Physicists Debate Hawking’s Idea That the Universe Had No Beginning

In 1981, many of the world’s leading cosmologists gathered at the Pontifical Academy of Sciences, a vestige of the coupled lineages of science and theology located in an elegant villa in the gardens of the Vatican. Stephen Hawking chose the august setting to present what he would later regard as his most important idea: a proposal about how the universe could have arisen from nothing.

Before Hawking’s talk, all cosmological origin stories, scientific or theological, had invited the rejoinder, “What happened before that?” The Big Bang theory, for instance — pioneered 50 years before Hawking’s lecture by the Belgian physicist and Catholic priest Georges Lemaître, who later served as president of the Vatican’s academy of sciences — rewinds the expansion of the universe back to a hot, dense bundle of energy. But where did the initial energy come from?

The Big Bang theory had other problems. Physicists understood that an expanding bundle of energy would grow into a crumpled mess rather than the huge, smooth cosmos that modern astronomers observe. In 1980, the year before Hawking’s talk, the cosmologist Alan Guth realized that the Big Bang’s problems could be fixed with an add-on: an initial, exponential growth spurt known as cosmic inflation, which would have rendered the universe huge, smooth and flat before gravity had a chance to wreck it. Inflation quickly became the leading theory of our cosmic origins. Yet the issue of initial conditions remained: What was the source of the minuscule patch that allegedly ballooned into our cosmos, and of the potential energy that inflated it?

Hawking, in his brilliance, saw a way to end the interminable groping backward in time: He proposed that there’s no end, or beginning, at all. According to the record of the Vatican conference, the Cambridge physicist, then 39 and still able to speak with his own voice, told the crowd, “There ought to be something very special about the boundary conditions of the universe, and what can be more special than the condition that there is no boundary?”

The “no-boundary proposal,” which Hawking and his frequent collaborator, James Hartle, fully formulated in a 1983 paper, envisions the cosmos having the shape of a shuttlecock. Just as a shuttlecock has a diameter of zero at its bottommost point and gradually widens on the way up, the universe, according to the no-boundary proposal, smoothly expanded from a point of zero size. Hartle and Hawking derived a formula describing the whole shuttlecock — the so-called “wave function of the universe” that encompasses the entire past, present and future at once — making moot all contemplation of seeds of creation, a creator, or any transition from a time before.

“Asking what came before the Big Bang is meaningless, according to the no-boundary proposal, because there is no notion of time available to refer to,” Hawking said in another lecture at the Pontifical Academy in 2016, a year and a half before his death. “It would be like asking what lies south of the South Pole.”

Hartle and Hawking’s proposal radically reconceptualized time. Each moment in the universe becomes a cross-section of the shuttlecock; while we perceive the universe as expanding and evolving from one moment to the next, time really consists of correlations between the universe’s size in each cross-section and other properties — particularly its entropy, or disorder. Entropy increases from the cork to the feathers, aiming an emergent arrow of time. Near the shuttlecock’s rounded-off bottom, though, the correlations are less reliable; time ceases to exist and is replaced by pure space. As Hartle, now 79 and a professor at the University of California, Santa Barbara, explained it by phone recently, “We didn’t have birds in the very early universe; we have birds later on. … We didn’t have time in the early universe, but we have time later on.”

The no-boundary proposal has fascinated and inspired physicists for nearly four decades. “It’s a stunningly beautiful and provocative idea,” said Neil Turok, a cosmologist at the Perimeter Institute for Theoretical Physics in Waterloo, Canada, and a former collaborator of Hawking’s. The proposal represented a first guess at the quantum description of the cosmos — the wave function of the universe. Soon an entire field, quantum cosmology, sprang up as researchers devised alternative ideas about how the universe could have come from nothing, analyzed the theories’ various predictions and ways to test them, and interpreted their philosophical meaning. The no-boundary wave function, according to Hartle, “was in some ways the simplest possible proposal for that.”

