The Beginning Of Infinity

Once you take explanation seriously, an old assumption starts to wobble. We tend to imagine knowledge as a finite territory: there's a fixed amount to know, we're steadily filling it in, and one day we'll approach the edges. Deutsch thinks this picture is not just wrong but backwards. Every good explanation we reach opens up problems that were previously invisible, and each of those becomes the starting point for the next round of discovery. Knowledge doesn't converge on a boundary. It generates its own frontier as it goes.

The engine driving this, he argues, is creativity — the specifically human capacity to conjecture explanations that were nowhere in the data. This puts him sharply at odds with the standard story, often traced to Francis Bacon, that we learn by patiently gathering observations and letting patterns emerge. Observations never speak for themselves; someone has to guess what they might mean, and the guess always outruns the evidence. Nobody deduced the axial tilt by staring harder at sunlight. They imagined a hidden arrangement and then checked whether reality could survive the test.

If there's no ceiling, why does progress feel so hard? Deutsch's answer is that we are, and will always remain, fallible. We get things wrong; every explanation we hold contains errors we haven't yet noticed. He treats this not as a cause for despair but as the load-bearing beam of the whole structure. Fallibilism — the frank recognition that we can always be mistaken — is what keeps knowledge open to correction. The alternative, believing we've reached final truth on anything, is what he calls the beginning of dogmatism, the moment a tradition stops improving and starts defending itself.

He compresses the whole outlook into a pair of statements that sound almost like slogans but are meant literally: problems are inevitable, and problems are soluble. There is no state of completed knowledge in which the problems run out, because every solution reshuffles the situation and throws up new ones. But there is also no problem that is soluble in principle yet permanently barred to us. What stands between us and any given solution is a missing explanation — and missing explanations are exactly the kind of thing creativity is built to supply.

Step back from the physics and the philosophy, and what Deutsch is really arguing about is us — what kind of thing a human being is. The dominant modern self-image is one of insignificance: a fluke of chemistry on a minor planet, adrift in a cosmos indifferent to whether we understand it. Deutsch finds this both false and quietly corrosive. If we are universal explainers — beings capable of understanding anything that can be understood, and of building anything the laws of physics allow — then we are not incidental to the universe at all. We are among the very few places, perhaps the only place we know of, where the universe generates knowledge about itself.

This reframes what he means by optimism, a word he is careful to strip of its usual softness. He is not saying things will turn out fine, or that history bends toward improvement on its own. His optimism is a principle about problems, not a forecast about outcomes: all evils are due to insufficient knowledge, and knowledge can grow. It says nothing about whether we will acquire the right knowledge in time. Progress is available; it is never automatic. A civilization that stops seeking better explanations doesn't hold steady — it decays, because the problems keep coming and it has stopped generating solutions.

Deutsch begins with people telling stories about the Sun and ends with a claim about the reach of human minds, and the through-line is the same the whole way: some accounts of the world can be varied at will and some cannot, and everything hinges on preferring the second kind. Good explanations are hard to vary, they reach far past the problems they were built for, and they can always be improved — which is precisely why the growth of knowledge has no natural stopping point. Progress isn't the slow filling-in of a fixed map. It's an open-ended process that keeps generating its own next questions.