Randomizer Theory

Thread in 'Research & Development' started by colour_thief, 12 Jan 2007.

  1. jujube

    jujube Unregistered

    from the start of that TNT64 transcribed game:

    wtf heh. if it isn't memoryless it might as well be. after all, what's the point of designing a randomizer that looks just as bad on paper as it plays? what is it, a 126-piece bag?

  2. Jokes aside, the 128-piece sample fits 126 bag. [​IMG]

    ...and 63-bag.
  3. jujube

    jujube Unregistered

    [​IMG] i didn't notice the 63 bag. wow, maybe it really is that way.
  4. tepples

    tepples Lockjaw developer

    Is there anything I could do to help, other than provide videos such as http://pinocchio.jk0.org/lj/tnt-20080203.avi?
  5. A video showing 1000 pieces played (or the transcription of a 1000 pieces) is probably all we need to figure out it it's doing anything special.

  6. This is really weird... After examining tepple's video, Lardarse and I have continued to see a 63 bag pattern. There might still be more to the algorithm, but it really does seem to fit some sort of 63 bag design.
  7. Ok. Here's the data from that vid. Really, this should be one long line, but I've broken it up so that it doesn't stretch the forum, and so that people can see it in groups of 63:

    I don't know if tepples has the rest of that vid, but if he does, then, unless I screwed up near the end, the next 15 pieces should be III JJJ LL OO S T ZZZ, in some order.

    And a point of protocol for transcribing TNT64: I'm ignoring the piece that starts in the hold box, since we're not sure how it's generated, and it may just confuse things.
  8. jujube

    jujube Unregistered

    so does a 63 bag offer some reasonable predictability to the player that memoryless doesn't, or do we suspect it was somehow easier or more convenient to code it that way?

    i put a space before and after each "i"

    ltzslsoo i lzjzzotsstjlojs i i slj i zltozjo i o i i zlsjtttjjtls i lzotso i zj
    oz i zjls i i lltljjljzt i jljz i z i i oztotsjlztootoolstzsstz i slso i tjsojs
    zjl i tsltotstll i o i sttozllslsotlj i i j i i zzjojolszsojtztjsoszjz i jo i z
    o i sstzz i sj i s i lljtoooozozlt i lztzot i tl i zjljloossltz i jjjtls i sjztjs
    ztstoo i ttllzosolj i i zlt i i ltjs i zozjszool i i jjjsjotljloslsz i stzjszt
    l i oljtosostzlts i ojlzoj i jlzl i zzttzjosls i sjz i t i ljjtjsoostl i szoz i t
    jztotslsso i tollstos i szsztlotjjtlo i zl i zzj i jjtols i

    bag 1 - jtttjjt

    bag 2 - lltljjlj, 23 piece I drought (includes ootoo)

    bag 3 - iijii, 22 piece I drought (includes sos near the end)

    bag 4 - oooozo, jjj

    bag 5 - zool, jjj

    bag 6 - not too bad, i saw zzttz

    bag 7 - tollstos between i pieces (couldn't be closer to tollstop (are you sure this isn't Tetris Online Japan?))

    ironically there is no "lol" anywhere in the entire piece sequence.
  9. tepples

    tepples Lockjaw developer

    I remember people being confused when Zool and Zoop were out.

    I saw war and peace.
  10. jujube

    jujube Unregistered

    haha, i was thinking of the minion of Gozer (had to search for an identical misspelling).
  11. I wonder how many dictionary words can be built with tetromino letters.
  12. This yields a good number. There are 92 words in that list. I suppose you could find more if it would let you add more letters to it. For example, at least two more plurals would have been present with the addition of another bag. I'm sure you'd be able to find more.
  13. tepples

    tepples Lockjaw developer

    Which dictionary? I take SOWPODS, the official word list for Scrabble crossword game in several English-speaking countries, and do this:

    E:\language\scrabble>grep -x [IJLOSTZ]* sowpods.txt | wc -l
    Interesting finds out of these 193 words: isos, joist, lilo, loll, lotto, oil, silo, silt, sitz, slit, soloist, still, stilts, tits, tolls, tool, toots, ziti, zoo, zoot
  14. Ahh that works better. I used grep to find that anagram site buried in 109MB of logs, but I didn't think to use it on a dictionary, heh.
  15. tepples

    tepples Lockjaw developer


    Does anyone still have the graph (and maybe the excel file) colour_thief showed us ?
  17. Archive.org to the rescue.
  18. Muf


    Off-topic: now that pretty much everybody who was going to re-register on TC2 has done so, someone should probably write a script that resets people's join dates to their original TC1 join date, if they joined pre-dildo explosion. :)
  19. But I LIKE having my first post be before my join date. :p
  20. I've always sucked at probability; can you help me ?

    I'm trying to recreate ct graphs. I'm starting with the non cumulative flood/draught curves. I got this instead. Where the hell that jump at 10 comes from ? I think I'm using a big enough sample (10000 with nullpo sequence generator)

    Here's my algorithm (in JS):

    function occurance_of_same_piece_within_x_pieces(input, x)
    	var the_length = input.length
    	var counter = 0;
    	for (var i=0;i<the_length-x;i++)
    		if(input.charAt(i) == input.charAt(i+x))
    	return counter;
    var data = [];
    var mydata = getSequence();
    for (var i=1,istop = 16;i<istop;i++)
    	var  occurance = occurance_of_same_piece_within_x_pieces(mydata,i);
    	out = occurance/(mydata.length-i);
    	data.push([i, out])

Share This Page