But two years ago, a paper by Turok, Job Feldbrugge of the Perimeter Institute, and Jean-Luc Lehners of the Max Planck Institute for Gravitational Physics in Germany called the Hartle-Hawking proposal into question. The proposal is, of course, only viable if a universe that curves out of a dimensionless point in the way Hartle and Hawking imagined naturally grows into a universe like ours. Hawking and Hartle argued that indeed it would — that universes with no boundaries will tend to be huge, breathtakingly smooth, impressively flat, and expanding, just like the actual cosmos. “The trouble with Stephen and Jim’s approach is it was ambiguous,” Turok said — “deeply ambiguous.”

In their 2017 paper, published in Physical Review Letters, Turok and his co-authors approached Hartle and Hawking’s no-boundary proposal with new mathematical techniques that, in their view, make its predictions much more concrete than before. “We discovered that it just failed miserably,” Turok said. “It was just not possible quantum mechanically for a universe to start in the way they imagined.” The trio checked their math and queried their underlying assumptions before going public, but “unfortunately,” Turok said, “it just seemed to be inescapable that the Hartle-Hawking proposal was a disaster.”

The paper ignited a controversy. Other experts mounted a vigorous defense of the no-boundary idea and a rebuttal of Turok and colleagues’ reasoning. “We disagree with his technical arguments,” said Thomas Hertog, a physicist at the Catholic University of Leuven in Belgium who closely collaborated with Hawking for the last 20 years of the latter’s life. “But more fundamentally, we disagree also with his definition, his framework, his choice of principles. And that’s the more interesting discussion.”

After two years of sparring, the groups have traced their technical disagreement to differing beliefs about how nature works. The heated — yet friendly — debate has helped firm up the idea that most tickled Hawking’s fancy. Even critics of his and Hartle’s specific formula, including Turok and Lehners, are crafting competing quantum-cosmological models that try to avoid the alleged pitfalls of the original while maintaining its boundless allure.

Garden of Cosmic Delights

Hartle and Hawking saw a lot of each other from the 1970s on, typically when they met in Cambridge for long periods of collaboration. The duo’s theoretical investigations of black holes and the mysterious singularities at their centers had turned them on to the question of our cosmic origin.

In 1915, Albert Einstein discovered that concentrations of matter or energy warp the fabric of space-time, causing gravity. In the 1960s, Hawking and the Oxford University physicist Roger Penrose proved that when space-time bends steeply enough, such as inside a black hole or perhaps during the Big Bang, it inevitably collapses, curving infinitely steeply toward a singularity, where Einstein’s equations break down and a new, quantum theory of gravity is needed. The Penrose-Hawking “singularity theorems” meant there was no way for space-time to begin smoothly, undramatically at a point.

Hawking and Hartle were thus led to ponder the possibility that the universe began as pure space, rather than dynamical space-time. And this led them to the shuttlecock geometry. They defined the no-boundary wave function describing such a universe using an approach invented by Hawking’s hero, the physicist Richard Feynman. In the 1940s, Feynman devised a scheme for calculating the most likely outcomes of quantum mechanical events. To predict, say, the likeliest outcomes of a particle collision, Feynman found that you could sum up all possible paths that the colliding particles could take, weighting straightforward paths more than convoluted ones in the sum. Calculating this “path integral” gives you the wave function: a probability distribution indicating the different possible states of the particles after the collision.

Likewise, Hartle and Hawking expressed the wave function of the universe — which describes its likely states — as the sum of all possible ways that it might have smoothly expanded from a point. The hope was that the sum of all possible “expansion histories,” smooth-bottomed universes of all different shapes and sizes, would yield a wave function that gives a high probability to a huge, smooth, flat universe like ours. If the weighted sum of all possible expansion histories yields some other kind of universe as the likeliest outcome, the no-boundary proposal fails.

