Book Report: Deep Thinking by Gary Kasparov

Do you remember the legend of John Henry? John Henry was a steel driller in West Virginia or somewhere thereabouts in the late 1800s. He was the best there ever was. Then one day the railroad bought a big steam drill that they said could drill faster than any man. Henry, secure in his abilities (and trying to avoid the unemployment line) challenged the drill (and the company) to a famous battle of “man against machine.” Using two 10-pound hammers, one in each hand, he pounded the drill so fast and so hard that he drilled a 14-foot hole into the rock. The drill, unable to clean off the bits of rock, got stuck nine feet in. But John Henry couldn’t celebrate for long, dying quickly of exhaustion. ⁽¹⁾Here’s Johnny Cash’s The Legend of John Henry’s Hammer.

I’m in the middle of the modern-day battle of John Henry and the steam engine. I’m a product manager. It’s my job to find out what customers want and what technology can deliver. Then I figure out how to get the computers to do the job that people need. So you’d think I was on the side of the steam engine, trying to make computers more intelligent so that they can just do everyone’s jobs.

Making computers smarter so they can do things like people is called Artificial Intelligence.⁽²⁾Artificial Intelligence is actually more complicated but humor me. A lot of people get very excited about Artificial Intelligence but it’s not as important as you’d think. While there are some things that computers can do better than humans (e.g., recommending movies, finding the quickest route), there’s a far larger and more important set of things that computers aren’t great at—at least by themselves.

In his book Deep Thinking, Gary Kasparov details his battle with Deep Blue and how computer chess, like many other forms of AI, go from laughably bad to incredibly good in just a few years. Kasparov lost to Deep Blue in 1997. He outlines a whole host of reasons including getting flustered in game 2 and IBM hiring a Russian speaker to spy on him. But he concedes that it was only a matter of time before computers were going to beat him.

With Deep Blue beating the world’s best chess player, you might think that would be the end of human vs. computer chess. But Kasparov had a question, “What if instead of playing against each other, the computers and humans could partner?” This eventually turned into a competition called “Advanced Chess” in Lyon, Spain in 2007. People could compete in teams and use computers. Kasparov called them centaurs. Traditionally the use of computers by human players would be considered cheating. With substantial prize money at stake, there were entries from the world’s greatest grandmasters and IBM’s newest supercomputer “Hydra,” which was many times more powerful than Deep Blue.

As it turns out, grandmasters with laptops could easily beat Hydra and the other supercomputers. The laptops could check the work of the Grandmasters and keep them from making stupid mistakes. But one team was a break-out winner. Many people assumed it was Kasparov himself with a supercomputer. Surprisingly, this was a pair of amateur players with 3 laptops. These were neither the best players, nor the best machines, but they had the best process. This led to something calls Kasparov’s law:⁽³⁾Kasparov, like many people with laws named after them, would rather it be named something else. “Weak human + machine + better process was superior to a strong computer alone and, more remarkably, superior to a strong human + machine + inferior process.”⁽⁴⁾Kasparov actually does a better job of discussing this collaboration in The New York Review of Books and this TED Talk than in the book.

Kasparov holds a view that Augmented Intelligence ⁽⁵⁾J. C. R, Licklider discussed Augmented Intelligence a long time ago in a paper called Man-Computer Symbiosis. Interestingly, Licklider thought that once computers could do all his calculations, he’d have fewer things to do and could focus on more strategic questions. is more important than Artificial Intelligence. Take the example of email. Artificial intelligence can do some cool things like spam filtering and sentence completion. But look at the far more powerful system of Augmented Intelligence. This is the system of human-computer email, where a human and a computer work together to communicate, organize, and act on information.

Computers and humans are good at different things. Essentially machines do the grunt work really well, allowing humans to focus on strategy. Here are two examples from the book:

• Computers are great at calculations. Kasparov quotes Charles Krauthammer in Time Magazine about the match with Deep Blue. Krauthammer said, “Blue ignored the threat and quite nonchalantly went hunting for lowly pawns at the other end of the board. In fact, at the point of maximum peril, Blue expended two moves—many have died giving Kasparov even one—to snap one pawn. It was as if, at Gettysburg, General Meade had sent his soldiers out for a bit of apple picking moments before Pickett’s charge because he had calculated that they could get back to their positions with a half-second to spare.” But Krauthammer points out that this would be a perfectly sane thing for Meade to do if he could calculate, and really know, that he had that time.
• Computers are bad at undefined strategic questions. In the Hitchhikers Guide to the Galaxy, the computer Deep Thought calculated “42” to be “The Answer to the Ultimate Question of Life, The Universe, and Everything.” Unfortunately, no one knew what the question was. Deep Thought was the original name of the computer that beat Kasparov before IBM bought it and renamed it Deep Blue.

The story of man vs. machine has morphed significantly since John Henry. It was once the story of horses being replaced by cars. Then it was weavers and other artisans being replaced by factories. Now it’s computers coming for white-collar workers. But to me, it’s not a battle if man vs. machine where you can bet on the winner. As a product manager, I believe that the bigger opportunity is about getting them to work together to do something truly great.

Note: I wrote a version of this as my Idea of the Year in 2015.