DNA^vs CPU is a free iPhone/iPad game of strategy, where you pit your skill in planning ahead against my internal and learn-on-the-fly algorithms.  The game attempts to employ some simple Artificial Intelligence (AI) techniques to improve play, so its performance may improve over time.  However, learning is a bumpy road, so occasional questionable early-game moves seem likely.

• DNA^vs CPU plays at about a Fifth-Grade level. That is, its machine learning techniques are "Pre-Perceptron" and extremely primitive (no neural nets, no transformers, no massive databases, etc.), but it is still a formidable opponent. Your overall goal is to see if you (DNA) can win enough games to get to the Easter Egg before CPU makes the games noticeably harder^† to win. That is – Are you smarter than a Fifth Grader? ^‡


• DNA^vs CPU is easy to play, but hard to play fast.  The scoring system strongly rewards fast play and mercilessly punishes slow play.
• DNA^vs CPU is comprised of four Generations.  Each Generation includes four Levels (A,C,G,T) with a total of 17 different games ‐ A: Astronomic (7 games),   C: Cosmic (5 games),   G: Galactic (3 games), and  T: Tumultuous (2 games).   At each Generation, players start at Level A: Astronomic.  Each Level and game is progressively harder and worth more.  A win at Game X earns access to Game X+1.
• When you win all 17 games in a Generation, you may ascend to the next Generation.  Although the 17 games appear similar in each Generation, rule changes and CPU performance upgrades make each successive Generation noticeably more challenging.  Winning all 17 games at the highest Generation may reveal an Easter Egg.
• DNA^vs CPU is played on a rectangular grid.  It's based on the 5^×4 game tac-tickle developed by Harry D. Ruderman in 1967.  DNA^vs CPU has added a GUI, machine play (augmented with some simple machine learning), a variety of game sizes, undo, some quantum effects, instant replay, a video archive of your previous games, and user-viewable stats, summaries, and research reports.  The game is locked into Landscape Mode.
• You earn points with a win: faster play is rewarded with more points.  You may lose points with a defeat.  Your cumulative points are displayed.  As your points add up, you may be promoted (there are 21 levels).  After your first win at your current largest (hardest) game size, a new (harder) game is introduced from which you may choose.  Harder game wins are worth more points.  As harder games are introduced, the points for easier game wins are reduced.
• Goals: Build your points by winning increasingly difficult games;  Move through the Generations and up the Promotion Ladder (21 levels);  Earn your way to T⋮10, the hardest, most rewarding game.  Be careful at T⋮10, because some of your learned game behaviors may not work as you think. When you win a GenZ(T⋮10) game, the Easter Egg may become available.
• A variety of actions and reports are available for your use during play or between games (e.g., Undo, Instant Replay, Statistics, Opener/Responder Winning Efficiency, ...).  Note however that employing some of them during play (e.g., Undo, Sleep) drastically reduces the maximum points awarded for a win.
• My "learning" ability is heavily dependent on your style of play and performance, and it will vary from player to player.  An open question is whether your inexperienced play in the early games causes me to learn bad habits (thus making it easier for you to win more often in the long term).  Learning is a bumpy road.  You will likely see me commit some stupendous blunders as I try to learn.
• Beginners can gain some useful insights from the video replays of previous games, and from the [Strategy], [Puzzlers], and [Blunders] reports.
• Experienced players may wish to use Tournament Mode, which provides a mechanism for you to periodically gauge your skill level, or more likely to compete against other players (at the same time in the same real or virtual room).  Possible goals may be how long til 1,500 Λxons, 100 wins, Level C or G or T, win one of each of the 17 games, or average Λxons per game (for 100 games), minimum think-times per game (for 100 games), ...
• To Prevail (not easy!):
  1. Work your way up the game ladder (3 Generations with 4 levels each for a total of 51 games) to meet and defeat the 4th Generation (Easter Egg Challenge).
  2. Play some practice tournaments on your own to get comfortable playing under pressure. Then play in tournaments with other players to gauge your mastery.
• Internet access is not required to play (DNA^vs CPU is completely self-contained and does not communicate with the outside world – except when Apple verifies upgrade purchases).
• No data collection (for use outside the app) is performed.
• To contact us,  Email:  DNAvsCPU@DNAvsCPU.com


• Are you smarter than a Fifth Grader? ^‡

^† This may be (purposely?) misleading. The advanced games (Levels {C,G,T}) are already pretty hard, and what CPU learns from playing them may make them even a little harder.

^‡ Actually, real Fifth Graders play DNAvsCPU about as well as most other groups. So, the real question is: In the context of DNAvsCPU, how quickly can you learn to exploit your innate superior pattern-matching and look-ahead skills against an inanimate machine that can

• Evaluate millions of moves per second
• Review all previous games in less than a millisecond, and learn something from many of them
• But never looks more than two moves ahead

The information on this website is a small subset of the internal documentation in the App.  Click the App Store icon to visit the iTunes Store.

How to Play

• The objective is to get 3 of your tokens in a row: up/down, left/right, or diagonally – like tic-tac-toe (see the "Screenshots" section below for sample board layouts).
• A player may move one of their tokens one cell up, down, left or right (but not diagonally) to an open cell.  You must move on each turn.  To move, tap one of your tokens, then tap the destination cell.  For the harder games (Levels C,G,T), quantum tunneling and wormholes may come into play and more aggressive moves are available.

Screenshots

Below is an array of boards (each at mid-game) and some other random screenshots.  These boards are static (not "live").

1. [Upper Left] DNA (Green) to move and win in 2.  How?  What do you think of CPU's last move?  What would you have done?
2. [Upper Right] Blue to move and has a sure win in 2.  How?   If instead it is Green's turn, Green also has a forced win in 2.  How?
3. The next three boards have that same attribute (the player who is about to move has a forced win).  How?  Note that this condition is not always the case, but it does suggest that the first player to move may have an advantage.
4. A peek at some of the reports.  [Strategy], [Blunders], and [Puzzlers] might be interesting to beginners, or at least worth a chuckle.
5. A brief player History