100 Million Times Faster? Quantum Computing Is In the House
Last Tuesday, the Google Research Blog announced the publication of a new paper by Alphabet’s (NASDAQ: GOOG) Quantum AI team, reporting on the results of experiments run on the quantum computer system that GOOG operates jointly with NASA. That system was produced by Canadian company D-Wave Systems; we wrote about Google’s collaboration with NASA in our letter on October 8.
What Is Quantum Computing, and Why Does It Matter?
Quantum computing aims to exploit some of the bizarre and counter-intuitive aspects of sub-atomic physics to create computer systems that are millions (or billions) of times more powerful than even the fastest of contemporary supercomputers. Those quantum behaviors include such things as superposition, in which a system exists in all potential states simultaneously; entanglement, in which particles, once they’ve met, remain “attuned” to one another no matter how far apart they are; and tunneling, in which particles can move from one place to another by teleporting through an intervening obstacle. (These are all observed realities of physics, not just theorizing on the part of physicists.) The prospect of exploiting quantum effects in a computer was first proposed by researchers at Oxford University back in 1985.
If such quantum machines could be built, they would have momentous effects on every data-driven industry — which in a modern economy, means every industry. Problems that have so far remained unsolved by engineers and mathematicians would be solved; so would problems that have not yet been conceived. The “Turing test” would be definitively passed — which is the ability of an artificial intelligence to convince anyone communicating with it that it is human. AI would move from being embryonic to being omnipresent in the lives of consumers and deciders.
The key idea is optimization. Most problems in financial, industrial, and consumer enterprises are optimization problems — given a problem with a large number of variables, how do you arrive at a solution which optimizes your desired outcome, or comes close enough for your satisfaction? (A simple example of this kind of problem is the “travelling salesman problem,” in which a person moving from place to place tries to determine the path he should take to visit every town on the map in the shortest time. Add enough stopping points, and it becomes a very difficult problem to solve quickly — so difficult that it has occupied generations of mathematicians.)
The problem with optimization is that in the real world, there are often tens of thousands of variables that are relevant to a given problem, so with current technology, exhaustive or even sufficient optimization is simply impossible within a reasonable human timescale. This is one reason why, even with all our technical knowledge, we still suffer traffic snarls, bridge failures, wireless outages and slowdowns, building collapses, and airplane crashes — and why we need to have expensive redundancies built into our systems to try to avoid such problems. (This is also why intuition remains a critical gift in some areas where sufficient optimization is out of reach.)
Quantum computing, fully realized, would put the optimization of unimaginably complex problems within reach. The most complex logistics and distribution problems would be solved almost instantaneously. The human presence in consumer-facing services would be rendered obsolete by perfectly reliable speech recognition technologies; those same technologies would allow voice computer interfaces to become universal and completely reliable. Telecom operators and internet service providers could optimize voice and data traffic to boost the efficiency of their networks exponentially. The analysis of big data to pinpoint and predict consumer desires and behaviors would become so effective and robust that it would prompt political turmoil over privacy concerns that will be far more severe than anything seen so far, even in Europe, where the issue has come under particularly close scrutiny.
In this regard as in others, we should bear in mind the paradoxically negative effect of a backlash against the uses to which the new technology might be put. That backlash would likely be not just popular, but regulatory, in the case of high-frequency trading algorithms using quantum systems — and then we could be treated to a scorched-earth war between technology and its would-be regulators.
Finally, as we commented in our letter on November 5, the full realization of quantum computing would spell the immediate end of the encryption methods that all businesses and governments rely on for secure communication, whether political or financial. (Such a development would, incidentally, also kill Bitcoin and any other cryptocurrency relying on similar technology.)
Those are just a few reasons why the potential arrival of quantum computing is something worth investors’ attention. Not the level of attention given to a transformative technology whose implementation could be imminent — but certainly something to keep “on the radar” and check in on periodically to keep a current view of the lay of the land.
What Has GOOG Achieved?
Although D-Wave produced its first quantum computing system in 2011, controversy has swirled around the company and its products, with skeptics maintaining that the machines showed no superiority to conventional systems and that no quantum effects were actually being exploited.
D-Wave’s Quantum Systems — Real or Not?
Gradually that skeptical position has become less tenable, and the paper that GOOG published last week and reported on its blog does indeed represent conclusive evidence that “something quantum is going on.”
Put simply, the GOOG/NASA D-Wave machine implemented a very specific form of optimization on a very specific subset of problems. Within those parameters, the D-Wave optimization process was 100 million times faster than its classical counterparts. GOOG says simply that the debate about whether quantum effects are actually being exploited in the D-Wave machine has now been settled. So, check: there is a quantum computer in existence, and it is functional.
However, D-Wave’s particular engineering strategy might have resulted in a dead end, where it will never excel against all its classical competitors. The D-Wave as programmed by GOOG’s engineers only outperforms with this specific heuristic, on this specific set of problems. Compared to strictly comparable classical computation methods, it excels by orders of magnitude — but there are other classical methods that it still can’t beat. MIT engineering professor Scott Aronson comments, “I think of [D-Wave] as taking the dirty approach… Of course, it is possible that the dirty approach will get somewhere before the clean approach does. There are many precedents for that in the history of technology, where the dirty approach wins. But it hasn’t won yet.”
So to sum up: D-Wave’s machine is indeed a quantum computer, and GOOG has successfully built and run an optimization program on it that beats the corresponding classical methods by a factor of 108. Whether D-Wave’s architecture will allow programs to be run that excel outside this narrow framework remains a subject of controversy — but the reality of quantum computing is now no longer subject to doubt.