2-state or 4-state?

I was having this debate with Rosti on IRC the other night, and I thought I'd continue the discussion on here seeing as some of you may find it interesting and have something to contribute. In summary, we were discussing which is the theoretically more logical behaviour for the S, Z and I pieces; 2-state rotation or 4-state rotation.

Seeing as the ideal solution of having the tetrominos rotate about their true geometric centre is not possible -- as it would cause them to rotate out of alignment with the grid -- the two main alternative solutions are 2-state rotation (as seen in Pajitnov's original, the Spectrum Holobyte versions, the GB and NES versions, SEGA Tetris and its descendants, Atari Tetris, TetriNet, and others), and 4-state rotation (as seen in Tetris 2 + Bombliss, Super Tetris 2 and 3, Tetris DX, most other BPS versions of Tetris, and all guideline games after Tetris Worlds). For the time being, we're going to disregard wallkicks and high gravity, because they're an extra layer of complexity which may obfuscate the main line of argument.

Rosti was arguing the case for 4-state, and insisted that having a constant centre of rotation, which lies as close to the geometric centre as possible, is the most logical and intuitive way to do things. As a result, the S, Z and I have 2 horizontal states and 2 vertical states.

I argued that at low gravity, all S, Z and I placements can theoretically be made with a single vertical state and a single horizontal state, and it doesn't seem logical to have redundant states merely to provide a constant centre of rotation. I went on to suggest that a lot of players (myself included) don't intuitively think in terms of rotation about a fixed point anyway, and instead simply visualise going from one state to another.

Rosti countered my claim of "redundant states" by saying that the two vertical states allow for faster finesse. But this seems like a non-argument to me; if you're going to introduce more states than the minimum necessary, just because it's faster, then why stop at two? From a theoretical viewpoint, introducing redundant states should be taken to the logical conclusion of Typomino.

This is where we left off; Rosti went to bed and I went to brush my teeth. But what do you guys think? Is it logical that we must have two horizontal states because one is a 180 rotation of the other about a fixed centre? Or do people see them both as simply "a horizontal state" and regard one of them as being redundant?

 

You're thinking about it backwards. They wanted rotations to be a spatial rotation about a certain point, and 4-state S/Z/I came as a result, not the other way around. Also, it makes more sense that it should be that way when you factor in the VRP theory of SRS wallkicks.

That being said, I am with you in that I just think of a rotation as getting me to a certain orientation. ARS works better with me for this reason, but I can see why the design of SRS is logical to an extent (as long as you stop at the basic rotations and the idea behind the wallkicks).

 

I defiantly do not see these rotations as redundant, however it mainly applies to the I piece, not so much the S or Z pieces.

From the Tetrisconcept wiki (SRS):


Whether or not this response specifically answers your question, if the game (TGM3) used a 2-state rotation, only one of these moves would be possible.

 

Uh, there's no reason to shortcut both of those positions. IRS+DAS is a completely reasonable way to reach the second case quickly. Having it pop back to that position instead of kick straight up is more of an annoyance than anything.

I'd say the horizontal orientations are redundant (barring twists) considering that they have the same footprint. If you want to avoid upper/lower oddities at 20G and still have the leftmost/rightmost bias, you can do something like DTET's 3-state.

I need to think about this a bit though, I'll probably weigh in with a more detailed post later.

 

Whoops... I can't read. Assuming 20G my previous post would be more valid.

 

I was assuming 20G for the sake of the side argument you presented, though. You don't need 4-state for that.

http://fumen.zui.jp/?m110@rdF3gbB3g...bEYecAwNAA7eDB?TqDAEizBAhUqGAy3WeD0oBAA5eBAAA

 

I'm actually not thinking about it backwards; I do see the 4-state rotation as a result of the fixed centre:

I assume the bit that led you to that conclusion was:

I guess that was poor wording on my part; I should have phrased it something like: "it doesn't seem logical to have redundant states merely because you wish to have a constant centre of rotation." Still a bit ambiguous, but meh.

 

Well, it does, because otherwise you don't have a constant center of rotation anymore

 

This has been on my mind recently.

How would people feel about 2-state, but with CW basic rotation always going right and CCW always going left? This lets you do the same 0G finesse as 4-state, without any of the vertical motion. And you would never have to remember which of the 2 visually identical states you are in, because left/right bias would be tied to the button and not an internal game state.

 

That's an interesting idea, but I think would introduce some odd behavior. I guess it would depend on the implementation though.

One method would require you to still know what state it's in, creating a 3 state rotation, and another would allow you to walk the piece with the rotate buttons.

 

Was so fucking confused reading this thread until I realised that OP was from three years ago...

 

The main issue is S and Z wallkicks. the ability to get it to round the other way would allow for hopping, but only to the right, unless a special exception was given to the Z piece to let it prefer left when rotating CCW.