CTWC Main Event Reset rule?

Thread in 'Competition' started by mikewiz, 19 Aug 2016.

  1. Is this a rule that say on your first game you get really crappy pieces and you want to power off NES to reset RNG... can you do that?

    I think you should be able to do it at least one time in your match.

    Mike
     
  2. would that change the likelihood of getting better pieces next time? if so, wouldnt it represent n advantage that the other player didnt have? also, if resetting the machine increased the likelihood of getting fairer piece distribution then why dont we do it every game? i am a little confused obviously lol. :)
     
  3. I've never heard of this thing. Having heard a few things about how the RNG picks numbers in NES Tetris, I don't see how it makes sense either.
     
  4. Well, there are some things that don't reset between games, but the actual impact is up for debate.

    1) The "pieces generated" counter is maintained. It can hold a value 0-255, so the thing flips over 2~3 times a game anyway. Moreover, if you look at its actual role in the calculation (it just gets added to the initial piece roll and doesn't affect the re-roll that happens if duplicate/dummy come up), the meaningful range for the value is 0-7 -- this thing is flipping over hundreds and hundreds of times a game anyway.
    2) The main randomizer value. It visits around 32767 values (I might be off by one). Since advances once per frame (sometimes two since the dupe/dummy roll rule consumes another value but we'll ignore that for a moment), that means it flips over in around 32767 / ~60 frames per second / 60 seconds ~= 9.11 minutes. So, you're liable to flip the randomizer value at least once per game. However, when you're on the level select screen, it consumes randomizer values stupidly fast because Reasons (read: either B-Type Height something or scrapped 2P mode code, I don't know which) and only visits around 7500 of the states, meaning it whips through everything in around two minutes. You could question where in the cycle you start the game, but my guess is that the specific piece placement timings are going to be the more critical component; the exact moment you press Down (or where you put the pieces in Level 19 play when you can't otherwise affect the timing without a Pause) could mean the difference between an I-piece and an S, so you're pretty much always just throwing darts at a spinning wheel no matter what.

    Further research:
    * You could argue that since the "pieces generated" count doesn't reset and isn't spinning between games that there could be some perceptible effect on the first two pieces generated based on wherever the last game left it. Again, I kind of doubt it will make a considerable since it's just simple addition of a value 0-7 in the end. This is calculable to prove one way or another but I'd have to go dig up my crappy randomizer research script again.
    * Since the state of the previous piece has some influence in the next (triggers a re-roll if it's a duplicate, and the "piece default rotation id" of the previous piece -- it's a value from 2-18, not to be confused with your usual 0-6 piece ids that come off the dice rolls -- gets added to the value of the re-roll step. So, maybe you could try to argue some form of persistent convolution between the "piece generation" count and how it aligns with the sequence of previous pieces dealt. Again, I find it somewhat suspect that there could be a continuous detrimental effect due to this alignment considering the extreme volatility of the rest of the randomization mechanics.
    * Similarly, you might argue that players on opposite ends of that 2 minute spin cycle on the level select screen could result in one player starting in the middle of a hypothetical "good" pocket of randomizer values while the other player has to fight with the supposedly "bad" pocket for a couple of minutes. I still point to how caustic the randomization process is in this case, but maybe you could formulate some sort of stilted dynamic programming solution to examine the likelihood of certain randomizer states to be visited in succession and figure out some rough approximation of the local probabilities.
    * Even if the above investigation yield some fruit, who is to say the best players haven't unconciously developed some improbable, uncanny ability to press Down at the times that will cause I-pieces to appear more often in spite of "bad RNG?"
     
  5. From what Alex knows and what this guys says:

    http://meatfighter.com/nintendotetrisai

    There's lot of stuff that determines the next piece.

    But from what I read... powering off the NES (not just reseting) provides the RNG seed.

    So powering off the NES might help or it might hurt. However most of the time it seems when I play my first game I usually do pretty well and get good pieces. So maybe just a mental thing for me.... but I would like the option. (at least once)

    >* Even if the above investigation yield some fruit, who is to say the best players haven't >unconciously developed some improbable, uncanny ability to press Down at the times that will >cause I-pieces to appear more often in spite of "bad RNG?"

    I try to use my brain powers when I need a I piece, but the force is weak with this one. haha
     
  6. It's unfortunate that you can't see the piece distribution from last year's competition because it only shows next piece, I piece count, and I piece drought. It would be nice if it could somehow be added to other side of screen.
     
  7. So did anybody look at the new rules on tetrischampionship.com?

    So for Sunday's main event head-to-head matches this implies that each player will get same seed and pieces?

    "4. The players will set their games to the rolled sequence and the agreed upon level and push start simultaneously at the prompt of the referee. Referees will settle any disputes."
     

Share This Page