Ideal Tetris Theory

Yahweh · 9th November 2005, 08:32 PM

I wanted to totally beat Ceptimus' highscore in Tetris at the Freethought Forum, and I found the scoring formula of Tetris on Wikipedia:

Quote:

The scoring formula for the majority of implementations of Tetris is built on the belief that more difficult line clears should be awarded more points. In Nintendo's implementations on the NES, Game Boy, and SNES, the four possible line clears are as follows:

1. Single = (level+1)*40 one line is cleared.
2. Double = (level+1)*100 two lines are simultaneously cleared.
3. Triple = (level+1)*300 three lines are simultaneously cleared.
4. Tetris = (level+1)*1200 four lines are simultaneously cleared.

Code:

Level 	00 	01 	02 	03 	04 	05 	06 	07 	08 	09 	10
Single 	40 	80 	120 	160 	200 	240 	280 	320 	360 	400 	440
Double 	100 	200 	300 	400 	500 	600 	700 	800 	900 	1000 	1100
Triple 	300 	600 	900 	1200 	1500 	1800 	2100 	2400 	2700 	3000 	3300
Tetris 	1200 	2400 	3600 	4800 	6000 	7200 	8400 	9600 	10800 	12000 	13200

I'm a pretty good Tetris player, my technique has always been placing my pieces to get as many lines as possible (I get a lot of single lines), but I think there might be a better way. In a game of Tetris, there ought to be a way to maximize points by getting single-lines some percentage of the time, doubles some other percentage of the time, and so on.

I've been looking, but I havent come across anything on the subject. Is there a serious ideal technique to play Tetris?

Matabiri · 10th November 2005, 03:27 AM

Like this?

http://www.lookatentertainment.com/v/v-53.htm

PatKelley · 10th November 2005, 04:12 AM

My dad had a technique: he would use a "slowdown.exe" program to slow it to nil, and give him as much time as he needed to contemplate his next move.

ceptimus · 10th November 2005, 04:37 AM

Originally Posted by Yahweh

I'm a pretty good Tetris player

Ha! Check out my highscores in the other two Tetris games at the freethought forum, as well as the one at mu.nu I hold all the Tetris trophies there.

On each of those games, you get to see the next piece as well as the current one. You have to consider what the playing area will be like after BOTH pieces have landed to play optimally.

I try to leave a channel down one side to drop the long pieces in (so as to get the 4-line bonuses) but once the pile gets more than 6 squares high, I abandon that strategy and try to complete as many rows as possible, so as to get the pile height back down to something manageable.

You don't have time to think on the later levels - you just have to react - it takes lots of practice. Don't change your mind once you've decided where to place a piece - changing your mind while the piece is falling usually ends in a disastrous mistake.

Edit to add: It was slimshady2357 who taught me the 'get the tetrises' technique. Before that, I used to go for just complete lines and not bother with the bonuses.

LW · 10th November 2005, 04:44 AM

Originally Posted by Yahweh

I've been looking, but I havent come across anything on the subject. Is there a serious ideal technique to play Tetris?

A version of Tetris where the player knows all the pieces in advance has been proved to be NP-complete, meaning that it is a pretty difficult problem. Optimized versions of NP-complete programs tend to be even more difficult ranging from Delta^P₂ to Sigma^P₂ depending on the definition of optimality. Without doing any formal analysis at all, my educated guess would be that Delta^P₂ would be the correct class.

Analyzing the real version where the player doesn't know the sequence of the blocks is more difficult and to my knowledge no one has done that.

And now some short primer on complexity theory. The difficulty of solving a problem is expressed in the terms of time or space that is needed to solve the problem. The most important complexity classes are:

P: the problem can be solved in a polynomial amount of time with respect to the problem size.
NP: the answer to the problem can be checked in a polynomial amount of time. (All existing complete algorithms for NP problems use at least an exponential amount of time to find the answer).
PSPACE: the problem can be solved using a polynomial amount of space.

The three classes order in such way that P is believed to be easier than NP and NP easier than PSPACE (The first person to either prove this or refute it will become a millionaire).

Informally, a problem is complete with respect to some complexity class if it as hard as the hardest problems that belong to it.

The two strange looking complexity classes that I included above are defined using the concept of an oracle. The idea is that an oracle is some mysterious black-box machine that can instantly solve some problem. In both classes we use an NP-oracle, that is, a black box that answers NP-complete problems.

