From issue 2601 of New Scientist magazine, 28 April 2007, page 28-33The cosmos - before the big bangHow did the universe begin? The question is as old as humanity. Sure, we know that something like thebig bang happened, but the theory doesn't explain some of the most important bits: why it happened, whatthe conditions were at the time, and other imponderables.Many cosmologists think our standard picture of how the universe came to be is woefully incomplete oreven plain wrong, and they have been dreaming up a host of strange alternatives to explain how we gothere. For the first time, they are trying to pin down the initial conditions of the big bang. In particular,they want to solve the long-standing mystery of how the universe could have begun in such a well-ordered state, as fundamental physics implies, when it seems utter chaos should have reigned.Several models have emerged that propose intriguing answers to this question. One says the universebegan as a dense sea of black holes. Another says the big bang was sparked by a collision between twomembranes floating in higher-dimensional space. Yet another says our universe was originally rippedfrom a larger entity, and that in turn countless baby universes will be born from the wreckage of ours.Crucially, each scenario makes unique and testable predictions; observations coming online in the nextfew years should help us to decide which, if any, is correct.Not that modelling the origin of the universe is anything new. The conventional approach is to take thelaws of physics and extrapolate backwards from the present. From observations dating back to the 1920s,we can see that galaxies are moving farther and farther apart: the universe is expanding. By reversing thatexpansion, researchers concluded that 13.7 billion years ago the universe was in a very small, dense andhot state. The big bang theory, first proposed in 1927 by Georges Lemaître, was bolstered in 1964 by thediscovery of the cosmic microwave background - the radiation filling the universe that is thought to be arelic of the big bang - and has ruled ever since.In 1981, a major addition was made to the big bang picture. Alan Guth of the Massachusetts Institute ofTechnology and others proposed that the expansion of the early universe happened much faster thanoriginally thought. This theory, called cosmic inflation, explained the surprising uniformity of the visibleuniverse by saying that it grew exponentially from a patch that was extremely tiny to start with (NewScientist, 3 March, p 33). Though highly successful in this regard, inflation still doesn't explain the initialconditions of the universe.That's because inflation would have taken place between 10-35 and 10-32 seconds after the big bang.Going back further in time, we hit a brick wall because the two pillars of modern physics - quantum fieldtheory and general relativity - break down. Physicists don't have a complete recipe with which to concoctthe behaviour of matter, energy and space-time under such extreme conditions, and it's hard to blamethem.To get around this, some are basing their ideas around an age-old tenet. The second law ofthermodynamics dictates that the entropy of the universe - a measure of its disorder - increases with time.So the universe began in its most orderly state and has been getting messier ever since. The problem is, itwould have been more likely to be chaotic and disordered, so what was this initial state? "It'stremendously important that any respectable model of the early universe explains why entropy is so lownear the big bang," says Sean Carroll, a cosmologist at the California Institute of Technology in Pasadena.Enter the first of the new models. The entropy question has led Thomas Banks of the University ofCalifornia, Santa Cruz, and Willy Fischler of the University of Texas at Austin to conclude that theuniverse in its earliest moments - when it was less than 10-35 seconds old - was a sea of black holes. Theycall this scenario "holographic cosmology".The idea is based on the holographic principle, which was proposed in 1993 by Gerard't Hooft of UtrechtUniversity in the Netherlands and developed by Leonard Susskind of Stanford University in California.Although it is unproven, many physicists think the holographic principle is right: all the information in agiven volume of space can be represented by physical laws that exist on its surface. Entropy can bethought of as a measure of information content - the more disordered a system, the more information ittakes to describe it. Cast in these terms, the holographic principle says the entropy in a given volume islimited by its surface area, and maximised in the case of a black hole.Now imagine turning back the clock towards the big bang. Matter and energy get packed together moredensely into each shrinking region of space until we reach the entropy density limit, which corresponds tofilling up these regions with a sea of microscopic black holes.According to Banks and Fischler, the universe began as this black hole "fluid" (see Diagram). From anyvantage point, black holes would fill the entire space around, but how densely they fill it would fluctuateaccording to the uncertainty principle of quantum mechanics. A fluctuation towards lower density wouldmean that in that region the black hole event horizons would not fill every last bit of volume, but wouldhave some ordinary space between them, free of black holes and filled with radiation.This creates the conditions for our observable universe to come into existence. If the black holes in theregion where ordinary space opens up are densely packed and moving fast, their collisions and mergersmake them grow until they fill the space, pulling it back into the black hole fluid. But if the black holesare far enough apart and moving slowly, mergers won't happen fast enough. In such a region the ordinaryspace filled with hot radiation would quickly expand, pushing the black holes further apart.About 10-35 seconds after the beginning of time, this bubble of ordinary space joins up with theconventional picture, in which inflation expands our universe to more than 1 kilometre across in a tinyfraction of a millisecond. Eventually, particles condense out of the radiation to produce the buildingblocks of stars, galaxies, planets and life.So how do Banks and Fischler explain the low entropy of the early