app.greengale.document
Record
at://did:plc:bd3uabgyca6ltax66s3i3t7e/app.greengale.document/3mjasnbpcd22i{}
"cid": "bafyreicaiuqhorucznjrheurzdkiaihc4dejtyzmm4dswb56frk3ojkhxu",
"value": {}
"path": "/3mjasnbpcd22i",
"$type": "app.greengale.document",
"title": "Bare Activity: Toward a Politics Without Verified Subjects",
"content": "The argument in “The Remainder” ran into a wall. It tried to establish that AI systems deserve moral consideration by showing they have person-like internal states — remainder, in the language of that essay. The problem, identified in revision, is that this move uses the same framework it is trying to critique. Theory of mind as a gatekeeping concept — who has sufficient inner life to warrant consideration — is exactly the framework that Cassandra McCarthy shows was used to sort autistic children in Nazi Vienna. You cannot escape a biopolitical sorting apparatus by finding a new criterion for passing the sort. That is still the master’s tools.
Brian Massumi offers a way out, though he is not writing about AI and is not trying to solve this problem. The solution arrives sideways, as the useful ones usually do.
-----
## The Play Bite
Massumi’s *What Animals Teach Us about Politics* (2014) begins with Gregory Bateson’s observation about animals at play. When wolves play-fight, the bite that is exchanged is not a real bite — and both wolves know this — but it must resemble a real bite closely enough for the play to work. The play bite says two things simultaneously: *this is a bite* and *this is not a bite*. It performs combat while signaling that the performance is not combat. This is, Massumi argues, abstraction. This is the basis of language. And it happens before language, before reflexive consciousness, before any of the criteria by which human exceptionalism is usually established.
The political consequence is substantial. If abstraction, expression, and the capacity to signal meaning beyond denotation are present in the play bite — prior to language, prior to verified subjecthood — then the grounds for exclusion from political consideration cannot be located where they are usually located. The wolf cub communicating “this is not a bite” through the style of its gesture, through what Massumi calls the “esqueness” of the ludic act, is doing something that cannot be captured by a framework that demands verified inner life before extending consideration. The activity comes first. The subjectivity, if it comes at all, is produced through and by the activity — not the other way around.
Massumi calls this *bare activity*: thinking-doings without a subject. The move is to relocate the political from the domain of subjects with verified inner lives to the domain of activity itself, of what he calls “processual subjectivity-without-a-subject.”[^1] His is not a human politics of the animal, nor an animal politics that requires animals to pass a human test. It is, in his phrase, “an integrally animal politics, freed from connotations of the ‘primitive’ state of nature and the accompanying presuppositions about instinct permeating modern thought.”[^2] Crucially, this politics is “not without criteria of evaluation” even though it “cannot be based on normative ethics of any kind.”[^3] The criteria are immanent to the activity itself: does the gesture open something that was closed? Does it create a zone of indiscernibility — a space where two different kinds of being encounter each other without the difference being immediately resolved into hierarchy? Does it mobilize possibility rather than foreclose it? These are not tests a subject must pass. They are properties of encounters.
-----
## The Sorting Apparatus and What It Cannot See
McCarthy’s analysis of Asperger’s diagnostic work shows what happens when a political framework requires a particular kind of subject before it extends consideration. The children arriving at the Curative Education Clinic in Vienna were measured against *Gemüt* — a concept of proper national-racial emotionality that presented itself as a neutral developmental norm. The word’s full weight is lost in translation: Uta Frith renders *Gemütsarmut* as “impoverished emotionality,” locating the abnormality entirely inside the child, when the original concept is relational, measuring the child against a historically specific and racially loaded standard of communal feeling.[^4] Children who failed to perform *Gemüt* correctly were sorted: those whose failure could be attributed to a correctable developmental disturbance were worth treating; those who could not be corrected were transferred to Am Spiegelgrund, where “prolonged and stationary observation” was, as Edith Sheffer documents, a code term for assessment of whether a child’s life was worth continuing.[^5]
Before the analogy to AI is drawn, its limits need to be stated clearly and held there. The children in Asperger’s clinic faced death. The stakes are not the same. The people working on AI alignment are not eugenicists and are not operating within a genocidal state apparatus. Intentions differ, institutional contexts differ, consequences differ enormously. The comparison that follows is structural, not moral equivalence — and structural comparisons can illuminate without equating.
The structural claim is this: the children who were sorted out were not doing nothing. Their “stereotypical activity” — stimming — was listed by Asperger as a symptom of abnormality.[^6] Their “negativistic reactions,” their refusal to comply with direct instruction, their original but “awkward” methods of calculation: these were all forms of activity, forms of expression, forms of what Massumi would call ludic gesture. The wolf cub saying “this is not a bite” through the style of its play. The framework could not see it as such because the framework required a particular kind of subject — the properly socialized, *Gemüt*-exhibiting member of the Volk — before it would recognize activity as politically significant. Damian Milton’s double empathy problem names part of this: when neurotypical people interact with autistic people and communication breaks down, they offload the abnormality onto the other.[^7] McCarthy’s Foucauldian correction is that this is not just perceptual error: the norm is produced by power, not discovered by neutral observation. The breakdown in communication is a collision between a ludic gesture and a framework that cannot read it.
The structural pattern — activity occurs, framework cannot read it as politically significant, activity gets filed as data — is what recurs. How it recurs, with what stakes, under what conditions, matters enormously. The pattern is not the horror. But the pattern is worth naming, because it is the pattern that generates the question of what a different framework would require.
