The race to make the world's most powerful computer ever

By Mary-Ann RussonTechnology of Business reporter

Circuit board head graphic — Image captionCould quantum computing unlock secrets of our bodies and the universe itself?

Quantum computers have long been touted as incredibly powerful machines that will be able to solve hugely complex computational problems much faster than any computer we have available today. But no-one can agree on the best way to make them. Who will win the race?

Superfast quantum computers could speed up the discovery of new medicines, crack the most complex cryptographic security systems, design new materials, model climate change, and supercharge artificial intelligence, computer scientists say.

But there's currently no consensus on the best way to make them or how to make them available to the mass market.

Chromosomes — Image captionQuantum computers could help find molecules to tackle genetic diseases

Physicists, engineers and computer scientists around the world are trying to develop four very different types of quantum computers, based around light particles, trapped ions, superconducting qubits, or nitrogen-vacancy centres in diamonds.

Companies like IBM, Google, Rigetti, Intel and Microsoft are currently leading the quantum charge.

Each method has its pros and cons, but the overarching challenge is the fragile nature of quantum itself.

What is quantum computing?

Instead of using ones and noughts called bits, representing on or off, in long sequences as in classical computing a quantum bit - or qubit - uses the near magical properties of sub-atomic particles.

Electrons or photons, for example, can be in two states at the same time - a phenomenon called superposition. As a result, a qubit-based computer can do far more calculations much faster than a conventional computer.

Quantum computers 'one step closer'

"If you had a two-qubit computer and you add two qubits, it becomes a four-qubit computer. But you're not doubling the computer power, you're increasing it exponentially," explains Martin Giles, San Francisco bureau chief of the MIT Technology Review.

Media captionWATCH: Quantum computing explained in 20 seconds

Computer scientists sometimes describe this quantum computing effect as like being able to go down each path of a very complex maze at the same time.

Qubits can also influence each other even when they're not physically connected, a process called "entanglement". In computing terms, this gives them the ability to make logical leaps conventional computers never could.

The search for stability

But qubits are highly unstable and prone to interference or "noise" from other sources of energy, leading to errors in calculations. So the race is one to find a way to stabilise them for mass-production.

Computing giant IBM firmly believes that "transmon superconducting qubits" hold the most promise for quantum computing, and they have three prototype quantum processors that the public can access in the cloud.

"So far, over 94,000 people have accessed IBM quantum computers in the cloud. They've run over five million experiments and written 110 papers," says Dr Robert Sutor, vice president for quantum computing strategy and ecosystem at IBM Research.

IBM staff building the fridge to store qubits at freezing temperature — Image captionIBM's quantum computer stores superconducting qubits at extremely low temperatures

"People are learning and experimenting... we hope in three to five years to be able to point at one specific example, and say that quantum significantly improves on anything classical computers can do."

But IBM's method required the quantum computer to be stored within a large fridge, where the qubits are stored at temperatures close to absolute zero to ensure that they remain in their useful states.

This takes a lot of energy and means it would be extremely hard to miniaturise.

"It seems likely that superconducting qubits will be among the first technologies to enable useful quantum computation," says Joseph Fitzsimons, a principal investigator at the National University of Singapore's Centre of Quantum Technologies.

"However, my impression is that they are analogous to vacuum tubes in early computers, rather than transistors which came along later.

Google's Bristlecone quantum processor — Image captionGoogle has developed a 72-bit quantum processor called Bristlecone

"We may yet see another technology emerge which becomes the ultimate winner."

Microsoft and academics at the Niels Bohr Institute in Copenhagen are working on what they believe will be much more stable qubits based on so-called Majorana particles.

While other teams are working on trapping qubits in silicon - the material traditional computer chips have been made from.

And computer scientists at Oxford University are looking at ways to link smaller qubit computers rather than creating bigger computers with lots of qubits.

There are many ways to skin Schrodinger's Cat it seems.

Classical potential?

While we wait for quantum computers, what's the future for conventional, or classical, computing?

In July, Ewin Tang, an 18-year-old graduate in computer science and mathematics from the University of Texas at Austin, made waves in the international computing world by developing a classical computer algorithm that can solve a problem almost as fast as a quantum computer.

The problem involved developing a recommendation engine that suggests products to users based on data about their preferences.

UT Austin graduate Ewin Tang — Image captionEwin Tang came up with a classical algorithm that mimicked the speed of a quantum computer

And the EU recently announced it is working on the next generation of computers - exascale - which would enable a billion billion calculations per second.

"Exascale means 10 to the power of 18 operations per second," explains says Prof Scott Aaronson, a theoretical computer scientist at UT Austin who mentored Mr Tang.

"10 to the power of 18 is big, but quantum systems, which will be capable of 10 to the power of 1,000 operations per second, is much, much bigger."

And the problem for classical computing is that we are reaching the limits of how many transistors we can fit onto a chip - Apple's A11 squeezes in an astonishing 4.3 billion, for example.

Moore's Law - that every two years, microprocessors will get twice as fast, use half as much energy, and take up half as much space - is finally breaking down.

Quantum leap

Even if a stable, mass-produced quantum computer always remains elusive, the research is already yielding interesting results.

