Top Ten Solitaire

Top Ten Solitaire -- reviewed by Michael Keller

I first played this program several years ago. The version I now have, bought through Big Fish Games using a coupon for a free game (from a very limited list), is listed as version 1.8. They seem to be the current publisher, but I don't know who originally developed the game. Surely by version 1.8 minor glitches (the kind that don't affect game play much but make the program look unprofessional) should have been eliminated, but there are a number of glitches in the current version. For example, there is a stray white pixel in the middle of almost every playing background. The program rates players on a point-scoring basis, with one of 16 titles (I think; they are not listed in help anywhere but that's how many appear in the ratings lists). All but one of the titles are capitalized, but the second-highest (200,000 points in one game or 2,000,000 total) is in lower-case: wizard. There is a pop-up stats frame on the main screen, but it is not wide enough for the longer titles, which spill outside the frame. This is just sloppy work and careless testing. Players start as Novices in each individual game and as a whole: 500 points in any game get you the title of Rookie; 5000 total points in all ten games get you overall Rookie status. Other titles are earned in incremental steps: a player scoring 70,000 points in an individual game is rated Maven in that game; if you have 700,000 total points in the ten games you have an overall rating of Maven. The highest rank which appears on the rating lists is Demi-god (apparently 350,000/3,500,000). I played this version extensively for several days early in 2008, and accumulated statistics in all of the games. When I returned in January of 2009, all of my accumulated stats had been erased.

There is an online server (accessible automatically through the game) which tracks your scores and ratings in all eleven categories (you set a user name and character PIN to keep track of your own stats). This could be sensational if the individual games didn't have so many problems. In a few hours I got up to Apprentice overall (I was at 36,650 before I got purged), and reached Authority in Accordion and Competitor in Golf. But I wasn't interested enough play much further, as the program and individual games just aren't good enough compared to the many other solitaire programs on the market. The stat screens list the top 20 players in any given category, and can be sorted by total points or points per game, and show where you rank, but the stats don't seem to be accessible online and there is no way to see any stats except the top 20 (compare this to the powerful and detailed system on NetCELL). There are numerous anomalies in the online scores; whether these are due to older versions which had different scoring systems, program bugs, cheating, or some undocumented features in the scoring system, I do not know. For example, in Penguin you start at -150 points and get +5 points for each card played to the foundations (many games are based on an archaic model sometimes called Las Vegas solitaire, similar to the payoffs at the old Canfield casino in the late 19th century). So if you win at Penguin you get 245 points for the 49 cards you play to the foundations, plus a 150 point victory bonus which erases your initial deficit. I have won all 7 deals of Penguin I have tried, but my average score is 244, not 245 (true of several other players whose stats I can see, though a few players also have 245 averages (all of them have perfect records, one player with 2315 straight wins). But three players have higher averages: the top player has 449, followed by 313 and 257. And none of them are even close to perfect; the top player has lost 22 of 11578. Those look like scores from some older version of the program. In at least one game, Canfield, it is difficult to get positive scores; I had to win a game (which took 25 tries) before my score went into the positive side.

But there are several larger problems which keep this system of competition from being as good as it could be. First of all, the main sorting method, the one for which titles are awarded, is by total points, which has more to do with how many games you've played rather than how good you are: the second-ranked player in Accordion has won less than 6 percent of the time, but has played over 6700 deals; third place has won less than 1% in over 11,000 deals. The secondary sorting method is by points per game, but the games vary widely in scoring systems. In Scorpion, for example, you start at -10 and get 1000 points if you win; there is no partial credit for anything else, even a completed suite of 13 cards. So the points per game rating in Scorpion might just as well be replaced by winning percentage. Scorpion is a hard game with a low skill factor, and giving credit only for full wins takes away most of that skill factor: a good player will more often make substantial progress, completing suites and uncovering face down cards, but no credit is given for any of that. In Golf you gain points only if you discard all 30 tableau cards; you lose points even if there is only one card left -- in choosing a version too easy to win, put giving no credit for partial success, Top Ten has eliminated a great deal of the skill in playing Golf. I managed 27 wins in 67 tries, for a 93 average (66th of 431), and 6365 points (357th of 519). Presumably the point system is intended to allow players to be given an overall rating, but the scoring systems vary so wildly that there is no good way to tell who the best players are. Since the top score in Accordion is 670 (I think), and a few strong players who have learned to deal with the inept interface have won every deal, it is possible to average 670 at Accordion. Top Ten's version of Pyramid is so hard that the best player on the server, winning over 10% of the time, is averaging only 76 points (I'm actually ranked 8th at 34 points, despite only 2 wins in 54 tries, and 11th in Canfield with 2 wins in 26). I think the best player is fontpasc1, who has a perfect average at Accordion, the second-best average at Monte Carlo 378/433 for a 253 average), and also perfect scores at Penguin and FreeCell, a near-perfect score in Yukon, and 13th-ranked in Scorpion. But fontpasc1 has not played enough to appear on the cumulative scores list (and has only modest titles) and does not even appear on the highest cumulative average list, having played enough of the harder games to push his/her average below the 345 which is currently in 20th place. I'd love to know who this person is; a search for fontpasc1 didn't turn up anything.