The problem is that the path integral over all possible expansion histories is far too complicated to calculate exactly. Countless different shapes and sizes of universes are possible, and each can be a messy affair. “Murray Gell-Mann used to ask me,” Hartle said, referring to the late Nobel Prize-winning physicist, “if you know the wave function of the universe, why aren’t you rich?” Of course, to actually solve for the wave function using Feynman’s method, Hartle and Hawking had to drastically simplify the situation, ignoring even the specific particles that populate our world (which meant their formula was nowhere close to being able to predict the stock market). They considered the path integral over all possible toy universes in “minisuperspace,” defined as the set of all universes with a single energy field coursing through them: the energy that powered cosmic inflation. (In Hartle and Hawking’s shuttlecock picture, that initial period of ballooning corresponds to the rapid increase in diameter near the bottom of the cork.)

Even the minisuperspace calculation is hard to solve exactly, but physicists know there are two possible expansion histories that potentially dominate the calculation. These rival universe shapes anchor the two sides of the current debate.

The rival solutions are the two “classical” expansion histories that a universe can have. Following an initial spurt of cosmic inflation from size zero, these universes steadily expand according to Einstein’s theory of gravity and space-time. Weirder expansion histories, like football-shaped universes or caterpillar-like ones, mostly cancel out in the quantum calculation.

One of the two classical solutions resembles our universe. On large scales, it’s smooth and randomly dappled with energy, due to quantum fluctuations during inflation. As in the real universe, density differences between regions form a bell curve around zero. If this possible solution does indeed dominate the wave function for minisuperspace, it becomes plausible to imagine that a far more detailed and exact version of the no-boundary wave function might serve as a viable cosmological model of the real universe.

The other potentially dominant universe shape is nothing like reality. As it widens, the energy infusing it varies more and more extremely, creating enormous density differences from one place to the next that gravity steadily worsens. Density variations form an inverted bell curve, where differences between regions approach not zero, but infinity. If this is the dominant term in the no-boundary wave function for minisuperspace, then the Hartle-Hawking proposal would seem to be wrong.

The two dominant expansion histories present a choice in how the path integral should be done. If the dominant histories are two locations on a map, megacities in the realm of all possible quantum mechanical universes, the question is which path we should take through the terrain. Which dominant expansion history, and there can only be one, should our “contour of integration” pick up? Researchers have forked down different paths.

In their 2017 paper, Turok, Feldbrugge and Lehners took a path through the garden of possible expansion histories that led to the second dominant solution. In their view, the only sensible contour is one that scans through real values (as opposed to imaginary values, which involve the square roots of negative numbers) for a variable called “lapse.” Lapse is essentially the height of each possible shuttlecock universe — the distance it takes to reach a certain diameter. Lacking a causal element, lapse is not quite our usual notion of time. Yet Turok and colleagues argue partly on the grounds of causality that only real values of lapse make physical sense. And summing over universes with real values of lapse leads to the wildly fluctuating, physically nonsensical solution.

“People place huge faith in Stephen’s intuition,” Turok said by phone. “For good reason — I mean, he probably had the best intuition of anyone on these topics. But he wasn’t always right.”

Imaginary Universes

Jonathan Halliwell, a physicist at Imperial College London, has studied the no-boundary proposal since he was Hawking’s student in the 1980s. He and Hartle analyzed the issue of the contour of integration in 1990. In their view, as well as Hertog’s, and apparently Hawking’s, the contour is not fundamental, but rather a mathematical tool that can be placed to greatest advantage. It’s similar to how the trajectory of a planet around the sun can be expressed mathematically as a series of angles, as a series of times, or in terms of any of several other convenient parameters. “You can do that parameterization in many different ways, but none of them are any more physical than another one,” Halliwell said.

He and his colleagues argue that, in the minisuperspace case, only contours that pick up the good expansion history make sense. Quantum mechanics requires probabilities to add to 1, or be “normalizable,” but the wildly fluctuating universe that Turok’s team landed on is not. That solution is nonsensical, plagued by infinities and disallowed by quantum laws — obvious signs, according to no-boundary’s defenders, to walk the other way.

It’s true that contours passing through the good solution sum up possible universes with imaginary values for their lapse variables. But apart from Turok and company, few people think that’s a problem. Imaginary numbers pervade quantum mechanics. To team Hartle-Hawking, the critics are invoking a false notion of causality in demanding that lapse be real. “That’s a principle which is not written in the stars, and which we profoundly disagree with,” Hertog said.

