Between February 10th and 17th in 1996, the then world champion chess player Gary Kasparov took on an unusual opponent in a match that grasped the intrigue of the masses. Kasparov’s opponent was Deep Blue, and Deep Blue is of course a computer. In a series of six matches, Kasparov came out on top winning three matches with Deep Blue winning one and two ties occurring. Deep Blue was reprogrammed and a rematch took place in 1997. The improved Deep Blue could now analyse 250 million potential plays per second as opposed to the 100 millions per second of the previous year. This time, machine came out on top with Deep Blue winning two matches, Kasparov winning one match, and three matches resulting in a tie.
This is the classic tale of man versus machine. The 1997 clash was called “the most spectacular chess event in history”. We are now over 20 years on from the second match and the man versus machine battle has not died down.
It is now pervasive throughout society. Let’s take a look at some of the things which have happened since then.
The dot-com bubble of 1999 and 2000 has came and gone. Many made money and many lost money. Regardless, this bubble served a purpose. It injected a huge amount of capital into the tech industry and this capital played a key part in the building of the infrastructure central to our systems today.
We are now all walking around with a supercomputer in out pockets. No amount of money half a century ago could provide you with the amount of computational power and information each of us now has access to on our phones. The amount of information available has also increased exponentially. It is believed that more information was created in 2017 than the past 5,000 years of humanity.
A new player has entered the game of financial speculation. Algorithmic traders utilise programming knowledge and the computational power of computers to speculate on financial markets. They can analyse enormous amounts of historical data and develop statistical models to speculate where the market goes next.
Many big trends have shaped our world. We have had big data, cloud computing, social media, and currently blockchain and artificial intelligence are topics gaining a lot of attention. Artificial intelligence is not only making it into the newspapers, it is also gaining a lot of attention on the screen.
(
As the title suggests, I am here to make a case for the humans. So there are things which computers can undoubtedly do better than humans. Computers perform extremely well in tasks which involve a lot of known factors, historical data and an environment which is not subject to too much change and is fairly predictable. Chess is a great example as there is often going to be an optimal move to play based on the remaining pieces and board position. A machine has the computational power to go through potential moves and determine the move which statistically gives it the greatest chance of success. Even if it fails to play the best move, this computational power is a force humans will not be able to compete with. This computational power enables machines to analyse enormous amounts of data on a time level which is just not feasible for a human to compete with. It takes a human approximately 120 milliseconds to even become conscious of something. This is about .1 of a second. Computers can operate in nanosecond and terasecond territory (billionths and trillionths of a second). Many things humans are currently doing can be done much more effectively by a machine.
With that being said, there are things humans can do which a machine simply cannot do currently. So machines are extremely effective when it comes to environments with known factors. When unknown factors and unpredictable events come into play, machines don’t perform well. Trading systems based on quantitive systems have been shown to break down in times. A famous example of a trading system breaking down was the story of Long-Term Capital Management (LTCM). LTCM was a hedge fund formed of prodigious academics containing a number of nobel prize winners. They used complex models to identify small price discrepancies in the market and utilised huge amounts of leverage to profit from these discrepancies. LTCM achieved extraordinary returns in their first few years but after only four years of operation, LTCM ended up blowing up and having to be bailed out by a number of Wall Street Banks. In an increasingly interconnected and globalised world with more and more complex factors to take into account, market environments are not optimal for the kind of predictive models machines are good at executing. LTCM’s complex models failed to adequately manage risk for unlikely circumstances such as Russia defaulting on it’s debt which occurred in 1998 and led to the unravelling of LTCM. For an interesting account of the LTCM story, Roger Lowenstein’s “When Genius Failed” is an informative and entertaining read. There have been many examples of machines failing to operate as expected in the market. The 2010 flash crash saw the S&P500, the Dow Jones Industrial Average, and the NASDAQ all sharply fall and recover in the space of just a few minutes. The 2010 flash crash has been largely accepted as the result of algorithmic trading gone haywire in the market.
(
Apart from the shortcomings of machines, humans also have powerful tools which machines do not have. Humans can perform second order, third order, and nth order thinking. First order thinking is essentially an opinion about the future. Second order thinking and beyond is a more complex process which is highly useful in today’s business environment. So when a machine processes an enormous amount of past data and predicts the market is going to go up, this is essentially first order thinking. On the other hand a human trader may go through a much more complex process. They may go through a process which could be something like this:
-Institutional investors tend to move the market so let’s identify times which are attractive places in the market for institutional investors to enter a position (2nd order thinking)
-Due to large position sizes, institutional investors will look for points of liquidity. They may identify these points of liquidity by taking into consideration where retail investors will find it attractive to enter the opposite direction positions, e.g. chart breakouts. So by identifying where institutional level players predict retail level investors will enter, this may be an attractive place to enter (third order thinking).
This is the kind of power a human possesses which a machine does not. Today’s environment also is optimal for these kind of skills. So through raw computational power computers can beat humans in games such as chess. But add some unknowns and different levels of thinking into the game and we have an environment which machines will perform extremely poorly against humans. So let’s take the example of poker. Computers may be able to quickly analyse their cards, the remaining cards, and determine the optimal move and bet sizing based purely on mathematical expected value. But in poker, those who play the most mathematically optimal strategy are often not the ones who win as the game is not about that. The game is about thinking at levels far beyond the level of what is the optimal strategy. If you know the strategy another player is playing in poker, then you can beat this strategy as you know what they are likely going to do and you can counter this. This is the beauty of games such as poker. How can a machine beat a game in where always playing the optimal strategy results in your opponent learning this and countering it. They could add some random deviance from the optimal strategy but if this can work better than an expert poker player is very debatable.
Beyond games, there are many real world fields where the powerful tools humans possess will outperform machines for the long foreseeable future. Politics, business-strategy, consulting, and discretionary trading all involve multiple layers of thinking and operate in ever-changing and unpredictable environments. On a final note, while I still have to tick a box to confirm that I am not a robot I am not going to be too concerned about the humans versus machines tale.
(1) https://pt.wikipedia.org/wiki/Match_Garry_Kasparov_vs_Deep_Blue
(2) https://www.goodreads.com/book/show/13151167-the-hour-between-dog-and-wolf
(3)https://www.scientificamerican.com/article/computers-are-becoming-fa/
(4)https://www.makeuseof.com/tag/geeks-weigh-in-does-a-human-think-faster-than-a-computer/
(6)https://www.goodreads.com/book/show/10669.When_Genius_Failed