Meet MAYA-II, the new DNA computer that can play Tic-Tac-Toe

While strands of DNA may be able to pick fights with other strands, one thing that they can't do yet is play computer games -- that is, until now. In what appears to be an early proof-of-concept for DNA computing, scientists at Columbia University and the University of New Mexico have created a basic computer, called the MAYA-II, which has a molecular array of YES and AND logic gates made up of 100 DNA circuits. This allows the MAYA-II to play a complete game of Tic-Tac-Toe against a human opponent, and apparently nearly always win. However, this isn't exactly a fair test, given that the MAYA-II always goes first, and always opens with the center square. Still, even though each move takes 30 minutes to complete, the lead researcher on the project, Dr. Joanna Macdonald of Columbia, says that a DNA computer would be perfect for injection into human subjects to combat disease, because in theory, it would be able to "diagnose and kill cancerous cells or monitor and treat diabetes by dispensing insulin when needed." Useful applications aside, we're ready to take on the MAYA-II in Connect Four whenever Dr. Macdonald thinks it's ready.

[Via MedGadget]

[Bob @ Oct 16th 2006 9:02PM]

Doom port?

[mike @ Oct 16th 2006 9:12PM]

I for one welcome our DNA-computing blood dwelling cell-killing overlords.

[Ryan Gardner @ Oct 16th 2006 9:54PM]

if it "nearly always wins" it's playing against idiots. It should "nearly always get a tie" - which is what happens when two people with an IQ above 30 play that game.

[PhilJ @ Oct 16th 2006 10:45PM]

Mike, damn good, and way funnier than what I was thinking. Bob, great comment. You both made me laugh pretty good tonight, thanks!

[Ferny @ Oct 17th 2006 1:23AM]

actually to guarantee a win, the first player ISN'T suppost to take the center. the second player is suppost to take the center. and it SHOULD result in a tie.

[Alex @ Oct 17th 2006 1:27AM]

Ryan, the "nearly always wins" refer to all possible games the automaton can play given a fixed strategy. This means that the automaton have to respond to all possible human moves, and obviously most of the games lead to a victory for the automaton, with a few resulting in a tie. Your comment is completely irrelevant to the design of a game-playing automaton.

[Dimitri @ Oct 17th 2006 2:07AM]

Alex,
First, it's "automaton has", not "automaton have". Second, Tic-tac-toe should always ends in a *tie* (if both players play correctly.) Your comment to Ryan that "obviously most of the games lead to a victory for the automaton, with a few resulting in a tie" is just plain false. Sorry dude.

[Chir @ Oct 17th 2006 1:30AM]

Ferny, I can assure you the strategy the automaton uses is perfectly valid. It guarantees either a tie or a win. Your statement is self-contradictory, since resulting in a tie defeats the guarantee of a win. In any case, no strategy guarantees wins in all cases. The perfect strategy used here guarantees wins in most cases, and ties in others.

[Nate @ Oct 17th 2006 3:56AM]

No you guys have it all wrong. Look at the photo. There are more than 9 spaces to fill. In molecullar tic-tac-toe, you are allowed to play past the traditional 3x3 grid.

[Kibi @ Oct 17th 2006 6:44AM]

It could "combat disease"? Hell, if it can play games it could combat death!
Remember "Bill and Ted"?
Grim Reaper: A hit. You have sunk my battleship!
Dead Bill, Dead Ted: Excellent! Yes!
Dead Ted: I totally knew he would put it in the J's, dude!
Dead Bill: Good thinking, Ted.
Grim Reaper: You must play me again.
Dead Bill: WHAT?
Grim Reaper: Um, best two out of three.
Dead Bill, Dead Ted: No way!
Grim Reaper: Yes way.

[Josh Chudnovsky @ Oct 17th 2006 8:15AM]

Kibi,

my friend that is what comedy is all about right there... take a completely forgotten quote, lost to the ocean of time and bad movie producing, and bring it back into the light 15 years later. Kudos.... Kudos....

And yes, Tic Tac Toe should *always* end in a freaking tie... and yes it definately looks like the grid is enourmous, in which case against any human it'll kick its butt because we don't exactly practice tic-tac-toe outside the 3x3 range. If after the age of 8 you somehow found a strategy for beating your buddy at this game, then your buddy is probably in special ed.

[Adam @ Oct 17th 2006 12:38PM]

Thats a 384 well plate to be exact. The purple box represents lower DNA concentration, the red is higher (if they are using OD260 readings, which it looks like they are). This is a fun play on an everyday biotech procedure, but can it really be useful? I can make it print a smiley face, woohoo.

[Chir @ Oct 17th 2006 3:01PM]

Dimitri, you are making the same mistake about the design of the automaton as some of the previous posters. The statement that the game should always end in a tie if played CORRECTLY is false. The thing to keep in mind here is what it means to play correctly and what the automaton has to do in response. The game of tic-tac-toe is a game of perfect information, which means if both players employ a perfect strategy for play, then the game will always end in a tie. The automaton does use a perfect strategy, but it cannot assume the human player does as well. The automaton must account for all possible scenarios that the rules of the game allow, not just those that will take place if the human player plays perfectly. This is the reason why under the majority of the scenarios, the automaton will win the game. These are the scenarios under which the human player does not follow the perfect strategy, but are nonetheless correctly playing the game as specified under the rules of the game. Under all the other scenarios (four, to be exact, since the automaton always plays first and always in the center square) the human player follows the perfect strategy and the game ends in a tie.

Nate and Josh: the automaton does play a regular game of tic-tac-toe. What you are seeing is a 384 well plate that is used to read the outcome of the experiments. In any game, only 9 of those wells are used, representing the 9 squares of a tic-tac-toe game.

Adam, those readings are actually not OD readings, they are fluorescent readings as the procedure makes use of fluorescent resonance energy transfer. The detection part of the experiment, the "everyday biotech procedure" part as you say, is not important. Printing a smiley face either using OD readings or FRET readings is a manual procedure that anyone can do. The value of this research lies in actually producing an automaton that can produce these outputs through the logical analysis of human inputs by following a set of programmed rules, in this case that of the game of tic-tac-toe.

[David @ Oct 17th 2006 3:14PM]

Greetings Professor Falken. Would you like to play a game?

[__redruM @ Oct 18th 2006 11:47AM]

Chir,
I see your point about all possible games, but that's not how the article is worded. We shouldn't have to be an automoton expert to decode the article. A normal reading of the article is misleading at the very least, and false to most interpretations. You have to look at all the possible interpretations of the article, to see why it's stupid.

[southcutt @ Dec 9th 2006 1:35PM]

David: Kudos. Far better quote than Bill & Ted.