If you're new here, you may want to subscribe to my RSS feed. Thanks for visiting!
The Official Google Blog put up a post, entitled “The Future Of Search” today, talking about some of the questions and challenges that Google is tackling in their ever expanding quest to make search better. While reading it I couldn’t escape the thought that the science of search is really the science of human cognition. It is the science of figuring out how human beings think and ask questions, and then teaching a computer to understand those questions.
For example, if I were to ask you, “What color is that?” Before the question can be answered, we would need a context, ie: identifying “that”. Is it a dog, a car, or are we talking about the color itself? As human beings, we naturally interpret what the other is asking with little effort, because we think alike. If we were talking about a car, then the question would be about the color of the car, or a part of the car. If we were talking about colors, then I would be probably asking you to identify the name of the color, such as “magenta”.
But those are simple, linear questions and answers, and we humans are neither simple, nor linear. We could be talking about cars, and the example question could be popped in the middle of the conversation asking about something totally un-car related. In the normal course of events, you, knowing that I am perfectly aware that the car is red, would then ask yourself your own question: “What is he really asking?” You would probably look at me, to gauge where I’m looking, and then look there yourself to see what we are talking about now.
And that’s not even getting close to our liveral use of metaphore and analogy in our daily speech. “I’m seeing red right now” is one example. Teach a computer to figure that one out.
Our daily person to person interaction relies on mechanisms like this, with incredible levels of subtlety and sophistication. Without it, our conversations would be long, drawn out, stale exercises in frustration. Yet, that’s exactly what we have to endure with search engines: long, drawn out exercises in frustration. The search engines rely on us to be clear and concise in our queries because they cannot interpret “that” to mean “car”. They cannot really predict the ebb and flow of our thoughts, nor can they switch topics on the fly without first being told that there is a new topic. We are still at a point where we have to speak the computer’s language, rather than it speaking ours.
While there are a variety of projects going on around the globe as we speak trying to teach computers to understand natural speech patterns, and behave in very human like manners, they are still a far cry from anything resembling human style thinking.
If Google wants to really revolutionize the search engine industry, my advice is that they should invest heavily into the science of human cognition. Only then will they be able to teach a comuter to understand a the rainbow experience of human interaction.











