Context Collapse
Then there is what danah boyd named context collapse: the flattening of a person’s many audiences into one. Speech that would be perfectly interpretable at the pub, among people who know your politics and your register and the running joke you are extending, becomes uninterpretable — or rather, becomes interpretable in a great many incompatible ways — when it is legible to your employer, your mother, your students and forty thousand strangers simultaneously. The famous cases of internet outrage are almost never disputes about what someone said. They are disputes about which community of interpretation has jurisdiction. The single most common protest of the person at the centre of such a storm is not I didn’t say that but that isn’t how it was meant — a plea, in effect, for a thicker description, and one which the medium has made structurally impossible to hear.
from Reading Clifford Geertz in the 21st Century: Meaning, Interpretation, and the Trouble with Culture (2026)
GPT-5.6 Sol:
The passage makes a strong Geertzian point: meaning is not carried by utterances alone; it is produced by utterances-in-context. “Context collapse” matters because digital media detach speech from precisely the contextual particulars that ordinarily make interpretation tractable: who is speaking, to whom, under what conventions, after what previous remarks, with what shared knowledge, and with what expectations about genre and tone. A joke between friends is not merely the same semantic object placed before a larger audience. Once detached from its original setting, it has become a different social object.
The connection to Geertz’s “thick description” is especially productive. The classic example is the difference between a wink and an involuntary twitch: physically similar movements can mean radically different things because the meaning resides in layers of convention and intention. Online, however, circulation tends to preserve the “movement” while stripping away those layers. A screenshot captures the sentence exceptionally well and the social situation exceptionally badly. The result is something like technologically enforced thin description: maximal fidelity to the visible sign, minimal fidelity to the interpretive world in which it originally made sense.
“Which community of interpretation has jurisdiction” is perhaps the most interesting claim here. It shifts the problem away from the familiar question, “What did this sentence really mean?” There may be no context-independent answer. Within one interpretive community, an expression can be unmistakably ironic; within another, unmistakably offensive; within a third, merely baffling. What social-media controversy often does is bring those communities into collision and then demand a single adjudication. The dispute becomes partly a jurisdictional struggle over which context counts. The speaker says, in effect, “interpret this according to the norms of the audience I was addressing”; critics answer, implicitly or explicitly, “once you spoke publicly, our norms are equally entitled to govern its meaning.”
That last disagreement reveals an important complication, though. “That isn’t how it was meant” can be a legitimate demand for thick description, but it can also be an attempt to claim excessive authority over meaning. Intention is part of context, not the entirety of it. Someone can sincerely intend a remark affectionately while using a convention that predictably reads as insulting outside an intimate group. A Geertzian approach therefore does not entail that the original speaker possesses the definitive interpretation. Thick description would require reconstructing the intentions of the speaker and the conventions of the originating group and the conventions brought by later audiences and the circumstances through which the utterance reached them.
This is where I would qualify the sentence that famous internet outrages are “almost never disputes about what someone said.” That is rhetorically effective but too categorical. Sometimes disputes genuinely concern textual facts: what was said, what was omitted from a clip, whether a quotation was altered, whether irony is linguistically plausible. More commonly, though, the literal words are unusually easy to establish precisely because the internet archives them. What remains unstable is their pragmatic force: whether they constituted a joke, threat, insult, quotation, provocation, flirtation, parody, hyperbole, or assertion. Digital media give us unprecedented certainty about the inscription while often giving us reduced certainty about the speech act.
There is also something slightly stronger than “many audiences become one” happening. They do not really become one audience. They remain multiple audiences occupying the same textual space. Your mother, employer, friends and strangers do not suddenly acquire a shared interpretive framework; they encounter the same artefact while retaining different ones. So “collapse” describes the collapse of the boundaries between audiences, not necessarily their fusion. The resulting pathology comes precisely from the fact that the audiences remain heterogeneous.
This recalls Erving Goffman as much as Geertz. Ordinary social competence depends heavily on audience segregation. We present ourselves somewhat differently to colleagues, parents, lovers, students and old friends, not necessarily because one presentation is fraudulent but because social identity is relational. The fantasy that a person has one authentic voice appropriate to every setting misunderstands how ordinary interaction works. Social media create an architectural problem: they ask users to perform before what Goffman would recognise as incompatible audiences while encouraging everyone else to treat inconsistencies between performances as revelations of hypocrisy.
There is a further asymmetry introduced by virality. Context is expensive; decontextualisation is cheap. It may take one screenshot to produce the accusation and a thousand words to reconstruct the exchange in which the screenshot originally functioned. The short version is more portable, more emotionally legible and easier to repeat. So I would soften the passage's final claim that thicker description has become “structurally impossible to hear” to something like structurally disadvantaged. Context can be restored—threads can be read, recordings watched, histories reconstructed—but the architecture of circulation systematically rewards fragments over reconstructions. The accusation travels at meme speed; the ethnography does not.
And there is an even darker implication. Once everyone understands context collapse, people can weaponise it deliberately. A hostile reader need not misunderstand an utterance. They can understand its original meaning perfectly well and nevertheless re-present it to another interpretive community under whose conventions it becomes incriminating. Context collapse then stops being an accidental feature of networked communication and becomes a technique: context laundering. The crucial act is not falsifying the quotation but changing its audience.
That makes the Geertz connection particularly apt. The problem posed by the internet is not simply that we lack context. It is that context itself has become politically and socially contestable. Participants disagree about which surrounding facts are relevant, how far outward interpretation should extend, who gets to supply those facts, and whether the original setting retains privileged authority after an utterance begins circulating. “Read it in context” sounds like a straightforward methodological injunction; online, the more difficult question is: which context?
The passage therefore points toward a broader irony. Networked media are superb machines for preserving text and remarkably poor machines for preserving situations. They remember exactly what was said while continuously rearranging the conditions under which saying it meant something. That is almost a perfect inversion of the ethnographer's task. Geertz wanted to move from observable behaviour toward the dense structures of significance that made it intelligible. The internet continually performs the reverse operation: it turns thickly situated human action into infinitely reproducible, thinly situated evidence.
