As appeared in New Scientist, 20 January 2015 by Lee Smolin
COSMOLOGY is in crisis. Recent experiments have given us an increasingly precise narrative of the history of our universe, but attempts to interpret the data have led to a picture of a “preposterous universe” that eludes explanation in the terms familiar to scientists.
Everything we know suggests that the universe is unusual. It is flatter, smoother, larger and emptier than a “typical” universe predicted by the known laws of physics. If we reached into a hat filled with pieces of paper, each with the specifications of a possible universe written on it, it is exceedingly unlikely that we would get a universe anything like ours in one pick – or even a billion.
The challenge that cosmologists face is to make sense of this specialness. One approach to this question is inflation – the hypothesis that the early universe went through a phase of exponentially fast expansion. At first, inflation seemed to do the trick. A simple version of the idea gave correct predictions for the spectrum of fluctuations in the cosmic microwave background.
But a closer look shows that we have just moved the problem further back in time. To make inflation happen at all requires us to fine-tune the initial conditions of the universe. And unless inflation is highly tuned and constrained, it leads to a runaway process of universe creation. As a result, some cosmologists suggest that there is not one universe, but an infinite number, with a huge variety of properties: the multiverse. There are an infinite number of universes in the collection that are like our universe and an infinite number that are not. But the ratio of infinity to infinity is undefined, and can be made into anything the theorist wants. Thus the multiverse theory has difficulty making any firm predictions and threatens to take us out of the realm of science.
These other universes are unobservable and because chance dictates the random distribution of properties across universes, positing the existence of a multiverse does not let us deduce anything about our universe beyond what we already know. As attractive as the idea may seem, it is basically a sleight of hand, which converts an explanatory failure into an apparent explanatory success. The success is empty because anything that might be observed about our universe could be explained as something that must, by chance, happen somewhere in the multiverse.
We started out trying to explain why the universe is so special, and we end up being asked to believe that our universe is one of an infinite number of universes with random properties. This makes me suspect that there is a basic but unexamined assumption about the laws of nature that must be overturned.
Cosmology has new questions to answer. Not just what are the laws, but why are these laws the laws? How were they chosen? We can’t just hypothesise what the initial conditions were at the big bang, we need to explain those initial conditions. Thus we are in the position of a computer program asked to explain its inputs. It is clear that if we are to get anywhere, we need to invent new methods, and perhaps new kinds of laws, to gain a scientific description of the universe as a whole.
Physicist James Hartle has talked about the “excess baggage” that has to be left on the platform before we can board the train to further progress in cosmology. In work together with philosopher Roberto Mangabeira Unger, we believe we have identified several of pieces of this baggage.
The first thing that must be discarded is the assumption that the same kind of laws that work on the scale of small subsystems of the world work, scaled up, at the level of the whole universe. We call this assumption the cosmological fallacy because it leads to a breakdown of predictability – as in the multiverse.
The second piece of excess baggage is the Newtonian paradigm, a method common to classical and quantum mechanics and general relativity. It is used to describe a subsystem of the universe, like a system studied in a laboratory, an atom or a star. This method depends on two elements: the set of possible initial configurations (or states) of the system and a law that specifies how the states change in time. Once we start the system off at an initial state, the law tells us what the state will be at later times. But if the laws and initial conditions are the inputs to the method, they cannot be its outputs too. If we want to understand why the laws hold and how the initial conditions of the universe are chosen, we need a new kind of law and methodology.
The Newtonian paradigm is ideal for describing systems in a laboratory but if we attempt to apply it to the universe, it explains both too little and too much. It fails to explain how the solution to the laws governing our universe is picked from the infinite number of solutions that don’t govern anything. But it also predicts an infinite number of facts about an infinite number of non-existent universes. This is one of the reasons why the Newtonian paradigm cannot be applied to the universe as a whole.
These concepts illuminate why the multiverse fails as a scientific hypothesis in spite of the fact that simple versions of inflation made some predictions that have been confirmed. The idea of inflation is plagued by the need to explain how the initial conditions were chosen. This was done in the context of a methodology that only makes sense when applied to a subsystem of a larger system. When applied to the universe, it forced us to treat the universe as a subsystem of a much larger system: hence we had to invent the multiverse. And thus with an infinite ensemble of unobservable entities we leave the domain of science behind. In some sense, the multiverse embodies the unreal ensemble of all possible solutions to the laws of physics, imagined as elements of an invented ensemble of bubble universes. But this just trades one imaginary, unreal ensemble for another.
Once we accept that we need a new paradigm to do science at the level of the universe as a whole, the next question to ask is what principle that new paradigm should be founded on. This is a question we hope to provoke cosmologists to think about. Mangabeira Unger and I propose three principles, which we argue are necessary to underlie any theory capable of explaining big cosmological questions – like the selection of the laws and initial conditions of the universe – in a way that is open to experimental test.
The first is that there is just one universe. The second is that time is real and the laws of nature are not timeless but evolve. The third is that mathematics is not a description of some separate timeless, Platonic reality, but is a description of the properties of one universe.
These principles take us beyond the Newtonian paradigm and the cosmological fallacy, and are a starting point for exploring the science of the universe as a whole.