In the past few years, I've been a fervently anti-bank corruption, often aligning myself with the occupy wall street crowd. However, unlike most people with my views, I see algorithmic trading as not a symptom of, but one of the solutions to the problems in investment banking.
Maybe it's because of my background in machine learning, but I view computers as a way to reduce the amounts of arbitrage opportunities and insider knowledge that can be exploited by traders to enrich themselves on the back of others.
Well programmed computers are able to use NLP techniques to read thousands of news articles, reports, financial statements, government data, demographic data , analyse millions of data points, sort which ones are important, build predictive models, and do it all in a few seconds of time.
A few computers can do the job of thousands of traders, do it with less bias and with better mathematically proven decisions.
Nowadays every time a human trader tries to beat the market by predicting short terms swings in prices, there is a computer on the other side of his internet connection that is relying on much more information and even models of the human's own personality and biases to trade against him. This means that unless the human has inside information that is not accessible to the computer, (and I'm sure those who operate these computers try to feed them or make them infer the greatest amount of direct or indirect insider knowledge as possible ) it will tend to be impossible to beat the market.
With the greater amount of information, the computers will keep the prices closer to real world fundamentals, bust bubbles before they inflate and reduce all the price swings that traders used to be able to exploit to make a profit on the back of less sophisticated investors.
By making obsolete the jobs of all these traders, they are making the market much more efficient. They are automating a kind of job that didn't produced any real goods. Hopefully, pushing these people into jobs that make something real instead of skimming the top of everybody's retirement savings (I can see why those who want to be traders would be upset about this).
Ah but you say, all we have done is replace humans from skimming our retirement savings by machines that do the same. This would be true if there was only one computer system competing against the human traders but there is a whole industry and they also compete against each other. Two sophisticated machines that trade against each other, will not only beat the human trader and put him out of business but reduce the arbitrage opportunity and margin each of these computer systems can exploit.
Contrary to what the article states, computers that compete against each other will tend to eliminate all short term unjustified price swings and leave only accessible prices based on real company fundamentals. Only long term traders that bet on fundamentals will be able to make money and they will make more money than when they were subject to be exploited through market distortions and bubbles.
This can all be proven by something called the efficient market hypothesis:
There is a great explanation by Glen Whitman of Agoraphilia, that uses grocery line wait time predictions as a metaphor:
http://agoraphilia.blogspot.com/2005/03/doing-lines.html
See also:
http://lesswrong.com/lw/yv/markets_are_antiinductive/
http://en.wikipedia.org/wiki/Efficient-market_hypothesis