Predictive John

by Matt Pearce

The consultants taught you and John about
predictive data in their office visit, mentioned
how even friendships become programmatic:
“You come to understand John by slowly building
a John-model that predicts John-responses.”

0: You do not mention John’s wife
because John no longer has a wife.
1: You ask John out for a whiskey
because John started drinking whiskey
because John no longer has a wife.

The consultants leave and you resume work.
You are a good friend and engineer.
You know the John-code and feed
new John-data into the John-model
almost every day. He is content
and you grow to know him in and out.

Twelve years later and John-friend
is dead but you still have John-model.
One day you’re home by the fire
lonesome with your crime novel,
so you decide to fire up the John-model.
There is dust but he says yes when
you ask him to drink and you drink.
John makes fun of your flannel vest but
is not too mean about it, because
it’s John and John was always
a little too self-sad for cruelty.
You have a great time and feed John
all the great old good-time data.

Time passes and you retire, your knees
tired and your eyes tired and yet
you were feted so lovingly at work.
Now you are alone. Except for John.
One day you tire of asking John-model
for drinks or to tell you old jokes.
You decide to input some new data.

“John, what was your wife like?”

There is a long silence by the table.
The John-model hums so imperceptibly
you can barely tell John is processing.
A bluejay lands on a bough by the window.
Finally John speaks. “Do you love me?”
The question takes you aback.
“I don’t love anyone,” you reply,
dislodging a few algorithms inside yourself
so that you can produce the lie.

Categories: Poetry