You may know the old joke about the supercomputer that was fed the question “what is the meaning of life”, and that came up – after long calculations – with the answer “27”. This joke has been told to illustrate many different points about computers, human computer interaction, philosophy, and life in general. Here, I would like to draw your attention to fuzzy questions and precise answers, or vice versa, and to better ways of interacting with computers.
Shyam Sankar gave an interesting talk on the subject at TED, called “The rise of human-computer cooperation”, where he “explains why solving big problems (like catching terrorists or identifying hidden trends) is not a question of finding the right algorithm, but rather the right symbiotic relationship between computation and human creativity”.
His first example is well known but deserves to be retold. It is the story of 2 world-class chess championships: World champion Gary Kasparov losing against IBM’s “Deep Blue” computer in 1997. In 2005, in a free-style chess tournament, in which men and machines could participate as partners, a supercomputer was beaten by a grand master with a relatively week laptop. But to everyone’s surprise, the tournament was not won by a grandmaster with a supercomputer, but by two amateurs using three relatively week laptops. Sankar argues that the way of interacting with their machines helped average men with average machines beat the best men with the best machines.
Now what has that got to do with Search or Unified Information Access?
Well, it may be a long shot and maybe I am not putting it into the most convincing sentences, but perhaps someone can help me by adding their symbiotic brain-and-computer power to my argument.
In “classic” computing, using databases, data warehouses, BI systems, etc., the (super) computer is asked precise questions – by people who understand the structure of their data and the way to ask these precise questions – and the computer comes up with an answer like “27” or a nice dashboard illustrating figures and possibly even trends. If you want to ask a question that takes you outside the structure of the data or the predefined logic of the “decision support” programs, you are out of luck.
Search, on the other hand, allows you to ask fuzzy questions in natural language and it comes up not with a “27”-type answer, but with a set of answers – be they documents or data base records – ordered in categories for you to navigate in. (Categories aggregate information from multiple sources, including business applications.) You can zoom-in on what your human intelligence recognizes as the most promising subcategory. It is easy to refine your question after insight from the first set of answers, indeed to ask any question you like without any need for reprogramming. And in a man-machine Ping-Pong of three exchanges, you have a good chance to discover an answer that your supercomputer with its elaborate programs would not have come up with. Or maybe it would – but with a few thousand man-days of development and “tuning”, and millions spent on top-notch hardware – just like Watson won Jeopardy.
At Sinequa, we like to think that our software is above average, but even if you assume it to be just average, the interaction of users with our Search and Unified Information Access platform is rather like that of the two amateurs with their laptops who beat the chess champion with his supercomputer.