What really makes a community like this work, though, is the opportunity for players to communicate with each other, to talk about strategies, and to compare scores and exchange difficult deals. Look at the flourishing communities on Pretty Good Solitaire and NetCELL to see how this works. But in order to exchange difficult deals, the deals have to be numbered, and Top Ten does not do this, although it takes only a few minutes of trivial programming. Of course the reason they will give is that they don't want to corrupt the statistics (in harder games especially) by letting players play deals known to be winnable. This is admirable, and the stats for Top Ten are in various ways more reliable than those of Pretty Good Solitaire, but there is a way to have your cake and eat it too. On NetCELL, you can play a random deal, and you are only told the deal number after the deal is finished (won or conceded). You can play a deal with a specified number, but it will not count in your stats. This would work splendidly well in Top Ten, allowing players to compare notes on interesting and difficult deals, but keep the competitive stats accurate.

There are ten games in Top Ten Solitaire, certainly ten of the best-known if not the best playing games: Accordion, Canfield, FreeCell, Golf, Klondike, Monte Carlo, Penguin, Pyramid, Scorpion, Yukon. There are no rules options for any of the games. This was probably done to keep the statistics clean and simple, a good idea if you program the games well and choose the best and most skillful versions of each game. But the versions chosen are a mixed bag. In the case of Golf, they probably made it too easy (one player is winning 88%) by using the wraparound version (good) but dealing only six columns of five cards (bad) instead of the traditional seven. For Klondike and Canfield, they use the hardest and most restrictive rules with a three-at-a-time deal (though they made an interesting choice in Canfield by using an open terrace instead of a closed storehouse). Monte Carlo, Yukon, Scorpion, and FreeCell are pretty standard, though even the simplest supermoves don't work in FreeCell, and the standard version of Scorpion is so hard that a less restrictive version which rewards skill more should have been chosen. Rather than program Scorpion properly to deal with the limited screen size, the programmers put in a lazy and wrongheaded rule restricting cards to a maximum of 30 per column (according to the help; I think it's actually implemented as 29), which occasionally makes an otherwise winnable game unwinnable. There is a bug in Penguin allowing you to make illegal plays, putting a card of the last rank on the first. But the two worst implementations by far are Pyramid and Accordion. In Pyramid, rather than using an excellent system like Par Pyramid (which would have made for very interesting and skillful competition), they chose an inferior one-pass version where the stock is dealt into three separate waste piles. This makes the game heavily reliant on luck: there is little chance of winning unless you can eliminate a lot of cards quickly before the waste piles start to accumulate. In Accordion, the display is limited to 41 cards, so it is not possible to play the skillful open version in the normal way. You need to scan through and figure out which rank is missing the most cards, then make enough plays to eliminate 11 cards so that the rest of the stock can be dealt. Only four ranked players (with 25 deals played) have learned to do this well enough to win even 50% of the time (before the purge I won 20 of 25, not enough to be ranked yet with 13535 points).

All of the games have unlimited undo, which actually does corrupt the game somewhat for closed games, especially in Klondike and Canfield where you can manipulate the stock rather badly. There is no active autoplay for any of the games, though the foundation games will autofinish once the cards in the tableau columns are in order. There is an autoend feature, intended to tell you when you are blocked and need either to concede defeat or undo to an earlier spot in the game, but it doesn't always seem to work properly. In Scorpion, it often tells you that you are blocked when you have kings that can move to empty columns or you still have the three-card merci left to deal. I had to turn it off anyway because it is part of a very annoying and obtrusive hints and warnings system.