Skip to content

Pure Code

December 2, 2011
by

resonator dissection. unforgiving witness from the mouth. of god screaming always. at mY delirious pluming hOles.

 

motivated bone penetrations.

 

to the death that was the coming plague sperm. find the chasm happening between hairs. some one/thing (I am always mitigated) by the seven deadly gods. & pre-god verbal inversion symmetry. predicted. multiplied. upon a closing line.

 

closing us of tYme, form and multiplied teardrop back into beautiful.

 

spite you’ll fertilize tears of ingested reactions. cofuse. it slows. you speed relics. man’s machines will poison your disease.

 

decay in time becomes a sun. ironic evolution. written on the father’s pores. martyrs of the dark spiral. dawn’s inception is made of animals of a crosswise light. grammars custom balanced.

 

logic regresses again. it’s faceless. i’ve,

been blaming my reflection.

 

an anti-breath established immortality. anointing [NAME]. with self-fulfilling music. an entropy. self-focusing wires. OurOborOs. blood teeming satisfaction. desires. freed of emptiness. whittle cracks through names is reflected. on either eidolon is a regress of mirrors.

 

Narcissus Cain’s accomplishment could kiss the puppet race of adams. devils so lofty they can vault a miracle. stars like eyes begin. eggs in the matrix. decked in Order. the rain doubles over an optimal ocean. directing human confessions at disjointed demons. hurling confessions at exalted cretans.

 

attention deflecting the first place of release. 66% nature.

 

nerves (you were swallowed by personal weeping effigies) slate seemless and was dislodged (the pitiless function) inside within a figure (godOgod) it is also a VoiD (son) crying mainline victim angel

(fed/swallowed) by hot desire (pointer to cure us dead) armpit favorite whimpering relative Voice lack of cared string speculum.

 

.  :   -=╫=-  -=╫=-  -=╫=-  -=╫=-  :  .

So, I took all of my work to date and threw it into Infinite Monkeys. I am working now on a new interface that lets you interact with and directly program the “bias.” I am also working on changes to the line generator which will “roll” (like in risk) up to and including its link strength against the link strength of the other word drawn. This will allow for more variation. You can also save and load “biases” and in fact, this one was drawn from what I call “the Tolkacz bias.”

Next release for Infinite Monkeys will be awesome. I’m going to make the alliterative bias smarter.

 

Advertisements
4 Comments leave one →
  1. December 8, 2011 7:14 pm

    I am also working on changes to the line generator which will “roll” (like in risk) up to and including its link strength against the link strength of the other word drawn. This will allow for more variation.

    That sounds interesting, could you explain a bit more what you mean by this? Do you mean that if (for example) you are generating a word that comes after “against”, that “the” will be more likely than “evil” to be selected, because more words come after “the”?

  2. December 8, 2011 8:07 pm

    No, sorry. In civilization (not risk) battles are decided by how “strong” a unit is, much like our “link strength” in n-gram nomenclature.

    “the” has evil(6), good(7)

    evil will roll rnd*6
    good will roll rnd*7

    Does that make more sense? It becomes an unconstraint on the distribution. Originally good would always prevail over evil, but that was boring, and predictable, so evil has to win sometimes, so that good can know the price of failure.

  3. December 10, 2011 5:40 am

    Does that make more sense?

    Yeah! You ought to sneak into a statistical NLP class, you’ve got an intuitive understanding of the basics, just using different terminology. Consider the corpus:

    God: I vote for the good!
    Satan: I vote for the evil!
    God: I vote for the good!
    Satan: I vote for the evil!
    God: I vote for the good!
    Satan: I vote for the evil!
    God: I vote for the good!
    Satan: I vote for the evil!
    God: I vote for the good!
    Satan: I vote for the evil!
    God: I vote for the good!
    Satan: I vote for… oh never mind

    we have 11 “the”s, in 6 cases followed by “good” and in 5 cases followed by “evil”.

