The Quantum Graph

Month: November, 2012

Dear Sir: Plea$e give us ¥our Mon€y

Yesterday I tracked down mining tycoon Rob McEwen. He infamously turned Goldcorp, a relatively small Canadian gold mining company into the world’s second largest — from $50 million to $10 billion market cap. Rob spoke at a Scotiabank mining conference at the King Edward hotel in Toronto. After unsuccessfully trying to haggle my way into the non-public conference ballroom, I sat in a stately lobby with fat Greek columns next to a giant Christmas tree and waited for him to come out. Recognizing his face from a Bloomberg interview, I noticed and followed him upstairs to the mezzanine. I approached Mr. McEwen at the snack table and got my introduction in before the nice gatekeeper lady who denied me earlier came to shoo me away. Luckily he’s a cool guy; we went back down to the lobby and sat near the pine tree to discuss a mixture of medicine and technology.

McEwen is familiar with what can come of good data and good software. In 2006 Canadian Business magazine named him the ‘Most Innovative CEO’. He’s also an investor, and as Noo Corp is seeking investment, I thought it wise to pitch him.

Athabasca tar sands, Alberta

Harvesting natural resources is probably an older profession than prostitution. It’s the backbone of any given civilization – famous or forgotten. When supplies collapse, so do empires. In recent times, as the Earth’s population has exploded, demand for things like copper or cadmium has increased dramatically. As a result, natural resource extraction has gained a well-deserved reputation as being downright dirty [July ’13 edit: it is truly ugly]. Yet with companies like Planetary Resources looking to the heavens as the next great frontier for the elements we use, it seems fair to say that our appetite for raw stuff is insatiable.

As a side note, before the San Francisco / Bay Area became Silicon Valley, it was the site of a great mining rush, as pointed out in this fascinating talk (thanks Pete).

Back to the story, McEwen’s insight was that the mining industry is mostly aloof regarding its use of data. This was a light bulb moment: noospheer has finally found its niche. Unifying the entirety of a given mining corporation’s data — both geospatial and logistical — implies an increase in efficiency. Overlaying this with open data on a given parcel of land from the wider network implies an increase in awareness. Efficiency and awareness in this context means less environmental impact. By integrating multiple data fields gathered from increasingly non-invasive exploration technologies, we can get more out of this planet while scarring her surface less. In the future, we should one day be able to teleport gold right from within the Earth’s crust.

Noospheer’s vertical is the natural resource mining industry as its a mess, and its data is too. Clean up the data, clean up the world. We’ve now met and spoken with numerous mining software companies for feedback, raw test data for piloting the system, and agreement to run the beta once ready.

Was Noo Corp successful with the ask? I’ll let you know.   ~Jordan

[July ’13: not yet]

Executive Bot

If you wanted to get rich, how would you do it? I think your best bet would be to start or join a startup. That’s been a reliable way to get rich for hundreds of years. The word “startup” dates from the 1960s, but what happens in one is very similar to the venture-backed trading voyages of the Middle Ages.

Paul Graham, Hackers & Painters

Web software companies are interesting from an economics perspective because they maximize leverage. If a piece of software becomes popular, the company behind it often becomes enormously successful.

Like any business, web companies have expenses. The two cold goals of any corporation are to turn a profit and simultaneously reduce overhead. In the recent past, a web-based company would have to buy and operate pricey server hardware. New web companies don’t bother maintaining hardware, they pay other companies such as Amazon do that. A cluster of 4 GPUs (6144 total cores) running at 100% utilization, with 1tb of storage, 1tb of data transfer in and 1tb of data transfer out comes to well under $2000 for the month. A great example of a company that uses ‘the cloud’ to host its services is soundcloud.

Besides compute/bandwidth costs, a web company must also maintain a payroll. At last check, Google was paying 53,546 individuals. With an average starting salary of $82k, that’s roughly $366m in pay per month. Also, bigger companies tend to get involved in expensive lawsuits.

So, with hardware being cheap and easy to deploy, how can a web company reduce its employee and lawsuit overhead? Open source, of course! Open source companies generally avoid lawsuits because they are never accused of patent infringement. They also don’t need as many employees for two reasons: 1) the use of open code written by others and, 2) free development from the wider community. By using cloud hosting services and open source, a startup may keep its core team small, outsource hardware, and in theory remain nimble / efficient well into maturity.

With miniscule expenses and some sort of scaling model (catch 22?), the coins flow freely. Business can avoid the inevitable clinging and bloat that comes with success.

The next question is, “will the role of corporate Officer soon be non-human?”

Assumptions (tsk tsk)

Human beings like to make assumptions. Assumptions seem to make the world a less demanding place to understand. Ironically, no kind of human is more susceptible to falling prey to assumptions than the scientist. In the strict definition, a scientist prides him/herself on proof through rigorous observation of well-constructed experiments.

However, if a question is hard, it’s difficult to devise an experiment which can answer it. When this is the case, many scientists fall back on beliefs which they assume to be true. This is unfortunate, as the universe is mysterious, and assumptions generally turn out to be false (ex: “all heavenly bodies orbit the Earth”).

A modern scientific assumption is that factoring is hard. Factoring involves taking a number and determining which two prime numbers multiply together to equal the original number. For example, 21 factors into 3 times 7. Despite the simplicity of the problem, there is no obviously efficient way to solve it. So, if the original number is sufficiently large, a computer may take a billion years to get the answer. Due to its hardness, factorization is at the heart of public key cryptography (PKC).

PKC is responsible for keeping computer systems secure. Virtually all institutions encrypt data with PKC. Virtually all ‘secure’ online transactions and digital signatures rely on PKC for their relative security. Symmetric-key cryptography (SKC – using a secret key that cannot be guessed logically) doesn’t directly rely on PKC, but the problem with SKC keys is a critical requirement that they be shared by both parties. This implies a secure connection must be established in order to share the private key in the first place — unless you share the key off-network. Simply, SKC needs factoring to be hard too.

Shor’s quantum algorithm factors in polynomial time (fast). Thus, Shor’s discovery ruins current security systems. However, it is a quantum algorithm and requires a quantum computer. Recently, Shor’s algorithm was implemented on a quantum computer and succeeded in factoring the number 21 in polynomial time. However, PKC uses numbers much larger than 21. So, it would be interesting to see whether Shor’s algorithm can be efficiently implemented on a classical computer, such as an iPhone.

It is proven here and extended here, that at least one major component of Shor’s factorization algorithm – the quantum Fourier transform (QFT) – can run efficiently on classical computers. The harder part of Shor’s is something called modular exponentiation (ME), and this is commonly assumed to be an impossible task for classical computers. However, it is also now proven that ME can be broken down and reconstructed in terms of the QFT. Therefore, the totality of Shor’s algorithm can run on an iPhone. If you find this alarming, rest assured such a breakthrough brings QKD much closer to fruition.

There are deep subtleties in the papers cited above, which we are carefully investigating. Should our assumption that factoring is classically easy prove true, we will publish our findings here.