i was playing today and got a five long S block "river" (as opposed to "drought"). i was wondering if anyone has seen a longer river? kitaru mentioned that is very unlikely as the game checks the last piece and rerolls before allowing you to have the same piece again.
Maybe the saving throw memory resets after failing (and the first identical block is spawned)? Otherwise the statistical chance is ludicrously low for this to happen.
I ran some numbers. If I understood Alex correctly: Roll 1 = 1/7 chance of getting any block Roll 2 triggers a third Roll if an identical block is rolled (saving throw) Your chances are then: (1/7)^2(*n-1) where n is the number of identical blocks in a row This means you have the following probabilities of getting any n identical blocks (of any of the 7 types) in a row: 1: 100% (duh) 2: 2.04% 3: 0.04165% 4: 0.00085% 5: 0.000017% So out of 100,000 sequences of 5 blocks, you might see 5 identical blocks about twice. And you caught it on video!
If the saving throw doesn't exist at all, then your changes go way up. n identical blocks in a row 1: 100% 2: 14.285% 3: 2.04% 4: 0.2916% 5: 0.0416% Since we've all seen 3 and 4 blocks in a row pretty frequently, it makes me question whether the saving throw applies after the second consecutive block. I'm guessing it doesn't, which would lead to probabilities somewhere in between those between these two tables (which seems pretty darn believable).
if the first instance wasn't weird enough... check this out. same day, same location- six Os in a row. can someone calculate this?!? this has been happening all day (or at least it feels like that to me but i may just be looking out for it now). i have been getting sequences of 2,3, and 4 identical pieces a lot. is this a malfunction or is the machine on the fritz?
In the past, we've used flood as an antonym to drought. River works OK too. (In that vein, perhaps "stream?") Note the dummy value -- if the first roll whiffs and comes up with no piece, then we need the second roll to get a result. So, it's 2 out of 8 values that result in a re-roll rather than 1 out of 7. The re-roll applies to every piece equally. The history is only cleared when a new game begins. So far, we've only discussed the theoretical implications of the high-level selection method. You could also consider the behavior of the low-level pseudo-random number generator, but that's probably getting a bit outside the scope of this thread. At any rate, while the conditions that would yield such a sequence are extraordinarily rare, it is still possible for them to come up eventually. (I wouldn't suspect a faulty NES just yet! I think you'd be witnessing some other significantly strange behavior if memory corruption were at play.) I won't lie, it was pretty damn eerie watching those floods. However, technically it's possible for much more unlikely things to happen. EDIT: Related thread: http://tetrisconcept.net/forum/showthread.html?t=512 -- though, it's more about intervals than repetition. I also thought NES's curve was discussed in this thread, though I guess it may have been something that came up in conversation on IRC instead. Also, if you look at all of the randomizer states possible going into the beginning of a game, only 0.98% of them result in a double O start. To go from that into whatever timing was necessary to yield four more O's is really quite something.
thanks for the response (and sorry i had to text you about it earlier- i assume you were busy at work but i HAD TO KNOW)! after i dissected, and eventually digested your post (you use so many god damned ten dollar words kid) i think the point is this: what i experienced today is one of those improbable-but-not-impossible types of things. still, the fact that they both happened on the same day freaks me out. eerie is the right word, sir. how long must i now wait for a 7 piece flood, river, or stream?
No worries. In fact, the timing coincided perfectly with the return trip from a company lunch, haha. Hahaha, sorry. Yep. I can't say I've ever seen a flood that persistent from NES, but you have to figure that enough games have been played out there that someone has run into them. :s The world may never know... Though, it's something I'm tempted to devote some time programming to figure out -- at least as far as brute-forcing for floods goes.
I don’t know if I’m hallucinating, but I thought I recalled something about an additional re-roll on consecutive line pieces…
from what i're read above (calculation uses conditional probability only) you have (1/2^n odd to get a specific tetromino n consecutive times after having drawn it once. For example if you get an O, getting 5 others O's in a row from this point has (1/2^5 probability (1 out of 17 million).
My fault. this misconception comes from me. from a misinterpretation of alex long ago haha.. Nobody ever listen to anyone but alex ever about anything from here on out
Hmm... /me bastardizes You know, you really shouldn't be giving me any more ideas... There are others here who know their shit. He just tends to be the one who can type faster than anyone else.
It's really cool to have those big floods on record, especially so close to each other. (All those deep blue s blocks linked up did kind of look like a river.)
Alex, I just read on one of your youtube videos that NES rerolls if you get two of the same pieces in a row. Did you find this to be false then?
I believe what Ben was referring to was a misconception that different pieces received different numbers of rolls, or something to that effect. The game uses the same procedure for each piece deal. Every time you see a repeat, this happened behind the scenes: 1) "the 8-faced die" came up with the repeat or "blank" 2) "the 7-faced die" came up with the repeat
