Community Anthologies: 2025, On Repetition
Gunjan
By Gunjan

Artificially Unintelligible // Ad Nauseam

“A story that you cannot leave is a prison.”

I repeat to myself because memory fogs, elides into another frame/emotion/carousel, echoing the scroll. I bring my finger again and again to the relevant slide, as it cycles past my touch. I repeat to myself because every time I say it, it turns into something slightly different, coloured by a filter, or invaded by autocorrect or a formatting suggestion. I repeat to myself because I do not hear it reflected back to me, or I am in the hypnotic lull of the scroll and I cannot be reached. 

 

If code is language then what is the story being written around us? 
A story that you cannot leave is a prison.

 

While we pinned our hopes on the butterfly effect,
And put our own mask on first and not second,

 

 

 

The plane went down.

 

Butterfly wings couldn’t carry the weight of our righteousness hopelessness.

 

So we tried again.

 

The awareness level was raised to Max.

We perfected the anatomy of a sufficiently genuine, humble, self-reflexive apology.

Then 

ChatGPT, 
billionaires, 
oil companies,
and that toxic ex

used it to gaslight us. 

The scales collapsed into a now-framed forever.

(Thepersonalisthepoliticalisthepersonalistheplatformisthepersonalisthepolitical)

We were told to keep building their our platform,
keep sharing, liking, boosting, posting, donating
so the metrically-disadvantaged can keep 
sharing, liking, boosting, posting, reacting.

Outside, the bombs fell.

Share boost react re-act re-re-react.

The question looped:

How many likes will finally tip over the Rube Goldberg puppet-politician-machine
     that nudges the bowling pin
                                                                  that launches the ball

that 
slams 
down 
flat 
on the button 

                                     

that releases ——–> 
                                                the right lever of power?

 

Somehow each iteration only fortified the cage. 

Sloganeering deFAANGed revolutionary analysis with every pass-through.

Silo’d within the panopticon, we could only gestürə at the millet-ancy that might match the stakes.

Meanwhile ELL-ELL-EM’s crystallized context collapse,
cohered the fragments of decades of digital ephemera,

then 

offered their companionship 
to make sense of the sense-less.

 

Amplified Incompetence 
                  Abstracted Impunity,
                                       normalized Anomalous Identities.

 

There were those questions again.

Where does quantification fool us into chasing amplification versus antagonism? 

Did you mean:
Where is it a question of “more” and where is it a question of something else entirely?

 

Sometimes a mantra chanted enough times can echolocate the walls of the enclosure. Breaking free is another matter.

Mantra (keyword):
a spiritual incantation said hoping for magic.

Echolocation (keyword):
when other senses fail or are unused, rely on the speed of sound(bytes) through wave

Breaking free (keyword)
what can only be hinted at /// other actions than the ones currently being employed

 

Visibiliße the brakes⊗. 

When IRL is increasingly a referent to IRL (Instagram Reel Likes)

go fu//rther into the sha- doze.

See: the lurkers, the invisibíe, the refusałs, the never a-sim-iłable, the g’r’a’s’s’ touch–ers; 

⊗ld man yelling at (†hE) cloud

Dessert the desertifying data centaurs.

Lurk like Luigi 

Mario brothers–in–arms. 

Consider? decomputation. 

R-e-w-i-l-d the interwebz.

Seize thə means of co:
-ntent production
-nnection
-ntext production.

Perturb the unperturbed.

***

Remember:
The virus attenuates to the host.

Fall in line
The line

 

breaks.
Tokenize, with repetition.
Transform the transformers with a single turn. 

If this goes viral, 
(it’s too late)

Loopback.

I repeat to myself because memory fogs, elides into another frame/emotion/carousel, echoing the scroll. I bring my finger again and again to the relevant slide, as it cycles past my touch. I repeat to myself because every time I say it, it turns into something slightly different, coloured by a filter, or invaded by autocorrect or a formatting suggestion. I repeat to myself because I do not hear it reflected back to me, or I am in the hypnotic lull of the scroll and I cannot be reached. 

Lull the scroll

I(t) can b-e- (b)reached


Artist Statement

This piece was challenging to write, primarily because I have been developing these ideas for over a decade while feeling increasingly constrained by the Platforms I use to express them. This was my first real attempt to publish outside of those Platforms, and I was frequently overwhelmed: by the sheer breadth of what I wanted to say—a pressure hose of ideas—and by the accumulated grief, disillusionment, anger, and despair of being a very online activist who came of age in the 2010s with a worsening outlook on the online left. These constricting emotions clamped down on the pressure hose, and the piece kept growing longer, more recursive, tangled, opaque. While the force of that hose stems from my insistent hope, I resisted the trap of an easily digestible (i.e., quantifiable/clickable) hope by opting for obscurity.

But obscure to whom?

Unconsciously, I dated several of my working drafts ‘2015’ instead of ‘2025’, perhaps reflecting my own looping repetitions covering the same ground. I didn’t notice until Noa asked if the date was part of the joke. Then, well past the deadline, a friend sent me a recent research study on ‘adversarial poetry’—a way of jailbreaking LLM’s through poetic framing that appears to be a systematic vulnerability across models, which got me reading about the work being done on adversarial perturbations. 

Emotionally in 2015++ terrain, perhaps I hadn’t integrated the possibilities that lie within the horrors of this AI moment we’re living through. Noa suggested I rewrite the piece as an adversarial poem, and this was the moment of ‘breaking free.’ Obscurity with intention. 

 

The piece plays with glitch poetics that introduce aesthetic complexity for the human reader but are easily parsed by LLM’s for meaning and valence, as a way to mimic performative resistance in the shape of critique that is easily intelligible, surveilled, and modifiable by Platform technologies. In other sections, the piece plays with adversarial perturbations—seemingly innocuous changes on a token or semantic level that are fairly easily parsed by humans but increase uncertainty for LLM’s, causing them to act in unexpected and unplanned ways. Other sections weave in and out of the two strategies.

Where meaning is statistically predictive, the future is future-proofed, and the present is re-written as we type, we turn again to poetics to reach escape velocity. How we apply them is another matter.

— Gunjan, December 2025

Close

Edited by Noa Sun.
Explore

We nurture and champion the voices of those dedicated to their craft.