Is the normal randomizer fair for Marathon mode?

[aoeu256]

Im talking about the randomizer that sets the next piece to any of the pieces letting you repeat pieces endlessly.


Yeah its a open ended question, but I want to know your opinions of it.

[Rosti LFC]

As in the Tetris DS randomiser, with a 7-piece bag?


I think it's probably the best randomiser without going to the complicated ones used in TGM2 and Ti. It keeps the gameplay balanced, and it allows experienced players to plan their stacking in anticipation of future pieces.

[kotetsu213]

I think he meant "pure random", as in, start with a Z, 46 pieces later get 4 I's in a row, followed by 5 O's and having nowhere to put them.


I don't know about you, but I couldn't stand that at all in older Tetris games.

[aoeu256]

Yeah I mean the purely random one, but for Marathon mode (i.e: not getting Tetris).


(I know about the S and Z piece, but that seems very improbably...)

[Rosti LFC]

I think he meant "pure random", as in, start with a Z, 46 pieces later get 4 I's in a row, followed by 5 O's and having nowhere to put them.

I don't know about you, but I couldn't stand that at all in older Tetris games.


Oh, like N-Blox on www.neave.com (http://www.neave.com/default.htm)

[colour_thief]

It's fair if you want to play slowly and/or not go for tetrises.

[tepples]

The "pure random" randomizer, called "Memoryless" in Lockjaw, is a lot better than bag or history if you're playing Square mode, which in general tends to be a slower, more cerebral mode.

[herc]

a question (surely asked a thousend times before, and surely deducible from the wiki, but i am a bit in hurry)


whats the exact advantage of the tgm 6history x-reroll randomizer over a 7 piece bag? only difference i can see is that 7piece bag has on bag seams the possibility of spitting out 2 same pieces in a row.

any other obvious differences? or is it about more subtle differences that tgm randomizer is generally more preferred here?

[colour_thief]

Randomizer Theory (http://www.tetrisconcept.net/forum/viewtopic.php?t=401)


Curve showing probability receiving the same piece again within x pieces:




Average predictability for each algorithm:

RND
Average: 14.29%
Std. Deviation: 0%
Not very predictable, but constant in it's unpredictability.

TGM1
Average: 31.77%
Std. Deviation: 2.70%
Fairly predictable, and fairly consistently predictable.

TAP
Average: 32.76%
Std. Deviation: 1.78%
Marginally more predicatble than TGM1 and marginally more consistently so. It shows the limit of the TGM randomizer design: Increasing the the number of rerolls will only bring it closer and closer to 1/3 predicatbility with 0 deviation.

SRS
Average: 37.04%
Std. Deviation: 28.07%
The most predictable randomizer by a fair amount, but also very inconsistent.

[Needle]

Note also the randomizer's tendency to avoid S/Zs in the opening of the game. (Never deals S/Z/O in the beginning; initial history state set to ZZZZ in TGM1 and ZSZS in TAP)

[Air Gear]

I don't see what's unfair about using a memoryless randomizer for a marathon-style game since the marathon form is mainly about surviving against all the unpredictable, nasty piece combinations the game throws at you. However, the best use for memoryless is in a two-player versus game where pieces given to each player are synchronized and, as such, both players will get screwed by the randomizer.

[DIGITAL]

I don't see what's unfair about using a memoryless randomizer for a marathon-style game since the marathon form is mainly about surviving against all the unpredictable, nasty piece combinations the game throws at you.
Hmm, I always thought Marathon encouraged score/time. Maybe such unpredictabilty is better suited in an Endless mode?

However, the best use for memoryless is in a two-player versus game where pieces given to each player are synchronized and, as such, both players will get screwed by the randomizer.

Even though both players get the same order of random pieces, their field will probably be vastly different; thus it may screw one player over more than another. And let's just face it, getting screwed by the randomizer is just not as fun as being able to incorporate it into a stacking strategy. Would you rather win against an opponent because he was screwed worse or because you had stacked well?


EDIT: Changed an error in wording the last sentence.

[colour_thief]

I strongly agree with DIGITAL.