According to Hertog, Hawking seldom mentioned the path integral formulation of the no-boundary wave function in his later years, partly because of the ambiguity around the choice of contour. He regarded the normalizable expansion history, which the path integral had merely helped uncover, as the solution to a more fundamental equation about the universe posed in the 1960s by the physicists John Wheeler and Bryce DeWitt. Wheeler and DeWitt — after mulling over the issue during a layover at Raleigh-Durham International — argued that the wave function of the universe, whatever it is, cannot depend on time, since there is no external clock by which to measure it. And thus the amount of energy in the universe, when you add up the positive and negative contributions of matter and gravity, must stay at zero forever. The no-boundary wave function satisfies the Wheeler-DeWitt equation for minisuperspace.  

In the final years of his life, to better understand the wave function more generally, Hawking and his collaborators started applying holography — a blockbuster new approach that treats space-time as a hologram. Hawking sought a holographic description of a shuttlecock-shaped universe, in which the geometry of the entire past would project off of the present.

That effort is continuing in Hawking’s absence. But Turok sees this shift in emphasis as changing the rules. In backing away from the path integral formulation, he says, proponents of the no-boundary idea have made it ill-defined. What they’re studying is no longer Hartle-Hawking, in his opinion — though Hartle himself disagrees.

For the past year, Turok and his Perimeter Institute colleagues Latham Boyle and Kieran Finn have been developing a new cosmological model that has much in common with the no-boundary proposal. But instead of one shuttlecock, it envisions two, arranged cork to cork in a sort of hourglass figure with time flowing in both directions. While the model is not yet developed enough to make predictions, its charm lies in the way its lobes realize CPT symmetry, a seemingly fundamental mirror in nature that simultaneously reflects matter and antimatter, left and right, and forward and backward in time. One disadvantage is that the universe’s mirror-image lobes meet at a singularity, a pinch in space-time that requires the unknown quantum theory of gravity to understand. Boyle, Finn and Turok take a stab at the singularity, but such an attempt is inherently speculative.

There has also been a revival of interest in the “tunneling proposal,” an alternative way that the universe might have arisen from nothing, conceived in the ’80s independently by the Russian-American cosmologists Alexander Vilenkin and Andrei Linde. The proposal, which differs from the no-boundary wave function primarily by way of a minus sign, casts the birth of the universe as a quantum mechanical “tunneling” event, similar to when a particle pops up beyond a barrier in a quantum mechanical experiment.

Questions abound about how the various proposals intersect with anthropic reasoning and the infamous multiverse idea. The no-boundary wave function, for instance, favors empty universes, whereas significant matter and energy are needed to power hugeness and complexity. Hawking argued that the vast spread of possible universes permitted by the wave function must all be realized in some larger multiverse, within which only complex universes like ours will have inhabitants capable of making observations. (The recent debate concerns whether these complex, habitable universes will be smooth or wildly fluctuating.) An advantage of the tunneling proposal is that it favors matter- and energy-filled universes like ours without resorting to anthropic reasoning — though universes that tunnel into existence may have other problems.

No matter how things go, perhaps we’ll be left with some essence of the picture Hawking first painted at the Pontifical Academy of Sciences 38 years ago. Or perhaps, instead of a South Pole-like non-beginning, the universe emerged from a singularity after all, demanding a different kind of wave function altogether. Either way, the pursuit will continue. “If we are talking about a quantum mechanical theory, what else is there to find other than the wave function?” asked Juan Maldacena, an eminent theoretical physicist at the Institute for Advanced Study in Princeton, New Jersey, who has mostly stayed out of the recent fray. The question of the wave function of the universe “is the right kind of question to ask,” said Maldacena, who, incidentally, is a member of the Pontifical Academy. “Whether we are finding the right wave function, or how we should think about the wave function — it’s less clear.”

Correction: This article was revised on June 6, 2019, to list Latham Boyle and Kieran Finn as co-developers of the CPT-symmetric universe idea.

This article was reprinted on Wired.com.