Delta^P₂ is the class of problems that can be solved in a polynomial time if we have an NP-oracle.
Sigma^P₂ is the class of problems whose answers can be checked in a polynomial time if we have an NP-oracle.

alfaniner · 10th November 2005, 09:14 AM

I always created a channel one block wide on the third column in from either side. This facilitated placing matching pieces (and those darn squares!) without leaving too many spires to deal with, and was always good for dropping the long piece into.

I haven't played Tetris in quite a while. I kind of miss it. A simple, elegant, addicting game.

Bronze Dog · 10th November 2005, 09:49 AM

I go for the empty column on one side, and I usually have the other side devoted to S and Z blocks.

Never had any trouble with squares. I always keep some place with two adjacent level blocks open. Not by conscious effort, though.

Can't remember where I read it, but apparently you can't play Tetris indefinitely, even with perfect reflexes: Sooner or later, the pseudorandom generator is going to give you a long, long stretch of nothing but S or Z blocks.

aerosolben · 10th November 2005, 03:17 PM

I usually build stacks of gold boxes on the two sides (silver if necessary), and do whatever I can with the middle.

http://en.wikipedia.org/wiki/The_New_Tetris

CurtC · 10th November 2005, 06:41 PM

Ahh, Tetris. I remember it from the old DOS days using a CGA color adapter. Crude graphics of the Kremlin etc. in the background. And I remember my delight in discovering finally that you could rotate the peices as well as move them left and right.

9th November 2005, 08:32 PM	#1
Yahweh Ayay ashay ayay Join Date: Apr 2003 Posts: 9,029	Ideal Tetris Theory I wanted to totally beat Ceptimus' highscore in Tetris at the Freethought Forum, and I found the scoring formula of Tetris on Wikipedia: Quote: The scoring formula for the majority of implementations of Tetris is built on the belief that more difficult line clears should be awarded more points. In Nintendo's implementations on the NES, Game Boy, and SNES, the four possible line clears are as follows: 1. Single = (level+1)40 one line is cleared. 2. Double = (level+1)100 two lines are simultaneously cleared. 3. Triple = (level+1)300 three lines are simultaneously cleared. 4. Tetris = (level+1)1200 four lines are simultaneously cleared. Code: Level 00 01 02 03 04 05 06 07 08 09 10 Single 40 80 120 160 200 240 280 320 360 400 440 Double 100 200 300 400 500 600 700 800 900 1000 1100 Triple 300 600 900 1200 1500 1800 2100 2400 2700 3000 3300 Tetris 1200 2400 3600 4800 6000 7200 8400 9600 10800 12000 13200 I'm a pretty good Tetris player, my technique has always been placing my pieces to get as many lines as possible (I get a lot of single lines), but I think there might be a better way. In a game of Tetris, there ought to be a way to maximize points by getting single-lines some percentage of the time, doubles some other percentage of the time, and so on. I've been looking, but I havent come across anything on the subject. Is there a serious ideal technique to play Tetris?
Yahweh Ayay ashay ayay Join Date: Apr 2003 Posts: 9,029

10th November 2005, 03:27 AM	#2
Matabiri Graduate Poster Join Date: Oct 2003 Posts: 1,734	Like this? http://www.lookatentertainment.com/v/v-53.htm
	__________________ "That's the kind of thing you can't look up on the internet, because it's the kind of thing you get taught at school." - Ashley Pomeroy

10th November 2005, 04:12 AM	#3
PatKelley Guest Join Date: Oct 2005 Posts: 610	My dad had a technique: he would use a "slowdown.exe" program to slow it to nil, and give him as much time as he needed to contemplate his next move.
PatKelley Guest Join Date: Oct 2005 Posts: 610

10th November 2005, 04:37 AM	#4
ceptimus puzzler Join Date: May 2003 Posts: 3,316	Originally Posted by Yahweh I'm a pretty good Tetris player Ha! Check out my highscores in the other two Tetris games at the freethought forum, as well as the one at mu.nu I hold all the Tetris trophies there. On each of those games, you get to see the next piece as well as the current one. You have to consider what the playing area will be like after BOTH pieces have landed to play optimally. I try to leave a channel down one side to drop the long pieces in (so as to get the 4-line bonuses) but once the pile gets more than 6 squares high, I abandon that strategy and try to complete as many rows as possible, so as to get the pile height back down to something manageable. You don't have time to think on the later levels - you just have to react - it takes lots of practice. Don't change your mind once you've decided where to place a piece - changing your mind while the piece is falling usually ends in a disastrous mistake. Edit to add: It was slimshady2357 who taught me the 'get the tetrises' technique. Before that, I used to go for just complete lines and not bother with the bonuses.
ceptimus puzzler Join Date: May 2003 Posts: 3,316	Last edited by ceptimus; 10th November 2005 at 04:51 AM.

