Talk:Playing forever

The random factor will make let's say 10 S figures appear in a row, thus wrecking your scheme. Btw I came up with exactly the same scheme a year ago, and proceeded to obliterate it with the current argument.

Cheers! -- 82.11.190.158
 * What you talk about does not apply to newer versions of Tetris. Older versions used a memoryless random number generator. Newer versions generate a bag of all seven pieces in random order, then another bag of all seven pieces, etc. See Tetris Guideline --Tepples 19:01, 9 July 2006 (EDT)

That won't work
TZS, ZTS:

I don't understand. If you get dealt ZTS, how are you going to put the T under the Z? You can't slip it under unless you've got the Z against the right wall, and I was under the impression you were putting this well on the left wall. Perhaps what you ment is this:

ZTS:

If so, why don't you enumerate all the possibilities so that it is clear.
 * Yup, I had "The New Tetris box art syndrome" (reversing Z and S in diagram). Besides, in both SRS and TGM, T can kick under an overhang. I'll go back and clarify that section. --Tepples 16:58, 3 August 2006 (EDT)

STZ@High-Speed Theory:
How about you just put the STZ into the middle, while making JLO a few blocks shorter on purpose? That way, at higher speeds, can't you make a hill with them, and thus move the other pieces into the left and right wells with more ease?

Having problems holding pieces
OK, this works well until I have to hold something. I put the piece in hold, and I can play on for a while, but when do I bring it back? I find I always seem to end up in a situation where I'm switching one piece I can't use for another that was already being held. The page doesn't seem to give any instructions on this matter. I've tried switching them for I pieces, because that doesn't really disrupt any patterns, but I'm not sure that that always works. Things still seem to get muddled. Am I just not thinking ahead far enough?

Also, I'm having difficulty working out exactly what the following bit means:


 * If the STZ heap becomes too high, and its floor is shaped as 1210, 0121, 2101, or 1012, the player can borrow a T tetromino from the STZ heap, placed on its side, to create a single line in the center. This results in no change to the floor shape, and it works especially in cases where the player would otherwise have to "hold the T and play as above".

What does it mean by 'a single line in the center'? The centre of the left pile, or the centre of the play area overall (as in the I pile)? And when it says 'on its side', which way is the point facing? A picture would be helpful. I find the left pile gets high quite quickly, so clarification would be appreciated. — James 11:00, 15 November 2006 (EST)


 * When you are instructed to hold a piece, you hold it until the seam between bags. And here, "center" refers to columns 5 and 6. Both the left pile and the right pile should pile up at the same rate. I'll see what I can do in the article. --Tepples 19:29, 15 November 2006 (EST)


 * OK, thanks. That was most helpful. — James 21:12, 16 November 2006 (EST)

Incomplete Looping
Hi Tepples. You explicitly show each of the 3 heaps looping. However, you don't show the "superheap" of the 3 heaps looping. You mention how the sides will increase faster than the centre, and also mention ways of bringing down the sides. This is good; all the tools are there. But there is no clear demonstration of a looping superheap.

When considering the superheap 2 bags at a time: The left heap will increase by 6 units with standard play and 4 units when using Ts for singles. The right heap will increase by 6 units with standard play and 4 units when using Js for singles. The center heap needs special consideration, subdividing it into column 5 and 6. Depending on how exactly you clear singles to bring the sides down, the 2 columns will not increase identically. The current proof does not show how an imbalance in the center columns can loop and be restored to balance. -- Colour Thief 17:32, 23 June 2007 (EDT)
 * To balance a too-tall column 6, put more I tetrominoes in 5. --Tepples 18:35, 23 June 2007 (EDT)

Oh, I can see that much... I understood the proof to begin with. It was more of a suggestion to bring it more clarity. You explicitly show the left, centre and right heaps looping, and the latter 2 cases are pretty trivial. Looping the relative heights of the heaps is easily almost as involved as the SZT heap, so I thought it would be good to show the explicit loop rather than inferring it. -- Colour Thief 19:19, 23 June 2007 (EDT)

Very nice addition with the .gif you added. :) But I think it highlights the point I'm getting at... The .gif shows placements of an I piece in the 7th column, and singles with pieces placed horizontal instead of vertical. There's nothing wrong with that of course, but these moves are not described in the article. Nor should they be, as I doubt they are necessary to play forever and thus would make for a less elegant proof. They main idea I'm trying to push with my comments here is that the current proof does not present a systematic way of choosing where to place the pieces. All the necessary tools are described... An intelligent person can survive forever with them... But because there is no deterministic way of placing the pieces the proof isn't airtight. I've got an idea for a simple way of looping things. If it pans out I'll edit the article with the extra placement rules (unless you beat me to it). -- Colour Thief 20:05, 24 June 2007 (EDT)


 * I admit that I made a mistake in that video; check its description page. I guess the I in the 7th column was part of a demonstration of error recovery to show that a valid state under this proof is reachable even after a play mistake. --Tepples 22:48, 24 June 2007 (EDT)

Well, that was pretty cool. I don't have a tetris game to check with, but I believe I have a relatively simple, non-ambiguous set of rules to balance the heaps in a cycle of 20 bags. Put another way, not only would you play forever, you would get a bravo every 140 pieces. -- Colour Thief 21:24, 24 June 2007 (EDT)
 * Please add it if you think it would improve things. --Tepples 22:48, 24 June 2007 (EDT)

