Epistemic status: speculative synthesis of timeless-physics and anthropic ideas; exploratory and possibly confused, offered to clarify thinking rather than claim correctness.
I recently read Carlo Rovelli’s The Order of Time, which explains the concept of time as something that exists only as our local, relational view of entropy. This seems to square with recent experiments suggesting that, under certain quantum descriptions, the universe may in some sense be mathematically timeless — for example, in Page–Wootters emergent time models[1], where temporal evolution arises from correlations rather than an external time parameter.
Entropy, in so many words, is a description of disorder or predictability — both in the thermodynamic sense and in the information-theoretic sense (Shannon entropy). Because improbable states are defined relative to macro-descriptions we choose, I sometimes think of entropy as functionally similar to the value of world-states according to some utility function: the predictability of free energy is only meaningful insofar as it enables useful work. This is not the standard physics definition of entropy, but it’s an intuition I find helpful and have trouble finding incoherent.
However, this raises a question for me: if entropy defines our experience of time, but the concepts of usefulness and predictability that we use to describe entropy are only evaluable in retrospect, then what kind of model of the universe could avoid this apparent circularity? What does a universe look like that does not presuppose time or entropy in the very process of defining them?
I had trouble squaring this away for a while until I started considering a multiverse-or-many-worlds-style viewpoint. Let me offer a thought experiment.
Imagine the universe and all its possible configurations like monkeys with typewriters (infinite monkey theorem) — where each moment, or slice of experience, is a completed “book” written by the monkeys. In a vast possibility space — something loosely reminiscent of Tegmark’s Mathematical Universe Hypothesis, where every mathematically definable state may have a place — every conceivable configuration could exist somewhere. Within such an enormous landscape, let us imagine that even Boltzmann brain–like observer-moments could appear (Carroll 2017).
If such a multiverse were large enough to contain every possible Boltzmann-brain-like moment we could have, then some of those states would contain apparent memories of previous moments and predictions of future ones. Another state would contain the “next” brain-state consistent with those memories, and so on. Across these “disconnected” partitions of state in the timeless multiverse we are imagining, the configuration of a Boltzmann brain in one world-frame would contain a representation of its current world-model, its anticipation of the next, which would then be instantiated in some other universe-partition, and so forth. The felt continuity of experience and time would then be, conceptually, a kind of logical correlation — stitching together structures across these different universe-partitions, each embedding its own model of the ‘previous’ and the ‘next,’ but containing enough similarity to the others to form some dimension of logical continuity. This is reminiscent of Julian Barbour’s idea of “timeless Nows,” as described in The End of Time.
In this view, our predictions, theories, and understandings of the world are not global declarations about the totality of reality, but local statements about the slices that contain us, looking back at what appears to be a past. The truths we articulate would then be just the ones encoded in observer-states with memories logically correlated to the one we currently inhabit — because such correlations are what bind otherwise separate universe-moments into something that can be experienced as coherent.
If such a notion were accurate, it would be instructive as to the boundaries of what is knowable. It also aligns with Rovelli’s point that we only know what “was” because the present contains records — lower-entropy correlations — that support useful predictions now.
Entropy, in this picture, becomes a way of describing the structures that make prediction possible for an observer — structures that also serve as records. And we remember a low-entropy past because only such states contain us; we expect a higher-entropy future because many of its branches do not.
The metaphysical implications of such a model give me new questions, but more particularly clarify the importance of anthropics in our reasoning about truth. If everything could exist at once, but is only understood across structures that contain continuous self-similarity — then how should we view the rules and emergent patterns we see within complexity?
If we are characters across every book the monkeys wrote that contained our lives, stitched across the multiverse of pages, do we know anything about such a multiverse? Or only its states that cohere with our story of it? Not because reality is constrained to exist in the ways that contain us, but because we are constrained to be that which had predicted it to exist in something close to this way. For everything else, time may not exist at all.