Ong observed that residues of the displaced medium survive roughly three generations before collapsing into ruins. The first post-transition generation retains the old knowledge as lived experience—these are the people who learned under the old conditions and carry, in their cognitive architectures, the capabilities that the old medium produced. They are fluent in both the old and new media, operating as bridges. The second generation inherits the residue as fading memory—transmitted not through direct experience but through the stories, heuristics, and pedagogical practices of the first generation. They know the old medium existed, may have studied it briefly, but do not possess the embodied understanding that sustained practice produced. The third generation has no access to the residue at all; the old medium is archaeologically interesting but functionally incomprehensible, belonging to a world whose categories have been replaced. The knowledge has become arcane. This three-generation arc governed the decay of oral residues after literacy, manuscript residues after print, and analog residues after digitization. For AI, the timeline may compress dramatically—decades to years, years to months—because the medium is changing faster than institutions can adapt.
Ong documented the pattern through the history of European literacy. The oral-formulaic tradition persisted as a residue in medieval epic poetry (Beowulf, the Chanson de Roland), taught and performed by bards who knew writing existed but who preserved oral compositional methods. By the Renaissance, the residue had thinned to stylistic echoes—formulaic phrases in print poetry, oral rhythms in prose. By the Enlightenment, the knowledge of how oral composition worked had collapsed entirely. The bard's method was a ruin, studied by philologists but unrecoverable as practice.
The computing case Segal documents follows the pattern exactly. First-generation engineers (Segal's cohort) learned Assembler, felt the hardware, and carry that knowledge into higher abstractions as 'architectural intuition.' Second-generation engineers know Assembler exists, may have studied it in school, recognize that 'low-level understanding matters'—but lack the embodied computational consciousness that years of machine-code practice produced. Third-generation engineers (entering the field now) learn Python or JavaScript as first languages, abstract from day one, and experience computation as high-level operations whose hardware substrate is invisible. The knowledge Segal romanticizes is fading from experience to hearsay to incomprehensibility, compressed into a timeline measured in years rather than centuries.
The Ong volume's most practical contribution is the warning: if you want to preserve a residue, you have approximately one generation to document it before the knowledge becomes untransmittable. The first-generation practitioners—those who learned under the old conditions—are the only witnesses who can articulate what the new medium does not reproduce. Their testimony will be partial, conditioned by nostalgia, shaped by the new medium they are trying to describe the old one through. But it is the only testimony available. Once the first generation is gone, the residue becomes a ruin, and the ruin is permanent.
Ong developed the timeline empirically, by observing how long oral residues persisted after literacy. The pattern was consistent: roughly three generations from lived experience to historical curiosity. He proposed the timeline as a structural feature of media transitions rather than a contingent fact about specific cases. The AI moment tests whether the timeline holds when the medium changes at digital speed.
Generation one: lived experience. Practitioners fluent in both media carry the old knowledge as embodied capability, operating as bridges between worlds.
Generation two: hearsay. Knowledge inherited from elders but not directly experienced—recognized as valuable but not fully possessed.
Generation three: incomprehensibility. The old medium belongs to a world whose categories have been replaced; the knowledge is arcane, not recoverable.
Compression in the AI age. The three-generation arc that unfolded across centuries may compress to years—residues dying before transmission is complete.