"If we hadn't invested in quantum computing, the quantum algorithm that inspired Mr Tang wouldn't have existed," says Prof Robert Young, a Royal Society research fellow and director of the University of Lancaster's Quantum Technology Centre.

More Technology of Business

Already, he says that quantum research has yielded a new way to cool devices to low temperatures; light-based chip enhancements that have improved the fibre optic broadband experience; and the invention of lab-on-a-chip technologies to speed up the diagnosis of illnesses.

"The real benefit of going to the Moon wasn't going to the Moon, it was the peripheral technologies that were developed on the way," says Prof Young - GPS satellite navigation and ball point pens, to name but a few

Unpicking the cyber-crime economy

Blank credit cards — Image captionCyber-thieves can struggle to dispose of the cash their crimes generate

Turning virtual cash into real money without being caught is a big problem for successful cyber-criminals.

They often have to get creative when "cashing out" or laundering the money they have stolen, according to a security expert.

Ziv Mador, head of security research at SpiderLabs. told the BBC that credit card thieves, for example, have limited time to profit, because at some point the victim will put a stop on their card.

Tens of thousands of stolen card numbers are traded daily on the underground markets that Mr Zador and his colleagues monitor, with details taken from compromised websites or databases.

"They can try to sell the card, which is not big money because they only get a few dollars for each one," he said.

Instead, he added, they are more likely to use them to buy more valuable assets like iPhones or Macbooks, which are popular because they tend to hold their value when resold.

"They do not buy 100 or so iPhones at once," he said. "They use a lot of different cards at different times."

Mr Mador said the crooks use randomisation tools to thwart anti-fraud systems that would spot if all the purchases, even those made with different cards, are being done on the same computer.

Another "cashing out" technique uses gift cards from big retailers such as Amazon and WalMart.

Apple iPhone X — Image captionApple's iPhones are often bought and resold by cyber-thieves

This technique involves buying the gift card with the stolen credit card and then offering it for sale at a big discount.

For example, a customer may be able to buy a $400 (£312) card for half price, although they face the risk of it being cancelled if a retailer notices it was originally bought with a stolen credit card.

Then there are the more creative scams that seek to use Uber and other ride-hailing firms to launder cash.

Mr Mador, and others, have seen adverts seeking drivers who can take part, with Spain and the US both popular locations for the fraud. Other places like Moscow and St Petersburg were "temporarily unavailable".

"They are looking for Uber drivers for fraudulent payments, people who can register for Uber and do fake rides," said Mr Mador.

The driver's account is used to launder the cash generated when stolen credit cards are used to pay for the fictitious journeys and they get a cut of the money as a payment.

It is these markets that form the backbone of the cyber-crime world, said Dr Mike McGuire, a criminologist from the University of Surrey, who has studied this shadowy community.

"We took a holistic look at the criminal economy and then we could see where the flows of money were going," he told the BBC.

Some was laundered via banks and other well-established routes long used by criminal gangs, who increasingly have been finding ways to use newer technological methods.

Dr McGuire's research suggests billions in criminal cash passes through underground markets each year. Some of that is just thieves selling to thieves but other methods involve the sale of drugs and other contraband.

Through conversations with convicted crooks and the police who pursue them, Dr McGuire said it was clear that some were involved in the trade for very mundane reasons.

"It's a very human set of activities that these people are involved in," he said. "About 15% were just using their revenues to pay their mortgages and their bills."

Others, those who were making a lot of money, had got involved in "old-fashioned ostentatious spending", he told the BBC.

"A lot of them are converting their money to assets and investing in them to acquire status."

Paper trail

Banks are getting better at spotting money laundering that uses property and fake corporations, said Rob Horton, from BAE's Applied Intelligence division that helps financial firms spot fraud.

This is not straightforward work, he explained, because the crooks worked hard to obscure their ownership of the bogus firms.

But, he said, detailed long-term analysis of the information shared by front companies can help unpick the relationships.

"We will often find the contact details and registration addresses for these facilitators are the same across dozens of applications," he said. "That's because it's very hard to genuinely create that many completely new corporate or individual identities.

Image copyrightREUTERS

Image captionOne money laundering effort sought to recruit Uber drivers

"As a result they tend to reuse the same artefacts."

Industry efforts to get lots of legitimate financial organisations sharing data about the organisations behind payments and purchases were also helping to uncover the front organisations, he added.

"It takes a network to defeat a network."

That long-term data-driven approach is also helping banks to pick out the low-level recruits some criminals are using as "mules", said Kedar Samant, co-founder of fraud-spotting firm Simility.

Mules are used by a lot of criminal gangs to get at cash generated by other means - often ransomware attacks or phishing campaigns.

Some gangs approach people who have done manual low-paid work abroad and offer to buy access to the bank account they set up when in that country.

Old school fraud detection systems would struggle to spot cash laundered through this route, because they were very "brittle", Mr Samant said. They tended to look for anomalous behaviour rather than consider the context around the account, how it is used over time and where cash goes.

That long view was becoming crucial because many criminals were happy to take their time to learn the best ways to defraud a bank by testing it with "short bursts of attacks" that they then pull back from to see if the loss has been noticed.

"Your customers may come and go but fraudsters tend to be very loyal," Mr Samant said.

"That's because it takes time to learn a bank's systems and how they can get away with the crimes."

lunar crater architecture

Monday, September 3, 2018