Almost 40 years ago, Gordon Moore made his observation that thenumber of transistors per square inch of printed circuits hasdoubled every year since the integrated circuit was invented. Thisobservation generally has been extrapolated into the future topredict our seemingly endless ability to create smaller and smallercircuits.
Current technology is creating chipsets with features that are 130nanometers in size (a human hair is about 60,000 to 100,000nanometers in diameter). The atoms in a silicon wafer (on whichmicrocircuits are built) are spaced about 0.235 nanometers apart.Obviously, we rapidly are approaching technologies that will beable to create structures at the atomic level.

Which brings us to a very interesting problemcurrent computertechnologies are based upon Newtonian physics, and Newtonianphysics doesnt work at the subatomic level. Interactions amongsubatomic particles are fundamentally different than interactionamong particles of molecular size and larger. Now, I suspectminiaturization of conventional computer circuits will be limitedby economics not science. But notice I said conventional computercircuits. If we look at the physics of subatomic particles, wediscover there may be other very powerful ways to process data. Onesuch theory is broadly known as Quantum Computing.

Physics 101

I think I can safely say that nobody understands quantummechanics.
Richard P. Feynman

Thats reassuring, coming from ar-guably the most brilliantpost-Einstein physicist. You can probably assume I do not intend topresent an in-depth study of quantum theory in these two pages.Fortunately, we can understand the basic principles of quantumcomputing with just a little bit of knowledge (which we all know isa dangerous thing). Max Planck started all this in 1900 when he wasinvestigating black body radiation. Traditional physics teachesenergy transfers among all particles are continuous (theacceleration of a body is directly proportional to the amount ofenergy applied). What Planck discovered is that at the subatomiclevel energy transfers are discrete.

When I was first introduced to atomic structure in school, wewere taught electrons orbited around the nucleus like the eartharound the sun. Newtonian physics teaches if the velocity of anorbiting body increases by applying additional energy, the radiusof the orbit will increase. Not so with electrons. There is nocontinuum through energy levels and thus orbits around the atomicnucleus. Electrons are observed at discontinuous levels around anucleus, and those levels are achieved with the transfer of adiscrete packet or quanta of energy. Plancks discovery states anyenergy transfer between two bodies is the sum of elementary butfinite transfers of quanta.

Let me make one more observation. When we observe a subatomicparticle, our observation forces that particle into a particularstate. Prior to our observation, there simply is a probability theparticle will be in that state. Interaction such as observationreveals a static subatomic universe, but the reality of thatuniverse is not static. In fact, it can be in a multitude of statesat the same time. We will come back to this later and explain someof this apparent contradiction.

Classic Computing

All present-day computing is based on principles and algorithmsthat predate the microprocessor. Data is represented as a series of1s and 0s. It is processed by being acted upon by gates. A not gatechanges a 0 to a 1 and vice-versa. An and gate takes two inputs andreturns a 1 only if both inputs are 1, etc. All the systems wecurrently use are based on these concepts. Data (that series of 1sand 0s) just as easily could be represented by a string of toggleswitches or marks on a rock as by voltages in a microprocessor.

Modern computers have achieved great efficiencies of scale andspeed but are still working in a serial world. Take this piece ofdata and do this to it; then take it and do that to itthats whycurrent cryptographic schemes are considered secure. The sites onwhich we carry out e-commerce are all secured using some sort ofsecurity system based on the difficulty of factoring large numbers.Algorithms for factoring numbers really are just brute forceattacks that repeatedly try different numbers after the obviouslosers are factored out.

Parallel computing may allow us to process complex processesfaster but is still limitedwe still are processing everythingserially, only with multiple processors or machines. Assume we aretrying to factor a 100-digit number. At some point in the process,we are going to test a particular 32-digit number, then another32-digit number, then another, etc. What if we could simultaneouslytest all possible 32-digit numbers at once? We could reduce thetime needed to factor large numbers from years to hours (or less).And that is what we hope to gain from quantum computing.

Bits and Qubits

This is the tricky part. We all understand bits. They are 0 or 1.There are no other alternatives. Enter the qubit or quantum bit.Just as the bit is the fundamental unit of information in digitalsystems, the qubit is the fundamental unit of information in aquantum system. A qubit can exist in a state that corresponds tothe classical logical states of 0 and 1. It also can exist in astate called a superposition that may be described as a blend ofthose statesand it can also exist simultaneously as 0 and 1. Let meexplain.

