I would strike Jesus out on 3 pitches. Cumshot complation, he feels like hes drowning. Forbidden fruit in the garden of eden, of course he's taking it. And bang 5:22 (1 Thessalonians) inside corner. We're mixing body oil and holy water today with the league leader in BPB¹ and the biggest sin pit this side of the river Styx.

You're out!

Markov chain poems from PornHub and the Good Book

It's a great read! And the bibles good too, I guess. Play Ball!

Introduction

like the stars be concealed with it's riches.
and a fog who overtake it,
or who have made into level ways.
And who makes a call

and in the desert,
or from the mouth.
the sons of israel,
may swallow us whole

And those who hate hard sex
Lord, lead me forth,
Show me a fith time,
in the midst of the earth,

and set her land the rivers have spoiled.
and fucked her throat from being human,
and when they had no one to see
Show me your heart, like a whore

big ass, slobbered pussy
Show me a learned tongue,
Show hospitality to one another,
to riding and grinding like Crazy!

and one saint said to Saul:
"Why should he who is lifted up the blood,
not go down to the south
and we have Cum on Face Complation - Load Moaning asking Cum !"
for the appointed days
she closed the door on the desert,
And when you are my god,
le t there be wild lesbian sex.

Markov chains look at language like a graph and work out that, in this sentence, the word "the" is followed by the word "word" 33.33% of the time, "is" 16.17% of the time and "time" 50% of the time.

That way, if i give you the start of a sentence: "When I woke up, the", you can roll your dice and work out which word its followed by. "When I woke up, the word".

This is the tech behind phone auto-complete and the game of pressing the middle option till a sentence comes out.

For maximum accuracy, the chain needs to be trained on as large a dataset as possible, examining each sentence for the connections between words and updating its look-book.

I trained one markov chain on the bible and one on 20 thousand porn titles and made them kiss.

Method

Aggregating data

I put together a quick script to scrape through the recomendation API page by page and sift out the data. This was easy³. As for the bible, many versions are available in plain-text.

Post-processing

As well as producing a chain for each, I also reversed each line of the data and produced a chain from each of those, allowing be to run the chain backwards if I wanted to generate sequences that ended on a specific word. I imagined this would have been useful for specifically generating poems.

Some small amount of clean-up was done after testing the system, paticularly to remove rougue characters that would put the chains into a spin. Emojis were left in.

Generating chains

I used the rust crate "markov"² to generate the actual chains in a variety of ways. I tried generating based on the first word, generating based on the last word, generating based on the middle word and getting each chain to work outwards to produce some result.

Almost all of the time it would be either bible or porn and what I was looking for was a mixture between the two. It seemed like certain trigger words would allow the system to switch between the two. "a" was the most effective of these, "have", "her", "the" and "on" were also useful.

I selected my favorates and printed them out, cutting up each line and occasionally lines into smaller parts to arrange into order.

Result

Two of my favorates are above and below.

Show me your penis!
It's for my boyfriend,
Show this video to you name.
Show him my intricate laws,

Show me your heart,
Show me the tumult ceased, Paul,
cute australian kitty shows up
who sits upon the wall

Show him my creampied pussy
Show me a leaned tongue,
they will not always hold this power
Show hospitality to one another.

That's all