Originally posted to 13.7 Cosmos & Culture on 4/9/2014


In the past two weeks, I explored here some of the consequences of the remarkable observation by the scientists of the BICEP2 experiment in the South Pole. The data, potentially revolutionary, points to a period of extremely fast cosmic expansion at the very beginning of time, a signal imprinted in the cosmos for 13.8 billion years.

This is really mind-boggling.

When I was starting my career in the mid-'80s, I never expected that the theories and models that my colleagues and I were working on could be tested in my lifetime. It was all way too eccentric, too far-removed from our everyday reality. And now, here we are, testing and putting limits on models that deal with physics at energies that are trillions of times higher than our most powerful machine, the Large Hadron Collider, in Switzerland, can probe directly.

But before popping the champagne bottle open, we need to make sure we are not being duped. (And this goes to many scientists who have already opened their bottles.)

This kind of observation is very different from what we can do in the laboratory, where we can control the system under study. We can't control the universe, and much less so near the beginning of time. On the other hand, observing distant objects is astronomy's bread and butter, and there is nothing controversial in this. It's true that we can't control stars in the lab (unless we do it through computer simulations, like 13.7's own Adam Frank does). But we can study them methodically by collecting the light and other kinds of radiation they emit.

Still, there is an essential difference between observing galaxies and stars — through direct study of the radiation they emit — and observing an ancient signal impressed on the cosmic microwave background radiation (CMB), as is the case with the BICEP2 experiment. Recall that the CMB radiation is the leftover fossil from the epoch when the first atoms of hydrogen were formed, some 400,000 years after the Big Bang that marked the beginning of time. The signal that BICEP2 detected was imprinted on the CMB radiation from events that happenedmuch earlier. It is as if we were measuring signals from radioactive isotopes — that existed pretty much since the origin of the Earth — on the blocks used to make the pyramids of Egypt, built much more recently.

Every scientific observation must be carefully scrutinized to avoid errors. It's easy, when the exciting data comes in, to get carried away. But nature, of course, doesn't get carried away. The BICEP2 team was extremely careful and meticulous in its data analysis, trying to eliminate all sorts of errors, including possible signals from galactic sources. And we do hope that the results survive, even though the signal is way more powerful than what has been so far reported, even surprisingly so. Many people remain skeptical, as they should be.

If the results prevail, many theoretical models that attempt to describe the mechanism that would propel the cosmos into a period of accelerated expansion would be wrong. In fact, most would.

For example, a model by the Italian physicist Gabriele Veneziano called "pre Big Bang" would be ruled out, as would models called ekpyrotic, where the cosmos emerges from the collision of flat slabs of space called "branes," as a mythic god that would create and destroy universes by clapping his hands. Also, most superstring-inspired models of the fast expansion would be in serious trouble.

This is all good; ruling out models with data is precisely what a mature science should do.

The curious aspect of all this is that the data won't be able to do more; we can't use it to select which model is right, only to eliminate those that are incorrect. This is a lesson on how science works: We can only rule out models, prove them wrong. The models that do work only do so temporarily, insofar as they survive testing.

This is a very peculiar situation, if you think of it. It means that we really don't ever know the "truth" of how nature works. All we have are approximations, tricks that explain the data we have at hand. Accepted theories are simply narratives that are good at describing what we can measure.

The 17th-century French thinker Bernard de Fontenelle knew this well when he wrote that "all philosophy is based on two things only: curiosity and poor eyesight ... the trouble is, we want to know more than we can see."

The reader should thus beware of claims of full understanding, especially when it comes to the boundaries of knowledge, as is the case here. For example, BICEP2 results could neverprove the existence of the multiverse, the hypothetical expanse that encompasses many universes, including our own. In fact, nothing can. We can say that our current models make the idea of an ensemble of universes compelling, but we cannot claim we know the multiverse exists.

There is a persistent fog out there, and we would do well to celebrate our achievements for what they are and not more.