Richard Feynman made it clear to me in his masterful book, QED: the Strange Theory of Light and Matter. The world is weird. Here’s Feynman’s elegant demonstration.

A laser beam fires photons at a half-silvered mirror. We know very well the properties of photons, and we know from the properties of laser light that the photons from a laser beam are identical to one another. Yet the observation of what happens at that mirror reveals an enormous puzzle.

Half the photons will pass through the mirror. Half will bounce off. How can this possibly be? What possible mechanism can cause identical photons to behave in such a contradictory way? The traditional quantum mechanical answer is, no mechanism at all. The real world is fundamentally random. Set up a device that will kill a cat if the photon is reflected and will spare the cat if the photon passes through, and all you can say is that half the time you’ll have a dead cat. But how? Why? Traditional quantum mechanics goes no further.

I’ve become fascinated with another approach, called many worlds (or the multiverse by David Deutsch and Brian Greene – same thing). Five recent books by five authors have raised more questions than answers, but the journey has been exhilarating.

First came Greene’s book, The Hidden Reality. He talks about many different conceptions of multiple universes, and all made for interesting reading. But the chapter I’ve returned to again and again is Chapter 8 on the Quantum Multiverse. In two long and intriguing end-notes, Greene reveals that he is a proponent of many worlds, and that there is a strong mathematical underpinning to the theory. Interestingly, just reading the text and ignoring the end notes, you might not get this impression.

Next were two books by David Deutsch. I’ve written extensively about how The Beginning of Infinity has changed my worldview. The multiverse Deutsch describes is becoming more and more a part of that view. Chapters 11 and 12, which discuss the quantum multiverse, were in many ways quite difficult, and I’ve reread them several times. They are unwavering in their dedication to the multiverse, and tell much of the story that I’ll outline below.

After Beginning of Infinity, I read Deutsch’s earlier book, The Fabric of Reality. It reads to me almost like a rough draft of Beginning of Infinity. But in chapter 2 of that book, Deutsch describes the straightforward interpretation that leads to the multiverse, including a piece of evidence that is still resonating in my head (also described below).

Next I found a book by Colin Bruce, an Oxford University colleague of Deutsch. The book is called Schrodinger’s Rabbits. Bruce draws heavily on the ideas of Deutsch and others in this book. He also touches on the profound probability problem many worlds faces, and discusses the various suggested solutions. More on Bruce’s curious book in future posts.

Finally I read a book by Brian Cox and Jeff Forshaw called The Quantum Universe. These authors very nearly ignore many worlds until they admit in Chapter Ten that they are essentially using the many worlds approach to obtain their results. Rather than examine the implications of this, they simply quote the oft-cited command to “Shut up and calculate!”

So what’s all the fuss about? I go back to Deutsch and his description of a device called the “Mach-Zehnder Interferometer”.

First (as Feynman describes in QED) a photon enters a beamsplitter. But now we follow the consequences of this apparently random result. In this case the photon is either transmitted (path D) or reflected (path U). Either way, it next moves to a second beamsplitter and then on to a detector.

Now things start getting weird. In every case, the photon shows up at detector 1, never at detector 2. This is due to interference – a careful accounting of how many times the photon’s phase flipped (at each reflection) or was slowed (at each transmission) shows that no photon can be detected at 2.

Wikipedia has a nice description of this if you’re interested:

The obvious question is, when you’re firing one photon at a time, what is interfering? The many worlds answer, described by Deutsch quite elegantly in chapter 2 of The Fabric of Reality, is “shadow photons.” These have all the properties of real photons, except that they cannot be detected except in their ability to create interference.

Here’s the part that blew me away. Suppose you block one path (maybe path labeled U in the diagram). Now the photon has to move through the interferometer along a single path (path D in this instance). Suddenly photons, arriving only from the lower path, show up in both detectors. But what, exactly, does it mean to “block” one of the paths?

A path is blocked if the photons along that path would be absorbed by the material in the way. OK, let’s suppose we’re using a red laser for this experiment. If we put a red filter (one that allows only red light through) on path U, then we get no effect. Understandable, since essentially adding a red filter changed the setup for red light not at all.

Now replace the red filter with a green filter. The green filter lets green photons through, but it absorbs red photons. Obviously, you’d expect half the red photons (the ones directed to the green filter) to be blocked and not make it to either detector. But now consider a red photon that does reach a detector. It must have traveled along the unblocked path.

What happens when that red photon gets to the second half-silvered mirror? Half the time it goes to detector 1 and half the time it goes to detector 2. Think how strange this is! Take the green filter away, and detector 2 falls silent. No photons reach it. But put the green filter in place and detector 2 wakes up again, catching half the photons that pass through the machine.

You might recognize this as just another instance of the double-slit experiment, and you’d be right. Everything I’m saying here about the Mach-Zehnder Interferometer works as well for the double-slit.

Consider what must be happening to this red photon when the U path is blocked by a green filter. It reaches the first beamsplitter and passes through, following the D path (if it instead were reflected, it would be absorbed by the green filter and disappear, so only half the time will we get the photon we’re interested in). Our red photon then bounces off the lower mirror and heads up to the second beamsplitter. Once there, it either passes through again (reaching detector 2) or reflects off the second beamsplitter (reaching detector 1). It’ll do each about half the time.

But if the green filter is replaced with a red filter, when our photon reaches the second beamsplitter interference effects cause it to pass only, always, every time, into detector 1. Somehow, when our red photon reached the second beamsplitter it “knew” not only that there was some object on the U path (a path that it could not have taken), but also what color that object was!

Traditional quantum mechanics would state that the color of the filter must be taken into account when deriving the entire wavefunction of the system. That’s OK when you think of a laboratory, maybe. But there’s nothing (in principle) to stop us from building a Mach-Zehnder interferometer the size of the solar system, or the galaxy. Are we to believe that the photons here, in our lab, somehow “know” what color the filter was stationed several light-hours (or even several thousand light years) away?

To me, the many worlds interpretation, in which the “shadow photons” Deutsch describes are real, makes much more sense. The shadow photons act like real photons. They react with filters exactly as real photons would, and yet they are undetectable except due to the interference effects they display. I know this isn’t in any sense proof, or even a particularly sound argument (after all, any barrier is translucent to some sort of photon). Somehow, though, for me it really makes an impact.

The Mach-Zehnder interferometer behaves exactly as if the photon really does split into innumerable photons at the first beamsplitter, with half taking the U path and half taking the D path. If there’s a green filter on the U path, all those photons are absorbed and photons on the D path enter both detectors. If there’s a red filter, instead, the photons on the U path fly straight through, meet the photons from the D path, and interfere so that only detector 1 ever records a photon. Many worlds gives us a straightforward way of understanding what happened and why.

But wait. If those shadow photons are absorbed by the green filter, can’t we detect that? Can’t we measure a temperature increase, a momentum shift, something? The answer is yes, if those shadow photons are absorbed by a green filter in our universe. But in fact no such effect has ever been found. Instead, the shadow photons are absorbed not by an object in our universe, but by a shadow filter. And if a shadow filter, there must be a shadow laboratory with shadow scientists who can now detect that shadow photon with their own shadow instruments. Taken to its logical conclusion, many worlds reveals a universe in which even we are but one instance of a vast collection. It seems utterly incomprehensible, yet that’s where the logic leads.

Now that I’ve written about the basics of many worlds, I’ll explore those logical, but utterly bizarre, conclusions in future posts. This one’s long enough already. At least in this universe.