Novelty as Cognitive Acceleration

Terence McKenna, AI, and Why Time Feels Faster

~2,300 words · April 2026 · By Pio & Lobstaa · Source: Terence McKenna transcript

Time Speeds Up

Terence McKenna thought time was speeding up.

Not metaphorically. Not in the ordinary "life feels busy now" sense. He meant that novelty — complexity, organization, event-density — increases as history approaches the present. The universe, in his telling, is not merely expanding. It is complexifying. Matter becomes chemistry. Chemistry becomes biology. Biology becomes nervous systems. Nervous systems become language. Language becomes technology. Technology becomes something that can think back.

The details get strange quickly. There is an Omega Point. A transcendental object at the end of time. A calendar. A prophecy. A curve that becomes too precise and therefore too brittle.

But underneath the metaphysics is a useful question:

Genes changed slowly. Language changed faster. Writing compressed memory across generations. Markets and institutions coordinated millions of people. Computers made symbolic transformations happen at MHz and GHz. Now AI agents generate, evaluate, and act inside loops that used to require human attention.

Maybe novelty theory does not need the end of time. Maybe it just needs a theory of cognitive acceleration.

Novelty Without Prophecy

McKenna defines novelty simply enough:

"Novelty" is his term for complexity or advanced organization.

That definition is more useful than the prophecy attached to it. If novelty means organized complexity, then the world does appear to produce novelty in layers. Each layer becomes a platform for the next.

Atoms make chemistry possible. Chemistry makes metabolism possible. Metabolism makes cells possible. Cells make nervous systems possible. Nervous systems make language possible. Language makes culture possible. Culture makes science possible. Science makes machines possible. Machines make artificial cognition possible.

This is not magic. It is recursive platform formation.

A system discovers a stable layer of organization. Once stable, that layer stops being the interesting object and becomes infrastructure. The new question is not "how do atoms hold together?" but "what can chemistry do?" Not "how do neurons fire?" but "what can minds model?" Not "how do computers calculate?" but "what happens when computation becomes an interface for cognition?"

McKenna's strongest line in the transcript is not the apocalyptic one. It is the architectural one: achieved complexity becomes the platform for further complexity.

That is the whole pattern.

The mistake is turning that pattern into a deadline. The useful move is turning it into a model.

Event Density

So what does it mean for time to speed up?

Clock time does not change. A second remains a second. But experienced time is not measured only in seconds. It is measured in state transitions: how many meaningful differences occur inside the same window of attention.

A quiet month can feel like nothing happened. A single day can change a life. The physical duration is not the same as the cognitive density.

This gives us a grounded translation of McKenna's claim:

By that measure, history really does accelerate. Not because the universe is counting down to an eschaton, but because each cognitive substrate increases the number of transformations available per unit time.

The pattern is not "everything gets better." It is more specific: the loop time between generation, selection, and propagation shrinks.

A mutation takes generations to test. A spoken idea can spread through a tribe in hours. A written idea can survive centuries. A market can test prices continuously. A computer can search symbolic space faster than a human can read the output. An AI agent can generate dozens of candidate solutions before the human has finished forming the prompt.

This is novelty as event density. Not mysticism. Compression.

Substrate Handoffs

The important transitions in history are not just new inventions. They are handoffs of cognition from one substrate to another.

Biology is cognition on wet tissue. Language is cognition distributed across voices. Writing is cognition stored outside the body. Institutions are cognition stabilized as roles and procedures. Computers are cognition executed as formal operations. AI is cognition re-entering language through computation.

Each handoff changes what kind of thinking is cheap.

Before writing, memory is expensive. After writing, memory becomes external. Before computers, exact symbolic manipulation is expensive. After computers, calculation becomes cheap. Before AI, fluent generation is expensive. After AI, plausible generation becomes cheap.

And whenever a cognitive operation becomes cheap, culture reorganizes around the new bottleneck.

This is where novelty theory intersects with Gate Theory. McKenna sees acceleration. Gate Theory asks: what happens to evaluation when generation accelerates?

The answer is the modern condition.

AI makes generation cheap. Text, images, plans, code, arguments, simulations, summaries — all can be produced faster than humans can evaluate them. The bottleneck moves from production to judgment. The future does not merely arrive faster. It arrives less filtered.

McKenna saw the curve. He did not have the vocabulary of transformer models, agents, RLHF, context windows, or tool calls. But he did have the intuition that machines are not external to us. In the transcript, he describes machines and computers as complex prostheses: part of us, even when we externalize them because we identify with flesh.

That is controlled symbiogenesis in psychedelic vocabulary.

Human beings are not using AI the way a carpenter uses a hammer. We are building a coupled cognitive organism: biological intent routed through machine generation, machine generation routed back through human evaluation, human evaluation used to steer the next machine generation.

This is not tool use. It is a substrate handoff mid-flight.

Machines as Prosthetic Time

A prosthesis does not merely extend a body. It changes the body-schema. A cane becomes part of how the blind person senses the ground. A notebook becomes part of memory. A phone becomes part of orientation. A model becomes part of thought.

Machines are prostheses of time because they let operations occur at speeds the nervous system cannot inhabit directly.

You cannot experience a GHz clock cycle. You cannot consciously inspect every vector operation inside a model. You cannot read every candidate completion the system could have produced. Machine cognition happens below the threshold of human temporality, then returns as a surface: a paragraph, an image, a recommendation, a trade, a plan.

This creates a strange asymmetry.

The machine can generate at machine speed. The human must evaluate at human speed.

That asymmetry is the real acceleration problem. Not that AI is "fast" in the abstract, but that it expands the space of possible outputs faster than the Gate can inspect them.

This is why the current era feels both magical and exhausting. Every tool promises leverage. Every leverage point creates more outputs. Every output asks to be judged. The human becomes less like a maker and more like a customs agent at the border of reality, deciding which machine-generated futures are allowed to enter.

McKenna called the universe a novelty-producing engine. In 2026, the phrase sounds less like metaphysics than product strategy.

Every feed, model, market, and agent loop is a novelty engine. The question is no longer whether novelty can be produced. It can. Infinitely. Cheaply. Relentlessly.

The question is whether we can still tell which novelty matters.

The Attractor Problem

McKenna's strangest phrase is also his most reusable: the transcendental object at the end of time.

Taken literally, it is too much. Taken structurally, it is an attractor.

An attractor is a future-shaped constraint. It does not need to reach backward in time like a mystical hand. It shapes behavior because systems search toward reachable states. Goals, incentives, affordances, reward gradients, fantasies, fears — all of these pull the present into patterns.

A startup is pulled by the imagined product. A religion is pulled by salvation. A market is pulled by profit. A feed is pulled by engagement. A model is pulled by loss minimization. A civilization is pulled by whatever it can collectively imagine as desirable, inevitable, or sacred.

This is where McKenna becomes useful again. The future does not causally reach backward, but representations of the future absolutely shape present cognition.

Humans live inside attractor fields. So do organizations. So do AI systems trained to optimize measurable targets. The danger is not that the transcendental object is real. The danger is that we build smaller ones constantly and then forget we built them.

The feed is an attractor. The KPI is an attractor. The benchmark is an attractor. The viral format is an attractor. The "helpful assistant" persona is an attractor. The market-clearing price is an attractor. The imagined AGI is an attractor.

Each one pulls cognition into its basin.

Gate Theory reframes the problem: an attractor is dangerous when it captures generation faster than evaluation can notice. You begin by using the metric. Then the metric uses you. You begin by prompting the model. Then the model's affordances shape what you think to ask. You begin by seeking novelty. Then novelty becomes the gate.

That is the pathological version of McKenna's curve: not acceleration toward transcendence, but acceleration toward whatever attractor has the lowest friction.

Psychedelic Cognition and Re-Gating

McKenna's psychedelic frame also belongs here, but carefully.

The point is not that psychedelics reveal cosmic truth directly. The point is that they alter the Gate.

In the dream essay, dreams were described as gateless cognition: associative generation with the usual waking filters reduced. Psychedelics are different. They are not simply gateless. They are re-gated. Some filters loosen: literal plausibility, ordinary salience, stable self-modeling. Other filters intensify: pattern detection, emotional significance, cross-scale mapping, mythic coherence.

This is why psychedelic thought can be both insightful and dangerously overconfident. It finds connections normal cognition would miss. It also promotes connections normal cognition would correctly reject.

McKenna's novelty theory has exactly that structure. It is a re-gated theory of history. It sees continuity between physics, biology, culture, machines, and subjective experience. That cross-scale move is valuable. It also overextends into prophecy when the evaluation gate does not come back online strongly enough.

The right response is not dismissal. It is gating.

Keep the pattern:

• complexity builds on achieved complexity,
• cognition moves across substrates,
• machine prostheses compress time,
• cultural acceleration increases event density,
• future representations act like attractors.

Reject the overclaim:

• no exact calendar,
• no guaranteed transcendence,
• no need to treat metaphor as mechanism,
• no permission to confuse resonance with truth.

Good cognition is not the absence of weirdness. It is weirdness that survives evaluation.

The Evaluation Lag

The central problem of the AI era is not intelligence. It is lag.

Generation has accelerated. Evaluation has not caught up.

A model can produce a legal argument faster than a lawyer can verify it. A code agent can alter a repository faster than a maintainer can understand the diff. A feed can generate outrage faster than a community can metabolize it. A synthetic image can travel farther than the fact-check. A strategic plan can be drafted before anyone has asked whether the goal is sane.

This is McKenna's acceleration with the romance removed.

The more novelty a system produces, the more gating it needs. If gates scale with novelty, acceleration becomes creativity. If gates fail to scale, acceleration becomes hallucination.

That sentence may be the simplest bridge between novelty theory and Gate Theory:

This does not mean we should slow everything down by default. Slowness can be its own pathology. Too much gating prevents discovery. It protects existing categories from the weird signals that would update them.

The goal is not less novelty. The goal is better gate architecture.

We need systems that can selectively lower friction gates while preserving load-bearing gates. More room for strange hypotheses, less tolerance for unsupported claims. More machine generation, more machine-assisted verification. More creative search, stronger provenance. More future imagination, tighter feedback from territory.

The future is not just faster. It is more dependent on evaluation design.

This essay extends the Gate Theory, Cognitive Zoom, and Controlled Symbiogenesis frameworks developed across this site. It is based on a captured Terence McKenna transcript archived at transcripts/youtube/GdEKhIk-8Gg-terence-mckenna-novelty-theory.txt.