays at the 'certified expert' level. At least that's what they say.

Now, the marketing. You've got to give an 'A' for originality, and quite a bit of time, effort, and expense surely went into this. Why does it take two full CDs to deliver this program? Because, it turns out, you play against three animated, talking actresses of the 'California girl' variety. During the game, they keep up an on-and-off chatter, complete with postures, gestures, facial expressions, and so on. Professional models were hired for this purpose. There is a lot of video, apparently over a thousand short clips, and it comes off quite well. The poses and dialogue fit the game well, too; when I went a piece up in one game, the opponent shook her digital head and said, 'I don't think I like this!' Or, making a nondescript move, she whispered conspiratorily, 'It's part of my plan.'

The display is well done, with a choice of beautiful background photos from around the San Diego area, including the harbor, the Hotel Del Coronado, and other famous and recognizable scenes. To the game's credit, the board is green and buff with red and white pieces. They did their basic research. There is a 3D-board shown in front of the actress, and a larger 2D board to the right where you click on piece, click on square to make your moves. The two boards update simultaneously. It is very appealing and very well done. Check out this screen shot; I've made it large enough to be representative, so it may take a little time to load.

There is also quite a bit of interesting background material on how the program was developed. Some of it appears to disparage or downplay the conventional approach to programming checkers in favor of the neural net approach; this seems a bit out of line especially considering the relative playing strengths, as will be seen. Or perhaps the point is simply that Blondie24 reaches a claimed high playing level by the new approach; but that isn't quite true either. The author wrote a whole book on this: 'Blondie24: Playing At The Edge of AI.' He took it all quite seriously; in fact, he did serious work with seriousness of purpose.

So, how well do these California girls play? You can choose three opponents with three levels each: Novice, Intermediate, and Advanced; or you can go all the way and play Blondie herself at the Expert level. I tried Novice and Intermediate and there was no competition at all, so I went on to the Expert level and did my usual test. Blondie24 faced off against Simple Checkers. Simple played White and had 5 seconds per move. Blondie24 took her own time (you can't set it)and used typically 30-90 seconds per move except for forced captures.

Result? With all the time Blondie24 took, she managed a draw against Simple Checkers. Simple had a little bit better game; Blondie24 made one terrible positional play but Simple didn't exploit it well enough. At a deeper search level (more than 5 seconds) Simple might have won. In the end, when a position repeated over and over with neither engine willing to try anything else, Blondie24 refused repeated draw offers, finally telling me, 'I don't THINK so!' and then no longer responding at all.

I did a later rematch with almost identical results. Blondie made one error but Simple Checkers didn't catch it, and the game drew again through repeating positions. You can view an animated, commented version here. The comments are my own interpretation of an analysis I ran with KingsRow. Notice where Blondie makes what might be a losing move but Simple Checkers just can't look far enough to see it. But as I state at the end of the commentary, we see here two Class C programs making only a single real mistake each in the course of a game. (There are numerous moves that could be argued as a bit less sharp than an ideal move, but this is usually by the finest of margins.) The state of the art in checker programming is pretty good these days, when even programs rated way down the line play this kind of checkers.

There are no game review features, move lists, position set-ups, etc; not even saved games. This, folks, is checkers for amusement, California-style, and not for serious competition - although at the expert level Blondie24 is going to defeat most casual human players. It rates a 'C' in playing strength by virtue of a draw with Simple Checkers, and thus escapes the 'Don't Waste Your Time" award, although the packaging and marketing certainly could have put it there. But this all overstates the merits of the program as a checker engine. It's good enough, but not remotely as good as Sage or Nexus, which sell for roughly the same price. It falls short of the small freeware program Damas99 in every respect; and while playing almost even with Simple Checkers (with a huge time advantage) it offers none of the features of the free CheckerBoard interface.

Despite the comments made in the supporting material provided with the game, the neural net approach doesn't seem to make it in checkers. Top checker programmers today find that the opening and ending databases are critically important. And they've got the search algorithms and corresponding evaluation criteria very highly tuned.

As mentioned in an earlier review, neural net programming dominates backgammon. Neural net backgammon engines play at the world class level. Perhaps this is because backgammon is a non-deterministic game. But it hasn't seemed to work out for checkers.

Should you buy Blondie24, "Checkers With An Attitude"? The game is a hoot to play, but ultimately, no more than that. It's amusement, not serious checkers. And that is a shame. This started out as a serious project to apply neural net techniques to checkers. It didn't come off in the end as a true expert player, but the attempt was well-made and shows it. Turning this into marketing silliness is a real disservice to the genuine scientific work done by the program author.

Feature Set: F

Interface: A

Opening Book: None.

Endgame Database: None.

Overall: A serious effort marred by lack of features and off-the-wall marketing that turn this into just a toy.

Simple Checkers

Simple Checkers is a surprise and a treat. It is indeed "simple" but that does not speak to the astonishing quality of this very small package.

This engine is used by XCheckers, but can be used by several others as well, especially CheckerBoard. It is free, and it is supplied with source code so you can see how it works. It plays a fine tactical game; in fact, it fooled me into thinking it had a small opening book, but in fact it was just doing a nice job of calculating plausible moves.

Positional play, as is typical with engines of this type, is not the strongest point. And it has enough trouble at times winning end games. But the search tree is quite well done. Simple Checkers is certainly weaker than Sage and Blitz, but it is fun to play against. I'd say it's more like playing another human than most other checker playing engines. I use it mostly because it will run under Linux (Cake++ and Kcheckers are the only others).

I enjoy playing against this engine, but I generally don't fare well. It can be frustrating to be outplayed by a small amount of C code!

Now, a few editorial comments. Simple Checkers, which makes no claims or pretenses, will outplay anything in Classes D and F, including some that really put on airs. Martin Fierz, the author, freely publishes the source code for Simple Checkers. It's indeed simple; maybe 2,000 lines of C code, much of which is a rudimentary text interface. The actual logic for choosing moves is quite brief and very easy to understand. There are some basic rules: a cramp on the enemy is good, a king in the center is good, etc. Just plain old basic principles. And it beats the daylight out of just about every non-serious program. In fact, I benchmark other programs against Simple Checkers, as you've seen here.

For fun and instruction, here is an example game between Simple Checkers and Zillions of Games, which has a respectable generic search algorithm but no checker knowledge per se. Simple Checkers wallops Zillions, although there is some real humor in the end, when, with Simple Checkers ahead by something like 6 pieces, it decides it needs six kings before going after the lone Zillions king. (The game was played at 2 seconds per move on a 2Ghz Pentium with 512 MB memory.)

So, with the very effective Simple Checkers rule set available for all to copy and improve upon, why do so many of the 'toy' programs make wild claims about playing strength and deliver next to nothing? Copy Simple Checkers and at least play a decent game! I simply can't understand it. Is it laziness, ignorance, or just lack of respect for the customer? Even more humorous (or sad) are rule sets that are incredibly stupid. One of the programs below thinks that kings are good, and advanced ranks are good, so it will make a single king and forever shuffle it back and forth between the seventh and eighth ranks.

Simple Checkers puts most of the checker playing programs out there to shame. Literally.

Feature Set: D with XCheckers interface; A with CheckerBoard interface

Interface: B with XCheckers; A with CheckerBoard

Opening Book: None.

Endgame Database: None.

Overall: One of the best of the non-pretentious programs out there.