12 Factors of Knowledge Capitalism
Back to home
Part II · The Evolution Interface Factor 3 of 12
Exploration Creates New Conjectures
•
6 min read
Principle

New knowledge begins as a creative guess. Exploration is the sovereign act of going out to meet the unknown and finding problems worth explaining.

In brief
  • Knowledge is created, not deduced: it starts life as a conjecture — an invented guess at how some piece of reality works.
  • Conjectures need raw material. Exploration surfaces the problems worth explaining, and nobody can do it on your behalf.
  • Exploration has a price — effort spent on the unknown is withheld from exploiting the known — but without paying it, the frontier never moves.

#The claim

Knowledge is not deduced but created. A new piece of knowledge does not arrive by reading reality off the world; it begins as a conjecture — a bold, creative guess at how something works, invented by a mind and only afterward tested against the world. Good explanations are authored, not transcribed.Deutsch (2011), The Beginning of Infinity. Explanations are creative conjectures about what is really there — authored, then criticized.

But conjectures need raw material. A guess is a response to a problem, and problems must be found. This is what inquiry is: the deliberate transformation of an indeterminate situation into a determinate one — going out to meet the unknown rather than waiting for it to arrive.Dewey (1938), Logic: The Theory of Inquiry. Inquiry as the controlled transformation of an indeterminate situation into a determinate one. The act is sovereign in a precise sense: no one can expand your realm of coverage for you. Another mind’s exploration produces another mind’s conjectures.

Exploration fans out into conjectures From a single seeker node, branches fan outward into a frontier of dashed, open conjectures. seeker conjectures
Figure 1. Exploration fans outward: each committed branch opens new branch points, and the frontier — dashed, still unproven — is made of open conjectures.

#The mechanism

Exploration is not free, and its price has a name. The explore–exploit tradeoff says that every unit of effort spent probing the unknown is a unit withheld from exploiting what you already know works.Sutton & Barto (2018), Reinforcement Learning. The formal home of the explore–exploit tradeoff. Exploit forever and you harvest a shrinking field; explore recklessly and you never harvest at all. Every learning system — a foraging animal, a research lab, a reinforcement learner — sits somewhere on this dial. What the tradeoff cannot say is which conjecture the exploration will produce. That remains a creative act.

The explore–exploit tradeoff From an actor node, a tight solid loop returns to known ground labelled exploit, while a long arrow crosses a dashed frontier line into unknown territory labelled explore. exploit the known explore the unknown frontier
Figure 2. The dial every learning system sets: the tight loop harvests known ground; the long arrow pays present yield to move the frontier. Only the second creates new conjectures.

Tools shift the dial’s economics by extending the seeker’s reach. The memex was imagined as a machine for trails of association,Bush (1945), “As We May Think.” The memex: association trails as an external substrate for exploration. and the augmentation program made the ambition explicit: not artificial thought, but amplified human intellect.Engelbart (1962), “Augmenting Human Intellect.” Tooling as amplification of the explorer, not replacement. Every generation of tooling since — search, simulation, language models — lowers the cost of a probe into the unknown, which means more conjectures per unit of effort.

#Why it matters

The frontier now has non-human explorers on it. As artificial cognitive systems move into open-ended learning, exploration becomes their engine for generating novel experience — and, more importantly, novel questions.Silver & Sutton (2025), “The Era of Experience.” Agents that learn predominantly from their own generated experience. A system that only exploits its training distribution is a library; one that explores is a seeker. The framework’s evolution interface begins here for carbon and silicon alike: everything downstream — criticism, action, capital — operates on material that only exploration can supply.

Exploration alone, though, produces only guesses. What turns a guess into knowledge is the subject of the next factor.

Related factors

Factor 4 — Criticism Hardens Conjecture into Knowledge
What happens to conjectures nextFactor 6 — Cognitive Actors Are the Nodes of Agency
The seekers, carbon and siliconFactor 1 — Knowledge is Negative Entropy
Why the frontier is worth moving
The factor, in full

Knowledge is not deduced but created. New knowledge begins as a conjecture, a bold and creative guess at how some piece of reality works, invented rather than derived. Good explanations1 are authored, not read off the world. Conjectures need raw material, so the cognitive actor must go out to meet the unknown, inquiring2 and exploring to surface the problems worth explaining. This act is sovereign, since no one can expand your realm of coverage for you. The explore–exploit tradeoff3 names its price. Effort spent exploring is withheld from exploiting what you already know, yet without it the frontier never moves. Tools extend the seeker’s reach, from the memex4 to augmenting human intellect.5 As artificial cognitive systems pursue open-ended learning,6 exploration becomes their engine for novel experience and, more importantly, novel questions.

References

  1. Deutsch, D. 2011. The Beginning of Infinity: Explanations That Transform the World. London: Allen Lane.
  2. Dewey, J. 1938. Logic: The Theory of Inquiry. New York: Henry Holt.
  3. Sutton, R. S., and A. G. Barto. 2018. Reinforcement Learning: An Introduction. 2nd ed. Cambridge, MA: MIT Press.
  4. Bush, V. 1945. “As We May Think.” The Atlantic, July 1945.
  5. Engelbart, D. C. 1962. “Augmenting Human Intellect: A Conceptual Framework.” SRI Summary Report AFOSR-3223.
  6. Silver, D., and R. S. Sutton. 2025. “Welcome to the Era of Experience.” In Designing an Intelligence. Cambridge, MA: MIT Press.
Previous Language Compresses and Carries Knowledge
Next Criticism Hardens Conjecture into Knowledge