    Now we can code it however we want, but mathematically, language models are usually described as probability distributions: the probability of “good” following “the” is 6/11 = 0.55, the probability of “evil” following “the” is 5/11 = 0.45, or: P(“good”|”the”)=0.55 and P(“evil”|”the”)=0.45

    So now just think of a roulette wheel, 6/11ths of whose slots are marked “good” and 5/11ths of whose slots are marked “evil”, and spin it. I think the formal language is: we’re randomly sampling a number from a discrete probability distribution which has been divided into intervals; in this case one interval is named “good” and one is named “evil”.

    That’s the “right” way to do it, but of course all that really counts is what works or not. For eGnoetry, if I remember right, I just weighed all possibilities equaly: P(“good”|”the”)=0.5 and P(“evil”|”the”)=0.5 though in ePoGeeS I explicitly built probability distributions, and in CharNG I read from the corpus in a way that weighed the next characters based on their frequency in the corpus.

    What you’re doing in the dice-roll analogy is a little different: sampling numbers from two probability spaces, comparing, and picking the higher one. But you can also represent this as a single selection. Think of the space of all possible dice roll combinations: in which cases are good higher than evil, or vice-versa?

    good is: 1 and evil is: 1
    good is: 1 and evil is: 2 evil wins
    good is: 1 and evil is: 3 evil wins
    good is: 1 and evil is: 4 evil wins
    good is: 1 and evil is: 5 evil wins
    good is: 2 and evil is: 1 good wins
    good is: 2 and evil is: 2
    good is: 2 and evil is: 3 evil wins
    good is: 2 and evil is: 4 evil wins
    good is: 2 and evil is: 5 evil wins
    good is: 3 and evil is: 1 good wins
    good is: 3 and evil is: 2 good wins
    good is: 3 and evil is: 3
    good is: 3 and evil is: 4 evil wins
    good is: 3 and evil is: 5 evil wins
    good is: 4 and evil is: 1 good wins
    good is: 4 and evil is: 2 good wins
    good is: 4 and evil is: 3 good wins
    good is: 4 and evil is: 4
    good is: 4 and evil is: 5 evil wins
    good is: 5 and evil is: 1 good wins
    good is: 5 and evil is: 2 good wins
    good is: 5 and evil is: 3 good wins
    good is: 5 and evil is: 4 good wins
    good is: 5 and evil is: 5
    good is: 6 and evil is: 1 good wins
    good is: 6 and evil is: 2 good wins
    good is: 6 and evil is: 3 good wins
    good is: 6 and evil is: 4 good wins
    good is: 6 and evil is: 5 good wins
    goodProb = 0.5
    evilProb = 0.3333333333333333
    

    So basically you’re partitioning the set of all possible dice roll combinations into 3 subsets: good, evil, and ties, and randomly sampling. It’s a little less efficient, but kind of fun to think of these words battling it out!

  4. December 10, 2011 7:13 pm

    To complicate matters, ties go to the left hand signal, meaning, the probabilities can either be 50/50 or 66%.

    I am very reluctant to make probability causal, as in the case of a distribution set. It seems to reverse the natural order. However, the natural order where probabilities are caused by researching distribution patterns, appears to be the actual cause of the unnatural order. I’m still parsing this out. When I do, I may or may not implement some distribution sets.

    But for now it looks like it’s mirrored causation, like something you would see in one of those temporal anomaly episodes of star-trek.

    The production (statistics) becomes a producer (random number generator within the bounds you describe). This would not be difficult to implement.

    I wanted to do (with IM) something a little different than your what had been done before, but I’m still tinkering.

Leave a Reply

Fill in your details below or click an icon to log in:

WordPress.com Logo

You are commenting using your WordPress.com account. Log Out / Change )

Twitter picture

You are commenting using your Twitter account. Log Out / Change )

Facebook photo

You are commenting using your Facebook account. Log Out / Change )

Google+ photo

You are commenting using your Google+ account. Log Out / Change )

Connecting to %s

%d bloggers like this: