Could this be applicable to search ranking? Say I was doing a search based on an input string, and I had a "closeness of match" heuristic. That expected value f

Could this be applicable to search ranking? Say I was doing a search based on an input string, and I had a "closeness of match" heuristic. That expected value for N=100 case seemed so good that I'm considering the possibility that you only partially search any data set and return a result that is pretty much good enough. (At least initially, and then you could asynchronously add better results to the list, in the background. Then you'd get very fast first results, at the very least).