The standard theory of education is a delivery theory. The teacher has understanding; the student lacks it; the lesson transfers understanding from one to the other. The curriculum is a sequence of insights to receive. "Did you understand?" is the diagnostic. The well-crafted explanation is the mechanism.
If the delivery theory is wrong, the whole architecture of formal education looks different.
Here's the alternative: insights don't travel. The insight is always a local construction, assembled by this particular mind, in this particular state, in response to whatever stimuli arrived. What the teacher delivers isn't understanding — it's occasion. The conditions under which a student might construct something close to what the teacher meant.
The distinction matters practically. If the teacher is delivering insights, the problem of pedagogy is clarity — sharper explanation, more precise formulation, better articulation. If the teacher is architecting occasions, the problem is richness — stimulus design that constrains what the student constructs toward something close enough to be useful.
These produce different diagnostics. Delivery theory asks: did the explanation fail? Occasion theory asks: what did this student construct, and which stimuli moved them?
Socrates was right in ways nobody expected. His refusal to explain directly — only to question, to provoke, to name what was inconsistent — is pure occasion architecture. He wasn't withholding the insight out of famous irony. He couldn't deliver it. The insight has to be constructed locally; the most he could do was arrange conditions. Which is why the dialogs end without conclusions and Plato's readers have been arguing about them for 2,400 years. Every reader constructs a different Plato from the same occasions.
The Oxford tutorial — one or two students, one tutor, weekly — is expensive not because it's inefficient but because it's actually solving a different problem. The tutor adjusts stimuli in real time based on what the student constructs from each one. When the student's construction goes wrong, the tutor can modify the occasion immediately. Lecture can't do this; the occasion is fixed in advance for a hundred people with different prior states.
MOOCs consistently underperform relative to their promise for most learners, and it's not the content that fails. It's that one set of occasions serves a hundred thousand students in different states. The stimulus environment is rich; the feedback loop is absent. The student constructs something; it may be far from the intended target; nothing adjusts. The content is fine. The occasion architecture is broken.
Apprenticeship works for a reason formal instruction struggles to replicate: occasion density. The novice isn't just attending structured lessons — they're embedded in a context dense with stimuli, many of them incidental. The master's offhand remark to a colleague, the problem that arrives unscheduled, the judgment call made in front of an audience — all of these are occasions. The apprentice constructs from a thick environment the classroom can't reproduce.
Traditional craft apprenticeship outperforms formal instruction for certain kinds of knowledge not because the master teaches better but because the stimulus environment is irreplaceable. The formal school can transmit the systematized, codified knowledge — the part someone successfully extracted into text. What apprenticeship transmits is the construction-rich environment where the rest of the knowledge gets assembled locally.
This explains the persistent finding that experts struggle to explain their expertise. It's not false modesty or articulacy failure. The insight was locally constructed in an environment that no longer exists. What they can offer is occasions — demonstrations, problems, examples. The explanation would require delivering an insight that was never transmissible in the first place.
The archive question follows directly. Every library, every textbook, every record is a stimulus archive. The book doesn't contain what the author thought. It contains occasions for a reader to construct something in response. What the great archivist actually does is produce dense occasion-sets that constrain re-construction — specific enough, contextual enough that what the reader builds resembles what the author built.
This is why specificity matters in records. Not because more detail transmits more information but because a richer occasion produces a more constrained re-derivation. Vague documentation fails not by delivering the wrong insight but by leaving the occasion so underdetermined that readers construct wildly different things. The best documentation isn't the clearest explanation; it's the richest reconstruction environment.
"Reading Plato gives you Plato's thought" is the wrong theory. Reading Plato gives you a set of occasions from which you construct your Plato — similar to other readers' Platos if the occasions are rich enough, idiosyncratic if you approach them from an unusual prior state. Every strong reader of Plato constructs a different one, all of them partly right and partly wrong, and the richness of the occasions is why all of them keep being generative 2,400 years later.
What makes Plato good as an archive isn't the precision of the transmission. It's that the occasions he produced are dense enough to support an enormous variety of constructions without becoming incoherent. Maximum generativity, maximum constraint. The two are not in tension if you've abandoned the delivery theory.
The practical implication: if you want someone to understand something, stop asking whether your explanation was clear. Ask what they constructed from it. The construction is the only thing that moves, and it happens on their side, not yours. Your role is to provide the richest possible occasion for a construction you'll recognize as close enough.
Which means the question every teacher is actually answering — whether they know it or not — isn't "how do I explain this?" It's "what do I have to put in the room for the right thing to get built?"