The design of the system is that it intelligently scrapes quantifiable information that can be put into a defined knowledge base structure and inter-related. Length, weight, oribital period, age, population, molecular weight, wavelength, numeric series, calories... values that are measured in units or physical properties of the world around us. By fitting this information into a defined structure the system has the ability to now extrapolate from it to answer questions... hence the words 'computational engine'.
Why build another text search "library index"? It's been done out the ying-yang. This system is orders of magnitude more ambitious and complex and while still in it's infancy, it's a pretty spectacular achievement already IMO. Just allow yourself to think outside of the 'search engine' box. While it contains some facts about the world, it's not a search engine.