Imagine a qubit as a sphere, not unlike the earth. Now imagine avector (or arrow) drawn from the center of the sphere to the NorthPole. Let that vector represent 1. Imagine a second vector drawnfrom the center of the sphere to the South Pole. Let that vectorrepresent 0. According to classic quantum theory, if we apply acertain amount of energy to the qubit in state 0, we can flip it tostate 1. Thus far we have described a system that could be used forpresent-day digital computing. We can switch the qubit from 0 to 1and back.

Now, lets assume instead of applying the proper quanta of energythat would flip the state from 0 to 1, we apply an amount of energyless than that discrete packet. What we have created is a new statethat is neither 0 nor 1. In reality subatomic particles exist instates that are probabilities. It is only when they are acted uponby an external agent (as when we observe them) that they appear ina discrete state. If our vector or arrow was pointing to a spot onthe surface of the sphere that corresponds to a latitude of 45North (where it would be in a state of superposition), then theprobability we would observe that qubit as a 1 is much greater thanwe would observe it as 0.

Now, assume we apply just the right amount of energy so that ourvector or arrow points to a spot at 0 latitude (on the equator).The probability that it will be a state of 1 or 0 is equaland thequbit is actually in both states at once. Really! This is notconjecture or speculation. Experiments using light (photons) andvarious split mirrors prove subatomic particles are able to traveldifferent paths simultaneously.

You notice I have defined a qubit abstractly, just as a bit isan abstract concept. It doesnt matter whether a bit is representedas a voltage, light, or stroke on a chalkboard. The understandingof a bit as a two-state device is enough_to build workingalgorithms. Likewise, it doesnt matter how a qubit actually isachieved. Whether it is done by measuring the spin on an electronor the charge on a photon is completely immaterial to our abilitiesto imagine and design quantum computing systems.

This Is the Cool Part

Lets hope we have a rudimentary understanding of a qubit. We willstart with a 512-bit piece of dataa string of 0s and 1s 512 unitsin size. This chunk of data has 2512 possible states. Using digitaltechnology we can perform an operation on a single state of thatdata at a time. If we want to perform that operation on everypossible state of that data we need to perform 2512 separateoperationsone at a time.

Pardon me while I fire up my quantum box. It now is possible toset each of the 512 qubits to represent both 0 and 1. It now ispossible to perform our operation on all 2512 states at once. Sinceeach qubit is a state of 0 or 1 simultaneously, every possiblestate of that data can be acted on in a single process. Thatsparallel computing without multiple processors but with the data inmultiple possible states. A complex processor-intensive operationlike factoring large numbers becomes easy. Current encryptionschemes become worthlesscomputing is changed forever.

OK, there you have a grossly oversimplified example of onepossibility that quantum computing presents.

Niels Bohr commented on Max Plancks research, saying: In thehistory of science there are few events, which, in the brief spanof a generation, have had such extraordinary consequences asPlancks discovery of the elementary quantum of action. It hasbrought about a complete revision of the foundations underlying ourdescription of natural phenomena. The same sort of statement couldbe made regarding quantum computing.

Fact or Fiction?

Did I mention quantum computers dont exist? True, true, butsmall-scale qubit operations have been demonstrated. Quantumcomputing began as a mental exercise. Richard Feynman firstproposed the idea of a quantum computer in 1981. The question was,If we use quantum states to encode and process data instead ofmacroscopic states, will it change our understanding of informationtheory? That early speculation has evolved into reality. Cor-porateAmerica is allocating R&D money to build these things. LucentTechnologies (at the famed Bell Laboratories) and IBM are just twoof the players. We may be years away from working machines, butwith the enormous revenues anticipated from such a radical change,research will continue. Every so often one sees an article about a5-qubit device that actually did something it was supposed to dofor a brief period of time. To me, that is encouraging.

What About Us?

Hey, we sell insurance. What the heck do we care about factoring200-digit numbers? Do you have gigabytes of data that you cantproperly analyze? Do you want to reduce the risk on each policy youwrite? Quantum computing will support incredibly complex datamining. It will provide you the ability to better identifyhomogeneous risk groups. It will allow you to determine patterns inclaims instantaneously. Who knows what possibilities it may bring?The insurance industry always has been at the forefront of computertechnology, and I suspect it will be there when quantum computingis a reality. If not, it is a wonderful mental exercise to imaginethe possibilities.