Computers are only human

Mr. Langer, in the 1970 science fiction film Colossus: The Forbin Project, a robot prophesizes to a human that, "In time you will come to regard me not only with respect and awe, but with love." Have we now, almost half a century later, come to the point where we are more or less dominated by computers, and are even emotionally controlled by them?

Langer: It's clearly tempting to think in such terms, and the topic has proven particularly exciting for authors and filmmakers. But for me the computer represents a technical resource - a tool that is intelligent to a certain extent, and which allows me to process large amounts of information. Maybe I'm too matter-of-fact in this respect, and maybe my attitude is itself a sign that I have already long been controlled by computers. Have you read the thriller The Fear Index by the British novelist Robert Harris? It tells a story of how in our world, the investment world, the computers take over and begin to learn on their own. In the end, they become this monster that seizes control of everything. We are a long way away from such horror scenarios with our programs. The limited ways in which we technically use computers "only" as a resource is almost disenchanting. By no means would I argue that computers direct us or even control us. The computer is, and will remain, a pure tool.

Fraikin: The idea that a computer can control us would be like talking in the past about a dictatorship of the typewriter. I also can't imagine that people would ever look at such a machine with awe and love. I do believe that software or computers have crept into certain areas of our lives, and they may even dominate our lives in these areas. But the fact is that we don't even really notice it in many cases. Think about the route planner that calculates your trips while taking traffic information into account. I'm not sure people really think about or are even conscious of the fact that they trust their trips to a computer. If we really took a minute to reflect on that, we would quickly arrive at questions that might make us uncomfortable. Basically, the key issue is: which areas of our lives do we want to entrust to computers, and which do we not?

Let's stay for a moment with your example of the route planner. Doesn't the computer in this case still have an influence on our emotional life? After all, women tend to choose a male voice for computers and men tend to choose a female voice.

Langer: Sure, it's obvious that psychological factors play a large role in our behavior, including how we interact with computers. But we are still a long way from being able to talk about computers controlling our emotions. Think about the amazing chess playing computer that IBM built. Deep Blue could calculate more than 100 million chess moves per second, and it also had the ability to learn over time. But at a fundamental level, the computer was just programmed so that it could think ahead when it came time to choose which move to make. And it works the exact same way with Go.

What you mean by Go?

Langer: I'm talking about the game Go, the Asian strategy game somewhat similar to chess. Chess is ultimately simple: it has 32 pieces that can be moved around on an 8 x 8 game board - a game board with 64 squares. Go, on the other hand, is played on a 19 x 19 board - that's 361 points that players can place their stones on, and the number of moves is essentially unlimited. Go is thus many times more complex than chess. Another difference is also that Go players who find themselves in a difficult position can almost always recover from it. Nevertheless, computer programmers were still able to successfully model even this high level of complexity. But that doesn't make the computer itself more intelligent. It is and remains a rational tool. The computer itself does not have emotions.

But if we look at the fact that a chess playing computer can beat almost any player in the world, doesn't that support the thesis that the computer is beating people at their own game?

Langer: Yes. Autonomous driving is another good example of that. When it comes to subways, for instance, we haven't actually needed drivers for a long time; the driver is only there now so the passengers feel more at ease. At the very least we recognize our dependency on computers when the power goes out. Our entire lives are managed by computers, even our water supply, everything. But I still don't think that we are at the mercy of computers. If something goes wrong, then it's a technical problem, just like a flat tire. It's just that an IT failure rapidly reaches a higher level of complexity, like in the summer of 2016 when the computer systems went out at major airline. That malfunction affected thousands of people who were unable to board their flights, and it caused multiple airports to descend into chaos. When a computer malfunctions, it rapidly leads to a breakdown of the system. A dependency certainly exists in this respect, but I would still describe it as a technical dependency.

In the finance business, we still hold a certain romantic image of the stock exchange: of traders yelling across the trading floor, gesticulating wildly and looking like they're about to have a heart attack. But the reality of the finance world today is totally different. Instead we see market players sitting around in open plan offices, working intently on their computers. When would you say this transformation began to occur?

Fraikin: The computer first entered the trading world at the beginning of the 1960s. One great milestone in the shift towards computers was the Black-Scholes model in 1973, a mathematical model used to calculate financial options that required the use of computers. Indeed, the calculation of complex derivative transactions is still only possible with the help of computers. Later in the 1970s, we also saw the emergence of the BARRA risk model, which is used to better forecast the risks associated with stocks and other financial instruments. That was the first time quantitative risk models were built that were based on an academic theory.

What would you say were the major landmarks in the history of computers entering the finance world?

Langer: From my perspective, there were two aspects. The first area of change was the processing of orders: the path taken from placing an order to its execution, processing, and posting. Previously, investors would call their bank advisor or securities dealer, and the dealers would then pass the order to their people trading on the stock market. The order was taken in a phone booth, written down on a piece of paper, and physically brought to the trader. Only then was the order transferred electronically to the stock exchange and processed by the traders. Today the entire process is electronic, from placing the order to its execution and processing, all the way to the accounting. The second large change was BARRA, the software I mentioned earlier, which was developed by Barr Rosenberg in Berkeley, California. He was one of the first quantitatively focused analysts, and definitely one of the most influential. This is the area where computers have changed the analysis of securities. By making it possible to process an enormous amount of data, computers have helped us base our analyses on a much more rational foundation. For me, these are the two biggest changes that computers have brought about in the financial markets: technical processing and analysis.

Why have computers had such an influence on the financial markets in particular?

Fraikin: In the beginning, it scientists were the ones who worked intensively with computers. The first people in the investment world to work with quantitative methods were actually educated as physicists and time series analysts. They really only took an interest in the financial markets because they were looking for data sets that they could work with. If you want to analyze a time series and test statistical methods, you need access to a time series. Financial markets are perfect, because they have time series with a long historical progression and a huge number of data points – namely, the entire series of stock and bond prices, which are precisely documented and some of which can be traced back into the 19th century. This kind of data can be found in the stock market world, but not necessarily in other places. The implication is, of course, that if researchers had found time series outside of stock market data they might have focused on a completely different domain.

And so computers were introduced to the financial markets by sheer chance…

Fraikin: Right: in...