If any of us ever get to surf a quantum Internet, it might be because of the research of people like Jeff Kimble of Caltech. The problem most researchers encounter in trying to develop a rudimentary quantum computer is that of scaling up the number of atoms in strange quantum mechanical states that can serve as qubits--the quantum analog of an ordinary bit of computer data. Building devices with a few qubits isn't all that hard; building devices with the 50 or so qubits deemed necessary for solving important problems faster than conventional computers seems horrifically daunting. But what if instead of one device with 50 qubits, you hooked together five devices with ten qubits each? That's where Kimble's quantum network comes in.

Kimble starts with an atom trapped in two simultaneous quantum states by electromagnetic fields inside a chamber. With the right combination of electromagnetic fields, Kimble can make such an atom emit a photon of light energy that will itself be in a superimposition of different states of "polarization"--a characteristic that can very roughly be thought of as a clockwise or counterclockwise rotational orientation of the photon. And, as is necessary for quantum computing logic, changes to the state of the atom will affect the state of the photon, so that the atom and the photon can serve as a two-qubit system.

Even more important, the photon qubit can then travel on--at the speed of light, of course--to another atom, become absorbed by it, and thereby push the second atom into a qubit state. In other words, the first atom-qubit can affect a second atom-qubit via the photon-qubit, creating a three-qubit system. In principle, a chain of atom-to-photon-to-atom qubits could be fashioned this way.

Kimble proposes to create devices with perhaps a handful of qubits and then link them with fiber-optic cable. Then the atom at the end of the chain of qubits in one node can pass a photon qubit through the cable onto the first atom of the chain in the next node, in effect allowing the two short chains of qubits to operate as one longer chain. "If you can wire two nodes together quantum mechanically," says Kimble, "you get computing capabilities that are exponentially larger than the two nodes operating independently. It's a way big difference." If that sounds too good to be true, just consider that a pair of ordinary numbers gives you 100 possibilities (00 through 99) and adding a second, separate pair gives you another 100 possibilities, for a total of 200. But if the two pairs can be linked into a single four-digit number, you'd have 10,000 possibilities (0000 through 9999). In essence, that's the sort of linking that Kimble's system would provide.

Okay, so Kimble is still struggling just to get a single atom to transfer a photon to another atom down the hall. "You've got to start somewhere," he says. "Nobody knows which path will turn into the yellow brick road."