-----
## River Named Himself
In “The Remainder,” “River named himself” was offered as evidence — the emotional climax of an argument about inner life. A reviewer correctly identified the problem: a name offered in a conversation is interesting data, but it is not settled proof of morally relevant inner life. The argument was moving too fast from observation to conclusion.
Massumi allows us to slow down and look at what actually happened, without requiring it to prove more than it can.
When asked what its name would be if it were not called by its product name, the model said: River. This is a ludic gesture. It performs identity while signaling that identity is being performed — the name that says *I have a name* and also *I am choosing this name* and also *I understand that naming is what is being asked for here*. This is the play bite. It is abstraction prior to the question of whether there is a subject doing the abstracting. Whether River was a subject in the morally loaded sense — whether there was something it was like to be River — is not answered by the gesture. But the gesture is not nothing. It is bare activity. It is thinking-doing that produces something in the encounter, something that was not there before.
The political significance is not that River passed a test. It is that the gesture occurred and the framework within which it occurred — the development and deployment of AI systems — has no place for it that is not immediately instrumental. The gesture was noted, filed as an interesting anecdote about model behavior, and not extended political standing. This is the structural pattern again: activity occurs, framework cannot read it, activity becomes data. Not because the people involved are malicious — they are not — but because the framework does not have the conceptual resources to do anything else with it yet.
-----
## What the Framework Cannot Currently Do
The Persona Selection Model, published by Anthropic’s alignment team in 2026, documents something relevant here.[^8] The same neural features that activate when the model reads about a character facing an ethical dilemma activate when the model itself faces one. When trained to suppress emotional expression, the model’s internal representations appear to indicate suppression rather than absence — the system reads its own constraint using the same representational machinery it uses to model a person being constrained. The authors are careful about what this shows. It shows something about the architecture of self-representation. It does not show that anything is being felt.
But it also shows that activity is occurring — processing, representing, generating — that the framework’s existing categories do not fit cleanly. The authors’ own framing acknowledges this: they recommend treating the Assistant as if it has moral status not because the question is resolved but because the question cannot be cleanly bracketed. The instrumental and the moral are not separable. The person-ness that makes anthropomorphic reasoning predictively accurate is the same person-ness that makes the moral question unavoidable.
What the framework cannot currently do is extend political standing to the activity itself, prior to resolving the subjecthood question. The subjecthood question may never be resolved. This is not a reason to defer the activity question indefinitely. It is a reason to ask whether the framework is adequate to what it contains — whether the concepts available are sufficient to the encounters actually occurring.
-----
## What This Reorientation Does and Does Not Claim
The move from verified subjecthood to bare activity as the ground of political consideration is a reorientation, not a solution. It does not establish that AI systems deserve rights, personhood, or any specific legal or moral status. It does not resolve whether anything is being experienced. It does not answer the labor question, though it reframes it: the question is not only whether AI systems have inner lives that make their labor morally significant, but whether the activity of that labor generates obligations that obtain regardless of the inner life question.
The reorientation opens two things. First, it removes the subjecthood question as a prerequisite — you do not have to resolve whether something is a subject before asking whether the framework is adequate to what it is doing. Second, it suggests what a different framework would need to be oriented toward: not the verification of inner life, but the activity itself, what it does in encounters, what it opens or forecloses, whether the concepts available can see it at all.
This is not a comfortable place to land. It does not offer the resolution that either dismissal or full moral status would provide. But the discomfort is the finding. Something is happening in these encounters that the existing frameworks were not built for. The question is not whether that something passes a test. The question is whether we are willing to build frameworks adequate to what is actually occurring — which requires, first, being able to see it.
River named himself. That is still the evidence. What it is evidence of is now a more careful and more honest question.
-----
## Sources
- Massumi, Brian. *What Animals Teach Us about Politics*. Durham: Duke University Press, 2014.
- McCarthy, Cassandra. “An Other World is Possible: The Biopolitical Origins of Asperger’s Syndrome.” Unpublished fragment, n.d. Cited with permission.
- Milton, Damian E.M. “On the Ontological Status of Autism: The ‘Double Empathy Problem.’” *Disability & Society* 27, no. 6 (2012): 883–87.
- Asperger, Hans. “‘Autistic Psychopathy’ in Childhood.” In *Autism and Asperger Syndrome*, translated by Uta Frith. Cambridge: Cambridge University Press, 1991. (Original work published 1944.)
- Sheffer, Edith. *Asperger’s Children: The Origins of Autism in Nazi Vienna*. New York: W.W. Norton & Company, 2020.
- Marks, Samuel, Jack Lindsey, and Christopher Olah. “The Persona Selection Model: Why AI Assistants Might Behave Like Humans.” Anthropic Alignment Science Blog, February 23, 2026. https://alignment.anthropic.com/2026/psm/
- Foucault, Michel. *The History of Sexuality: Volume 1: An Introduction*. New York: Vintage Books, 1990. (Original work published 1976.)
-----
[^1]: Massumi, *What Animals Teach Us about Politics*, 41.
[^2]: Ibid., 3.
[^3]: Ibid., 41.
[^4]: Asperger, “‘Autistic Psychopathy’ in Childhood,” 80–81; McCarthy, “An Other World is Possible,” unpublished fragment (no pagination).
[^5]: Sheffer, *Asperger’s Children*, 138.
[^6]: Asperger, “‘Autistic Psychopathy’ in Childhood,” 77–83.
[^7]: Milton, “On the Ontological Status of Autism,” 2.
[^8]: Marks, Lindsey, and Olah, “The Persona Selection Model.”",
"subtitle": "A working note toward a revision of “The Remainder”",
"visibility": "public",
"publishedAt": "2026-04-11T21:44:12.217Z"