Talk:The New Tetris

63 piece bag

Zaphod here. The game most definitely uses 63 piece bag, and is the game that bag was developed for. In order to build gold squares, you need 4 of the same. Building squares is not very practical using a history based randomizer or 7 piece bag. Either way, you won't get the 3rd or 4th piece soon enough. Memoryless allows repeats, but can just as easily deny you the critical 4th piece until is too late. The 63 piece bag allows for the pieces to clump up, and thus form golden squares more easily. If it shorts you on a piece for a while that means you'll get a run of them later to build a gold square. In fact it probably gives MORE repeats then straight memoryless (statistical analysis, anyone?). As for how I know, you 'll just have to ask me on IRC or PM me.

Future testing

CT suggested a couple more scenarios:

Does gold dominate over top?

More difficult: Does top dominate over left?

CT said this one was deemed impossible to test (and thus immaterial to a description of the algorithm), as the other O piece can't fit through. However, it certainly is possible (though difficult) to test this by making a hurdle:

The O pairs at top and bottom can be replaced with I pairs, J pairs, or L pairs, but the Os in the center must remain O.

--Tepples 20:22, 30 November 2006 (EST)

Actually, the "O pairs" at the top and bottom can't possibly be Os or else the Gold square priority will ruin the test. --Colour Thief 21:25, 1 December 2006 (EST)

Ok, I've tested these in Tetris Worlds on GC. It is left biased and down biased, which is different than TNT. When I tested the relative priorities (note this took the mirror image of the general hurdle test above) I found that Gold > Down > Left. I'm not sure how you'd like to fit this information into the articles, but go about it as you see fit. And be sure to test TNT as well; I'm curious to see the results. --Colour Thief 21:58, 1 December 2006 (EST)