Editor’s note: Like mighty Samson of old, in a single-handed effort, physicist Bruce Gordon has heroically forced up the Internet’s collective IQ by several points. On The Telegraph website, he responded to 1,870+ comments on Stephen Meyer’s op-ed reflecting on the 20th anniversary of Richard Dawkins’s The God Delusion. See here for his earlier reply. Dr. Gordon’s second response:
Interestingly, the comment feed contains the assertion that no cosmologists have ever claimed that the universe came from nothing. Some notable counterexamples come to mind. First, Alexander Vilenkin titled his 1982 paper on quantum cosmological origins “Creation of Universes from Nothing.” That seems to qualify. Then there was Stephen Hawking and Leonard Mlodinow’s problematic remark in their bestseller The Grand Design that “because there is a law such as gravity, the universe can and will create itself from nothing.” On this view, the universe would have to exist before it exists in order to create itself. And we shouldn’t forget Lawrence Krauss’s book A Universe from Nothing. It would seem that “nothing” has its champions among cosmologists.
If the ground is now shifted by objectors, saying that the origin of the universe finds its foundational explanation in an inflaton field, or colliding branes, or some primordial quantum state, then [Stephen] Meyer’s point has been conceded: “nothing,” in physics and cosmology, is always something. It is some sort of physical structure that is described by some kind of mathematical “law.” But laws, as just noted, are merely descriptive. They do not cause anything; they just give a description of what happens.
Even Stephen Hawking himself, in one of his more metaphysically lucid moments earlier in his career, asked “What is it that breathes fire into the equations and makes a universe for them to describe?” He implicitly recognized that even in quantum cosmology, the universe requires an explanation for its existence, quite aside from the fact that both the Hartle-Hawking (after the necessary conversion back to real time) and the Vilenkin semi-classical models have a temporal beginning.
Incomplete to the Past
There is also the matter of the Borde-Guth-Vilenkin (BGV) theorem, which requires neither general relativity nor any energy conditions, yet shows that any universe that has on average been expanding along any past-directed world line is incomplete to the past. The routes around this incompleteness require a rejection of average expansion, leading either to a contracting universe, whether an eternal contraction with a single bounce or endless cycles, or quasi-static phases that lasted throughout a past-eternity only to fluctuate into our universe about 13.8 billion years ago. But cyclic universes either succumb to intractable entropy problems or they dilute entropy density by a net growth in the scale factor, as in the 2019 Ijjas-Steinhardt model, landing them subject to the BGV theorem again, as Kinney and Stein have shown.
An eternal contraction leading to a single bounce would require conditions that were infinitely fine-tuned in the infinite past and would suffer from runaway anisotropy. On the other hand, models in which there is a quasi-static cosmic seed succumb to quantum instability problems that prohibit past-eternality. A beginning to the universe may be hard to swallow for some, but it’s also very hard to avoid.
Our Universe and Its Properties
The contingency of any physical explanation that might be proposed still leaves open the question of why this universe exists with the properties it has, rather than another. To take it as a brute fact is to say that some things happen for no reason at all — whole universes, for instance. But down this path lies madness. There is no objective basis for assigning probability if no explanation is the “explanation” on offer, so there’s no way of saying that “no explanation” is unlikely to be the “reason” that something happened.
Furthermore, if it’s possible that some things happen for no reason at all, then this possibility competes with every explanation we think we have. Even worse, it becomes possible that our current thoughts are happening for no reason at all and we’re completely disconnected from reality. This not only undermines the practice of science, it undermines all human knowledge. This is not the path that anyone should wish to follow, so if it is the path you are on, take note.
The Anthropic Reply
Finally, we might consider the anthropic reply to fine-tuning that functions as an observer-selection effect: “If the universe weren’t fine-tuned, we wouldn’t be here to see it.” Fine-tuning may be a necessary condition of our observation, but naming it does not do anything to explain why it is satisfied. The observer-selection effect explains why we don’t observe a lifeless universe, but apart from a multiverse with different laws and constants (more on this momentarily) it has no relevance to explaining why a life-permitting universe exists at all.
The Canadian philosopher John Leslie gave the example of a prisoner who survives a firing squad with 50 marksmen. It is hardly sufficient to say that he wouldn’t be here to make the observation if they hadn’t missed. The relevant question is to ask why they all missed, just as the relevant question in cosmology is why the universe is fine-tuned for life. The Everettian many-worlds of quantum mechanics, even if it were tenable, would not solve this particular problem, because all of the branches in the Everettian universe presume the same life-permitting laws and constants.
Shifting the Design Problem to a Metalevel
The primary alternative is the inflationary string landscape, where each of the vast number of different ways of compactifying the extra spatial dimensions in string theory produces universes with different laws and constants. But such a multiverse has contingent metalaws (an M-theoretic construction) that govern the production of different universes and that need their own precise mechanisms and constants in order for the theory to function in its universe-generating capacity. Barring an infinite explanatory regress of metalevels — which would itself be contingent and need an explanation — it is evident that the multiverse just shifts the design problem to a metalevel; it doesn’t dispense with it.
To the fine-tuning that multiverse models have relocated but not avoided, we may add the fact that they generate an intractable measure problem. Everything that can happen does happen infinitely many times, so the probability of its happening is undefined (infinity over infinity) until a measure is chosen, and different measures give different answers. Under many of these measures, a Boltzmann brain problem results in which evanescent observers are typical and vastly outnumber the ordinary ones. Sean Carroll calls such theories “cognitively unstable.” In short, this consequence of the model undermines the very epistemic basis on which the model itself might be asserted.