10th November 2005, 04:44 AM	#5
LW Master Poster Join Date: Jan 2002 Location: 60°N 25°E Posts: 2,800	Originally Posted by Yahweh I've been looking, but I havent come across anything on the subject. Is there a serious ideal technique to play Tetris? A version of Tetris where the player knows all the pieces in advance has been proved to be NP-complete, meaning that it is a pretty difficult problem. Optimized versions of NP-complete programs tend to be even more difficult ranging from Delta^P₂ to Sigma^P₂ depending on the definition of optimality. Without doing any formal analysis at all, my educated guess would be that Delta^P₂ would be the correct class. Analyzing the real version where the player doesn't know the sequence of the blocks is more difficult and to my knowledge no one has done that. And now some short primer on complexity theory. The difficulty of solving a problem is expressed in the terms of time or space that is needed to solve the problem. The most important complexity classes are: P: the problem can be solved in a polynomial amount of time with respect to the problem size. NP: the answer to the problem can be checked in a polynomial amount of time. (All existing complete algorithms for NP problems use at least an exponential amount of time to find the answer). PSPACE: the problem can be solved using a polynomial amount of space. The three classes order in such way that P is believed to be easier than NP and NP easier than PSPACE (The first person to either prove this or refute it will become a millionaire). Informally, a problem is complete with respect to some complexity class if it as hard as the hardest problems that belong to it. The two strange looking complexity classes that I included above are defined using the concept of an oracle. The idea is that an oracle is some mysterious black-box machine that can instantly solve some problem. In both classes we use an NP-oracle, that is, a black box that answers NP-complete problems. Delta^P₂ is the class of problems that can be solved in a polynomial time if we have an NP-oracle. Sigma^P₂ is the class of problems whose answers can be checked in a polynomial time if we have an NP-oracle.

10th November 2005, 09:14 AM	#6
alfaniner Penultimate Amazing Join Date: Aug 2001 Location: Sorth Dakonsin Posts: 11,384	I always created a channel one block wide on the third column in from either side. This facilitated placing matching pieces (and those darn squares!) without leaving too many spires to deal with, and was always good for dropping the long piece into. I haven't played Tetris in quite a while. I kind of miss it. A simple, elegant, addicting game.
	__________________ Science doesn't lie.

10th November 2005, 09:49 AM	#7
Bronze Dog Copper Alloy Canid Join Date: Mar 2005 Location: Homebrew D&D Campaign Setting Posts: 5,007	I go for the empty column on one side, and I usually have the other side devoted to S and Z blocks. Never had any trouble with squares. I always keep some place with two adjacent level blocks open. Not by conscious effort, though. Can't remember where I read it, but apparently you can't play Tetris indefinitely, even with perfect reflexes: Sooner or later, the pseudorandom generator is going to give you a long, long stretch of nothing but S or Z blocks.
	__________________ Stop Sylvia Browne Warning: Beware of contaminated water supplies! Suspected source of contamination: Sarah-I A non-Rockstar Rambler and dissector of Doggerel

10th November 2005, 03:17 PM	#8
aerosolben Evil Genius Join Date: Aug 2001 Posts: 2,270	I usually build stacks of gold boxes on the two sides (silver if necessary), and do whatever I can with the middle. http://en.wikipedia.org/wiki/The_New_Tetris
aerosolben Evil Genius Join Date: Aug 2001 Posts: 2,270	__________________ You can tell a lot about a fellow's character by his way of eating jellybeans. - Ronald Reagan

10th November 2005, 06:41 PM	#9
CurtC Illuminator Join Date: Aug 2001 Location: Dallas, TX Posts: 4,758	Ahh, Tetris. I remember it from the old DOS days using a CGA color adapter. Crude graphics of the Kremlin etc. in the background. And I remember my delight in discovering finally that you could rotate the peices as well as move them left and right.