By: Dick van Mersbergen
In one of the meeting rooms at Be Informed I came across a famous quote by Sun Tzu:
Strategy without tactics is the slowest route to victory. Tactics without strategy is the noise before defeat.
As a vivid (but not very strong) chess player, I study strategy and tactics. So I grinned and nodded - every chess player knows that Sun Tzu was right.
Besides, working at Be Informed, I am interested in computers and artificial intelligence. So Sun Tzu’s quote also made me think of the link between Be Informed and chess computers. The two proved to be very similar.
A short history of chess computers
The first chess machine was constructed by Wolfgang von Kempelen in 1770 to impress Hapsburg Empress Maria Theresa. It was called The Turk, after the mechanical Turk that executed the moves on the chess table. The machine travelled Europe and beat strong players and even Napoleon in 1809. Soon after that, the machine was unveiled as a hoax. In the small chamber underneath a master player was hidden, he made moves via a ingenious array of mechanics.
The David Levy bet
As soon as computers appeared, programmers tried to make it play chess. In the 1970s and 1980s many thought no chess program would ever be able to defeat a top human player. In 1968, International Master David Levy bet that no chess computer would be beat him within ten years. He won this by beating Chess 4.7 (the strongest computer at the time) in 1978, but admitted that it would not be long before he would be surpassed. He was right: in 1989, Levy was defeated by Deep Thought.
Turning point was the 1996 match between IBM's Deep Blue and Gary Kasparov (one of the strongest chess players of all time), where the latter lost his first game to a computer. But Kasparov regrouped to win three and draw two of the remaining five games of the match, for a convincing victory. In May 1997, an updated version of Deep Blue defeated Kasparov 3½-2½ in the rematch. Note that athough the Cold War was long over, this was still a match of Russia vs the USA. A new era began: chess programs had overtaken human players.
How did computers become unbeatable at chess?
The first study on the subject was published in 1950 by Claude Shannon, long before anyone had programmed a computer to play chess. He predicted two possible search methods that would be used and labeled them Type A and Type B. Type A programs would use a brutal force: exame every possible position for a fixed number of moves. Shannon thought this would not suffice, because:
- the computer would be to slow. With about thirty moves possible in a typical position, looking just 3 moves ahead for both sides would take about 16 minutes(!), even in the very optimistic case that the chess computer evaluated a million positions per second.
- it ignored the problem of quiescence. When only calculating a number of moves deep, you never know if there is a move later on that refutes the combination.
- only looking at good moves for each position
- using quiescence searches.
Programmers worked on these improvements for years, but the problem is that it is very hard for a program being to decide which moves are good enough to be worthy of consideration. Chess players rely on their ability to recognize patterns and on their intuition. Because computers don’t have an intuition and don’t have a strategy, this proved to be counterproductive.
Quiscence: look for stability as a goal
Instead of imitating human thought processes and knowledge it turned out that computers need to do what they are best at: calculate. But they needed to understand what to calculate. Not having to check all possibilities, modern chess programs look for steady positions and the ways to get there; just these are evaluated. This was the key to world domination.
The Be Informed approach
Similar to chess computers, traditional business applications are very good at Processing, but very weak at Management. This is where Business Process Management goes wrong. The BPM suppliers at first used brutal force to deal with the zillion possibilities, then tried to improve that by making the software to think like humans.
As with chess programs it is better to define goals instead of define and manage all possible paths. This is exactly what Be Informed did: the knowledge models within Be Informed let the application work towards goals and look for the paths itself. Be Informed gave BMP its quiescence search.
So where chess programmers found that it was useless imitating a chess player, we at Be Informed don’t want our business process platform to imitate the knowledge worker. We want it to free the knowledge worker. Let’s make them both do what they do best: calculate and and decide what is good respectively.
Being a chess player I can truely sum up Sun Tzu’s view with the pleasing words: Check and mate.