Super Random notes probabilities

DeusExMachina · 03-13-2007, 01:47 AM

I noticed the probability for a different colored note is not equal for color combinations at a given moment.

Ok that probably doesnt make any sense but an example is...

suppose you have 30 single red notes in a row in any particular song. On super random, if the randomness was evenly distributed you should see 10 greens, 10 reds, and 10 blues as that is what would be statistically favored. Based on what I observed thats not really the case, and in that situation it would be more like 20 red, 5 green 5 blue. Anyone else notice that?

If the original note is red, in super random the probabilities are like 50% red 25% green 25% blue, instead of 33% red, 33% blue, 33% green. Thats what I think

oni-ice · 03-13-2007, 06:38 AM

It looks totally randomness to me. No real numbers to SRA IMO.

legsweeper · 03-13-2007, 07:33 AM

I've never noticed anything like that.
Of course sometimes the notes wont change at all but sometimes it'll be "30" blues or greens.
The only thing that isn't random is of course open and all color notes.

Melchior · 03-13-2007, 10:23 AM

I noticed that as well~ getting a lot of the same color in a row just shouldn't happen with normal probability, so I've wondered about it to.

Maybe in the same way that random only switches between measures, super random only switches between beats or half-beats? Or maybe it doesn't switch unless (a) it's on a beat and (b) there was a minimal amount of distance from the last note. Or maybe something is actually programmed in deeper. Maybe it only switches notes if the mapped sound is actually changes.

*shrugs* Haven't really looked into it myself (don't play super random here), but those are a few theories that might be worth looking into.

jubrany · 03-13-2007, 11:19 AM

is it just me, or was the SRA algorithm much crazier (and cooler) back in the older AC mixes (like GF 8th for example)? I could've sworn that the older mixes would take single note patterns and even turn those into complex switching patterns.

so for example for a simple pattern like:

R
R
R
R
R
R
R
R

on Random would be

B
B
B
B
B
B
B
B

and on SRA might be

R
R
B
B
G
G
B
R

but I remember in the older mixes, SRA would turn that into

RG
GB
RB
G
B
RGB
GB
RG

or something crazy like that, thereby making almost any easy song rediculously hard in SRA.

am I just remembering things wrong, or has Konami really watered down SRA?

BobbyS · 03-13-2007, 12:44 PM

well, if it was 33/33/33 then a lot more songs become inhumane. I'm not complaining. SRA is great and with the exception of very few songs (like metal queen), it offers a great and do-able challenge.

Honestly.... 33/33/33 might actually suck on some songs.

aeriyu · 03-13-2007, 03:25 PM

A few things

- If the original notechart has 1 note, then SRA will randomize that 1 note. If there's 2 notes, then SRA will produce 2 notes; and if you hold down all 3 frets, then there isn't any randomization

- SRA actually has its own notepattern, I believe, but to prove this I'll have to actually pay attention the next few games I play. I don't know why I believe this... perhaps because notecharts sometimes feel "familiar" (ie: Micro Fin SRA in the 2fret stream part) on SRA. It'd definitely make songs like Famiresu and Fairy Tales on SRA a lot easier if they had a distinguishable pattern... but then again, maybe this is all in my head

kknd · 03-13-2007, 04:12 PM

... hey! If you think about it this way, it might be logic afterall
25% unchanged
25% red
25% blue
25% green

so a red has 50% of chance of staying red and 50% of changing!!!

BobbyS · 03-13-2007, 05:15 PM

I don't pay attention when I play SRA, but I seem to think it has a pattern too. To me, it just feels like it creates a more difficult note chart and doesn't really change (but I guess it's never the same twice, eh).

DeusExMachina · 03-14-2007, 01:15 AM

Ill do an experiment with a song then. I will post results soon.

AdamW · 03-15-2007, 08:41 PM

on a technical level it's entirely possible there could be some kind of pattern to it, purely because it's theoretically impossible to produce true randomness with a computer (since they're deterministic machines) and it's pretty damn hard to fake it convincingly.

*TTRlovesMittens* · 03-16-2007, 07:17 PM

Quote:

Originally Posted by AdamW

on a technical level it's entirely possible there could be some kind of pattern to it, purely because it's theoretically impossible to produce true randomness with a computer (since they're deterministic machines) and it's pretty damn hard to fake it convincingly.

Um... Random.org?

It IS possible to produce true random number generation

AdamW · 03-17-2007, 03:24 PM

yeah, sorry, I should have said without using an external source of entropy.

see the essay on that site: random.org - introduction to randomness and random numbers

Random · 03-21-2007, 12:47 AM

It doesn't seem all that illogical to me - the game doesn't randomize sets of notes, it randomizes one note at a time. The only thing that stays the same is the number of buttons you press per note - so open and three-button notes don't change at all. That leaves you with three combinations each for single- and double-button notes.

Think of it this way - if you flip a (perfectly balanced) coin ten times, and it comes up heads every time, does that mean the next flip has to be tails to even out the probability? No, it's still 50-50.

Same thing here - just because there haven't been many greens in a certain run doesn't mean the next button is going to be a green - it's still a 1 in 3 chance. Even if it might seem as though the game is doing some weird calculations at times... well, that's probability for you. :P (I just finished a long run of s-ran on MPG, and there really wasn't a pattern in the end. Just things that seemed like patterns at the time, and then they didn't happen again.)

Personally, I think it'd be cool if they introduced a new "super super random" or "ultra random" mode, where each note has an equal probability of getting set to any of the seven [eight] other combinations of buttons. (R, G, B, RG, GB, RB, RGB, [open]) I'm sure you can imagine what kind of challenge that would be. ^_^

<== also, Random. lol

Hortence:The Muleface Mod · 03-21-2007, 11:34 AM

You're not very super though. :<

ps jk

I've noticed this a lot in Pop'n also, sometimes it seems like SRan affects clusters of notes at a time rather than just individually, then switches it up. So what Sran might do is grab a random number of notes and apply a normal random to them, making some patters stay somewhat in tact, but breaking them up in the process. Just a thought, and I'm not sure if Sran would work the same across different games, even if it is the same concept.