When you play with a memoryless randomizer, you have to play expecting the worst because it can (and somewhat probably will) happen. To play for consistent results is to play extremely conservatively. How this gets extra stupid is that the scoring system encourages you to build tetrises... a strategy that, if followed, leaves you very much at the mercy of the randomizer. Following this strategy there are times when, before you even begin, you are doomed to failure. Respect to the guys who have pwned early Nintendo Tetris games, but for each successful superplay that you see there are inumerable failures through no fault of the player.


I'd rather not have a game waste my time like that. I'd rather play games where I have only myself to blame.

[Pineapple]

So, on the one hand, we have a randomizer that is completely fucking unpredictable. Even with 8 previews, memoryless will still fuck you up the ass. And it's not just the scoring system that encourages you to play at the mercy of the randomizer. If you're playing for speed, you also have the extremely long line clear delay (91 frames!) to contend with as well.


On the other hand, you have the bag randomizer, which (do I really have to whore out the link the the proof again?) has already been shown to be grossly consistent at low speeds. Even if we never find a loop in less bags, the very fact that our current solution exists is bad for the chances of the bag randomizer ever being taken seriously.


There's still a lot of middle ground between these 2 possibilities to be explored. But so far, no-one (apart from Mihara) has dared to explore it...

[caffeine]

i don't get it. how's any randomizer not fair?

[colour_thief]

i don't get it. how's any randomizer not fair?

It's fair if you want to play slowly and/or not go for tetrises.


It's not unfair in the competitive sense... everyone is using the same thing. It's unfair to players because it wastes their time with worthless sequences.

[Taratang]

i don't get it. how's any randomizer not fair?

Substituting "fair" for "reasonable" is probably better. It's unreasonable to expect a player to stack for tetrises only with purely random piece selection.

[tepples]

If you're playing for speed, you also have the extremely long line clear delay (91 frames!) to contend with as well.
That's the first Game Boy version. The NES version, the Super NES version, and the revised Game Boy version have a much shorter line clear delay.

There's still a lot of middle ground between these 2 possibilities to be explored. But so far, no-one (apart from Mihara) has dared to explore it...
Once word of our solution reaches Mr. Rogers, there might be a possibility.

how's any randomizer not fair?

"That video game took my quarter and didn't even give me enough I pieces."

[Pineapple]

If you're playing for speed, you also have the extremely long line clear delay (91 frames!) to contend with as well.
That's the first Game Boy version. The NES version, the Super NES version, and the revised Game Boy version have a much shorter line clear delay.
I looked on the wiki. Was going to say 40 frames before looking it up.

There's still a lot of middle ground between these 2 possibilities to be explored. But so far, no-one (apart from Mihara) has dared to explore it...
Once word of our solution reaches Mr. Rogers, there might be a possibility.

There's several possibilities, in fact. I'd be interested to see what our Hawaiian friend says, though...

[Needle]

Well, when talk about the guidelines happen, most of the time it only scratches "WHETHER it's against the guidelines or not" and almost never delves into "WHY it's against the guidelines" or "WHY the guidelines were made that way", so I don't really have high hopes...

[herc]

Once word of our solution reaches Mr. Rogers, there might be a possibility.
"

what is "our" solution? TGM style randomizer? ore do you have something even better?


On the other hand, you have the bag randomizer, which (do I really have to whore out the link the the proof again?) has already been shown to be grossly consistent at low speeds. Even if we never find a loop in less bags, the very fact that our current solution exists is bad for the chances of the bag randomizer ever being taken seriously.
"

i dont quite understand (mostly because of my lacking english capabilities)!

1. yes, please give me that link

2. "grossly consistent" sounds bad. but isnt "consistent" good?

3. "never find loop in less bags" i do not understand this.


please, explain, its a hot topic for me (and blockout II)

[colour_thief]

He's referencing this (http://www.tetrisconcept.net/forum/../wiki/index.php?title=Playing_forever). A game is broken if it has all of:


7-bag generator

Hold

4 previews

Gravity no higher than about 1G