Canary Wharf, London

London is the biggest financial market in the world but New York is the most automated with about 80 per cent of its trades completed by machinesPhoto: Lars Plougmann

Written in Greece at Ithaca harbour and dispatched to silicon.com via a 2.3Mbps 3G connection three days later from a bay off Meganisi.

No doubt about it. Our world is speeding up and latency is being minimised. However, our past experience in the physical and virtual domains says instability usually lies at the end of this road, if not along the way.

A new undersea cable linking the UK and US is being planned to follow the great circle route for the first time, and may be dedicated to just one industry – finance.

How come? It will cut six milliseconds off the flight time for transiting photons – about a 10 per cent improvement – which could translate into hundreds of millions of pounds in profit in high-frequency trading markets.

As a precursor, a direct cable between New York and Chicago was laid last year to shave off three milliseconds from the existing more circuitous route.

While London is the biggest financial market in the world, New York is the most automated with about 80 per cent of its trades completed by machines. Each operation has its own trading algorithms and network, and traders know first-hand what a difference a few microseconds can make, let alone milliseconds.

High-speed traders make money from small transient offsets and instabilities in markets. In this fractional trading, fractions of a second matter, and the race is on for the fastest hardware, software and networks.

Chip, circuit and system designers have been making steady progress in this arena for decades, but their contributions in the race to zero latency are measured in nano- and microseconds.

The new transatlantic cable will eclipse everything that has gone before. But of course, as each institution signs up for this service, the end point is a zero-sum game with everyone back to square one.

Even the financial markets cannot speed up photons, or increase the speed of light – no amount of money affects the basic laws of physics – and zero latency will never be achieved.

In a vacuum, photons travel at 299,792,458 metres per second, but in an optical fibre the refractive index of glass slows photons down to around 200,000,000 metres per second. There is some latitude for adjusting this index to gain further small reductions in the transit time, but that’s about it.

However, there is one obvious move that might be a really big winner. You could…

…locate the trading machines’ mid-cable span between financial markets – on a boat or in a submarine perhaps?

But if you haven’t already spotted the travesty here, permit me a short rant.

All this effort does not benefit mankind in any way. It changes nothing fundamental and does not create one iota of value. All it does is shave off money from the wholly illusionary activity of creating virtual value. None of this technology existed until about 1995 and it’s already been cited as contributing to numerous market flash crashes since.

The present global crisis came about through bankers and markets creating over $50 trillion of nothing. Monetary numbers not backed up by commodities, manufacturing, services or anything else – just numbers in a spreadsheet. And so it might be that high-speed trading is a small mechanism waiting to stimulate a big crash.

In the meantime, the race goes on. Faster and faster electronics and software speed up trades and make more virtual money. However, there is a snag.

These systems have multiple feed-forward and feed-back loops to the extent that they defy mathematical analysis. But we do know that such systems tend to be unstable. In fact, just like a helicopter they tend to be conditionally unstable, rather than well behaved like conditionally stable fixed-wing aircraft.

What does all this mean? If the system were analogue it would be prone to oscillate, while digital systems can go into limit-cycles and chaotic action.

Has this actually happened in a stock market in the past? Yes. Black Monday in 1987 was attributed to machine trading instabilities – and that was well before high-speed markets were established.

Since then there have been many more incidents, including flash crashes and chaotic transients rooted in high-speed trading. This problem isn’t going to go away, and it’s a risk being amplified by the continual removal of delay.

Based on recent evidence I think we can safely say no one understands how the financial system of the planet works. And in the case of automated trading, that lack of understanding is a sure-fire certainty.

Sadly, I fear that we are in for more of the same, powered by ignorance and greed – and the outcome will not be pretty.