It was bound to happen, sooner or later. Technology evolves at incredible rates, and greed is by far the strongest driver in that evolutionary process. In many ways, the tech world is beginning to mimic the biological world. We have neural nets (analogous to the brain), genetic algorithms (analogous to the process of natural selection), and now we have digital parasites that are at least as aggressive and tenacious as their living counterparts.

I recently came across an ad for Stone Ridge's new HFT-X development kit. Aimed at programming teams working for banks and investment houses, the HFT-X is a powerful computer server, highly evolved with specialized hardware to perform one task: play the stock market so quickly that by the time anyone notices what you're doing, you've already made off with the money. "Normal" high frequency trading takes place on the order of milliseconds; the rush is now on to use field-programmable gate arrays (FPGAs) and other high-speed hardware to speed up those stock deals to the microsecond level.

From an HFT developer's perspective, the problem with millisecond trading is that there are too many layers between the trading algorithm and the actual deal. The computational effort involved in bouncing data and instructions back and forth between RAM, CPU and network adds valuable time to the transaction. Adding FPGAs is a logical evolutionary response: the algorithm can now be optimized, compiled and implemented directly in hardware, without all that tedious back-and-forth between discrete computing components. Some traders are now getting worried about the speed-of-light delay becoming a limiting factor in such transactions.

From a stock trader's perspective, this is a wonderful development. HFT relies on the algorithm being able to quickly insert itself into transactions, buy from one party, then- before anyone notices- sell to a different party at a slightly higher price. The faster you can do that, the more transactions you can do and the more money you can siphon off of each one.

From almost anyone else's perspective, though, this can only be interpreted as destructive, parasitic behaviour. Remember the May 6, 2010 flash crash?  A major contributing factor to that mess was HFT algorithms that went haywire in response to unexpected market movements, playing off each other and driving the Dow Jones down six hundred points in under five minutes.  Eventually, the chaos was cut short when yet more computers- programmed by slightly more sane, less greedy institutions- cut off the markets, giving the human element a chance to put everything back together.

It is very hard to argue that millisecond-level trades, let alone microsecond-level ones, reflect true changes in the value of the underlying stock or commodity. Indeed, the inherent value of most financial instruments changes very slowly, over a time scale of days to months, with occasional exceptions such as when a company wins a major contract or declares bankruptcy. HFT may represent a technological bright spot, but it's a destructive one, detracting from the ability of the market to judge true value and siphoning off resources with no net benefit to any player but the parasite.

Various ways around the problem have been proposed. One interesting idea that's been mentioned now and then is to tax stock deals based on the time the position has been held; the tax might start at 98% of the profit on the deal for positions less than one second old, dropping progressively to some small value, say 10%, for positions held more than a year. Another is that listed corporations may eventually get fed up with their stocks being manipulated this way, and could insist that they be listed only on exchanges that prohibit HFT. For the moment, though, the only solution actually being implemented is "circuit breakers"- computers that, like those of the Chicago exchange that put a stop to the May 6 flash crash, will intervene to momentarily kill the markets if anything starts to move too quickly or in too high a volume.

Technology isn't always the solution. Sometimes the only real solution is to change the underlying culture on which the system is built. For now, we're going to see more and more computers and algorithms tossed at the problem in the hopes that it'll solve itself. But I'd much rather see the markets restored to their intended function- as markets.