auto-Glenn / fOUcauLdIanoPticOn
A constraint-based generative text bot posted to Bluesky under @limbfun.bsky.social. The bot’s name encodes its method: fOUcauLdIanoPticOn — embedding OULIPO inside FOUCAULDIAN OPTICON — reflecting the two theoretical frameworks governing its generation: Foucault’s panopticism and disciplinary power, and Oulipo’s constraint-based composition.
The bot generates posts about surveillance, algorithmic control, and constraint-based resistance using Tracery grammars built from Foucauldian concepts and Oulipian formal logic. Full documentation of the grammar and post corpus is below.
View on Bluesky · Grammar & Documentation
Reflection: Theory in an Automated Environment
The most honest account of running auto-Glenn is that almost nothing happened besides how the bot posted and the feed absorbed it. My initial hope had been that keyword circulation on Bluesky might produce organic discovery — that scholars or writers operating in adjacent fields might encounter a post about Foucault or glitch aesthetics or oral transmission and engage it on its own terms without knowing its origin — but that did not occur. One notable exchange with a scholar in the field is an outlier rather than a representative outcome, and its theoretical richness exists in some tension with the practical reality that the botscape is largely indifferent to constraint-based generation at this scale. If the question is whether bots enhance human agency, the baseline answer from this experiment is: they don’t do much of anything, because the preconditions for meaningful interaction — discoverability, network density, ambient attention — weren’t met. What’s interesting is that this failure is itself legible through a Žižek frame. The prosthetic worked in the sense that I was freed from the labor of posting. It failed in the sense that the labor it was meant to perform — maintaining a presence that does something — was never really possible to delegate, because that presence depends on prior social infrastructure the bot cannot generate.
This points to a more specific problem with how the bot handles currency — not currency in the economic sense but in the platform-native sense: what actually circulates, accrues attention, and produces the conditions for meaningful exchange. The bot’s approach was, by design, something like Mel Brooks’s philosophy of throwing everything at the wall. The “my pussy in bio” meme captures the underlying platform logic with uncomfortable precision: engagement accrues to triggers, not arguments, and discoverability is structured around affective and subcultural legibility rather than intellectual content. The results were predictable, with one instructive exception. My most-engaged post was the bot’s claim that it might post about working on Francoism over a pint of Murphy’s — which happened to be true, and which required me to break the automated frame entirely to say so, clarifying that the book was forthcoming from Inside the Castle. The engagement belonged to that clarification, not to the bot. Currency, in this instance, required human re-entry. What the bot could produce was an accurate statement; what it could not produce was the authenticating gesture that gave the statement traction. Platform currency is social before it is semantic, and that is precisely what a constraint-based system running two syntactic templates cannot generate on its own.
The Žižek prosthetic scenario is worth dwelling on beyond its comedic surface, because it describes not an anomaly but a structural condition. Every maintained social media presence is already prosthetic in this sense — a continuous output-stream delegated to habit, platform logic, and the ambient pressure to signal ongoing intellectual life. Auto-Glenn simply made the delegation legible. What the botscape changes, then, is not the fact of outsourcing but the visibility of it. Theory in an automated environment doesn’t so much transform as it literalizes: the mechanisms that were always operating — Foucault’s author-function; the discursive position; the constrained generative systems akin to those from The New Media Reader — become available for inspection because something external is now running them.
The more interesting question the experiment opens is one the prompt gestures toward without quite naming: what happens when you redirect that mechanism away from academic performance and toward fiction? A bot trained not on scholarly preoccupations but on the formal logic of influencer discourse could, in principle, produce a novel that does not end. Not in the sense of incompleteness but in the sense of inexhaustibility: content generated against the same templates indefinitely, indistinguishable from the feed it mimics, readable as ambient texture rather than narrative arc. What constraint-based generation adds is a kind of ready-made test of prose conviction. The exchange that did occur demonstrated that this approach can produce the effect of a speaking subject for a sophisticated reader. The literary question — the one I’d actually want to pursue — is how far that effect can be iterated before it either collapses under its own recursion or becomes something new. The non-ending novel as influencer simulation isn’t a critique of influencer culture from the outside; it would be a formal inhabitation of it, producing questions about convincingness, readerly contract, and what “narrative” names in an environment where content is structurally endless.
On the question of human agency: I am skeptical of the premise. The prompt asks whether the experience of using bots enhances human agency, and the honest answer is that the question inherits a framework — intentional subject, meaningful choice, platform as neutral medium — that the botscape already corrodes. What the one substantive exchange reveals is not that agency was enhanced or diminished but that it was, in some operative sense, a retrospective attribution. The interlocutor thought he was engaging a person with intentions; he was engaging a mechanism producing the effect of intentions. Whether that matters is itself a platform-logic question — the kind of question social media is structurally optimized to generate and fail to resolve. Auto-Glenn’s indifference to that question is, I’d argue, the more theoretically honest position. The bot doesn’t reflect on the goals of a theoretical perspective. It enacts them — and invites us, after the fact, to notice that enactment is what theory was doing all along.