I added it and took quite a bit of liberty re-organizing the rest. Feel free to change it as you see fit or slap me for any logical errors. -- Colour Thief 02:40, 25 June 2007 (EDT)
 * Why is it constructed such that 5 previews are needed? TGMA, Tetris Evolution, and a few cell phone versions have only 3. --Tepples 21:21, 25 June 2007 (EDT)
 * I think I found your purported worst case: LZSTIxx. You claim to need to see the first x in order to know where to put the L. But in fact, you could hold it and bring it back out when the I comes out. The new worst case is TLSZIxx, where the pattern requires that the T be put down last (so hold is in use), meaning you need at least 4 previews to do it as advertised. --Tepples 22:54, 25 June 2007 (EDT)

Animation example has a subtle error OH WAIT
I would complain that the article claims 4 previews are needed while the animation clearly has only 3... But only 3 are needed now! Do I hear a boo-yeah? --Colour Thief 02:04, 14 October 2007 (EDT)
 * Maybe he can predict the future... --Lardarse 02:16, 14 October 2007 (EDT)
 * It was recorded with more previews and played back with fewer. Besides, it took a few rerecords to get one that ends on a bravo and not a worst case. --tepples 10:28, 14 October 2007 (EDT)

Tetromino stacking != Tettromino tiling
''Is it possible to do a bravo in 5 bags? No, it is not. The reason is because you cannot pack an odd number of bags into any rectangular shape. The proof is simple, and linked at the third reference below. Therefore all solutions MUST be a multiple of 2 and of 5, which means all solutions are a multiple of 10.''

''This leave the problem of finding a sustainable loop with Ts in it that only uses two of them, and leaves sustainable loops for the other pieces that are also complete in 2 loops. Can this be done?''

My reason for reverting is as stated in the title of this section. This remains an open problem, and at least 2 people here believe that it might be possible with 5 bags, if you get the pieces in the correct order. If you can demonstrate that it isn't possible, either by proof, or by brute force, then it will be included in the article.

I see that you ammended the sentence just before my revert:

''Is it possible to do a bravo in 5 bags? Perhaps, but because it is impossible to tile any group with an odd number of Ts into any sort of rectangle, any such solution will require holes to be made and then recovered from as part fo the loop, making it very difficult to sustain.''

I still don't think that this is completely correct, but it's a lot closer... --Lardarse 19:50, 6 February 2008 (EST)

given a memoryless randomizer, the least number of tetrominoes that can produce a bravo from an empty state is 5. With the "memory 4-try randomizer" as seen in TGM (among other games), the least number of tetrominoes that can produce a bravo from an empty field is most likely 10 tetrominoes. here's my reasoning behind that:


 * you are almost guaranteed an I-block in any given set of 10 pieces. the likelyhood of a continuous cycle of 5 pieces is astronomically unlikely. I'll have the solid numbers to back that statement up later today.


 * assuming hold is permitted, the I-block can be at almost any position in the set of tetrominoes. there are a few patterns in which this would not be the case, but I'm trying to figure those specific cases out right now.


 * There is no probable sequence of 5 tetrominoes that can result in a bravo given a memory 4-try randomizer. I say "probable" because there is always the chance that you'll get a repeat within a set of 5 tetrominoes, and at least 1 repeat would be needed (a pattern of I-J-O-J-I, for instance, can produce a double without disrupting the field)

the above, however, assumes you know what the pattern is going to be ahead of time, which is also impossible in a standard environment.

with a 7-piece bag, the number of tetrominoes required to produce a bravo could increase tremendously AND the absolute minimum required is guarenteed to be 10 tetrominoes due to the lack of repeats within a bag. I'm willing to bet that quite a few bags of tetrominoes with 10 pieces (with a maximum of 1 duplicate of any given piece) can be found.

and before anyone jumps on me, this is all rough theory that I haven't analyzed too far. --Zeta 08:21, 16 May 2008 (EDT)


 * You can bravo with 5 pieces in TGM. The most likely sequence is one containing LJIOO. Also, in terms of playing forever, it is straight up impossible with a simple, canned approach like the one in the article. Even if you assume large floods and droughts are impossible. This is because, just like pure random, as the number of pieces dealt approaches infinity, the difference in piece counts can get arbitrarily large. Only "closed" bag randomizer (like 7-bag and 14-bag but not 8-bag) do not have this property.--Colour Thief 14:13, 16 May 2008 (EDT)

assuming no repeats within the memory, the absolute minimum is 10. I just pulled off a bravo in exactly 10 lines (25 pieces), so every multiple of 5 pieces holds the possibility of a bravo. --Zeta 14:55, 16 May 2008 (EDT)

Making this more readable
So I decided to do something crazy, and took a look at what this page looks like in Lynx (the text-only web browser). And it looks kinda bad... The empty blocks just disappear, leaving pfrow diagrams look like theyv been hit with a "move left" item.

So I was wondering what could be done to make them look better. Any ideas? --Lardarse 14:57, 15 May 2008 (EDT)
 * I don't see the point in changing things to work with the lowest common denominator. Tetris is graphic in nature, and anyone using Lynx should expect an unsatisfactory experience from such graphic intensive pages. Make changes if you want, but definitely nothing that hinders maintainability or the user experience with standard browsers. --Colour Thief 13:15, 16 May 2008 (EDT)
 * It's probably easier to just make a text-only version... --Lardarse 14:12, 16 May 2008 (EDT)