Published online 13 August 2002 | Nature | doi:10.1038/news020812-2


Is physics watching over us?

Our Universe is so unlikely that we must be missing something.

An expanding universe is destined to repeat itself.An expanding universe is destined to repeat itself.

In an argument that would have gratified the ancient Greeks, physicists have claimed that the prevailing theoretical view of the Universe is logically flawed. Arranging the cosmos as we think it is arranged, say the team, would have required a miracle1.

An ever-more-rapidly expanding Universe is destined to repeat itself, say Leonard Susskind of Stanford University, California, and his colleagues. But the chances that such re-runs would produce worlds like ours are infinitesimal.

So either space is not accelerating for the reasons we think it is, or we have yet to discover some principle of physics, the researchers conclude. Like a guardian angel, this principle would pick out those few initial states that lead to a Universe like ours, and then guide cosmic evolution so that it really does unfold this way.

The incomprehensibility of our situation even drives Susskind's team to ponder whether an "unknown agent intervened in the evolution [of the Universe] for reasons of its own". But creationists should not rejoice: even a god such as this can't explain how things got so strange.

The problem stems from the observation in 1998 that the Universe's expansion seems to be speeding up. The most popular explanation for this is that there is a cosmological constant - a repulsive force that opposes gravity.

As things stand, other galaxies will eventually disappear as they zoom away from us faster than the speed of light. Then nothing that happens in those parts of the cosmos can affect us. Our world - and everywhere else - will be isolated behind a boundary called a de Sitter horizon.

This view holds that the Universe will fragment into a foam of bubbles separated by de Sitter horizons: a de Sitter space. Each bubble would eventually settle into a bland, lifeless uniformity. And that would be the end of history.

Here we go again

Or would it? Thermodynamics says otherwise, reckon Susskind and his colleagues. Wait long enough, and everything that can happen, will. There is nothing to stop a drop of ink dispersed in a glass of water from gathering its molecules back into a single drop. It's incredibly unlikely but, with infinite patience, we'll see it happen.

In the same way, a Universe driven to become a de Sitter space by a cosmological constant will, after an absurdly long time, return to something like its original condition, the researchers say. A new cosmic history will then unfold, including the reappearance of life. But the chances that such a cosmic recurrence will produce a Universe like ours are extremely slim.

Cosmologists have a rejoinder to this kind of argument, called the anthropic principle. This says that, no matter how unlikely the Universe seems, the very fact that we are here to ask such questions resolves the paradox. If things were otherwise, life wouldn't exist and the question would never arise.

But Susskind's team show that the anthropic principle won't help, because a vast number of Universes would permit life and yet look quite different from this one. All of these habitable Universes would result from 'miraculous' statistical events. But there are so many of them that they would vastly overwhelm a cosmos like ours.

Even if 'something' had set the peculiar initial conditions of our Universe, this would only apply for its first run. Subsequent recurrences would produce a quite different Universe.

In that case, we'd have to conclude that we are in the first unfolding of this carefully crafted Universe. This all seems too much like special pleading, the researchers say.

So either there is no cosmological constant after all - in which case, why is the Universe accelerating? - or we're missing something fundamental. 

  • References

    1. Dyson, L., Kleban, M. & Susskind, L. Disturbing implications of a cosmological constant. Preprint, (2002).